AlastairThomson.com

“Don’t just do something, stand there!” is one of my all-time favourite quotes. Often attributed to former US President, Ronald Reagan (online sources vary), whoever said it opened the door to an important concept – one we probably need to think more about in our business lives.

When something happens, there’s often an implied pressure to “do something”.

That might be taking action to fix a problem, or taking action to repeat or accelerate something that is perceived (at the time, at least) to have gone well.

Often, however, we might be better “standing there” for long enough to think more deeply about the issue, rather than just diving into action mode.

How these issues present themselves is a little different depending on whether you’re trying to fix a perceived problem or to go all-in on something that’s perceived to have gone well.

So let’s take those issues one at a time.

Houston, we have a problem…

A long time ago, I worked with someone who, when presented with a proposal for action to address a problem, always asked “is this a fix or a solve?”

What he meant by that was whether the action proposed was just papering over the cracks – however important that might be in the short-term – or whether it was intended to change the underlying systems and processes to make sure the problem never happened again.

Spending extra budget on an express courier to get a delivery to our customer on time is an example of a “fix”. Ascertaining why our existing delivery system was so unreliable that we had to resort to the extra expense of express couriers from time to time, and doing something to stop that happening, is a “solve”.

We could, of course, spend £000s every day of the week on express couriers and leave everything else the same. Customers would be happy enough as they’d get their deliveries on time, but our carriage costs would go through the roof and our margins would take a hit.

But action is being taken. Someone gets to play the hero: “don’t worry, boss, I’ll make sure the delivery gets there on time!”. That’s got to be a good thing, hasn’t it?

Well, in the short-term, it may be necessary to retain an account worth £millions. And sometimes stuff just happens and you have to do your best in trying circumstances. But when this sort of response becomes a habit, you start to build endemic problems into your business which ultimately, one way or another, will backfire badly.

Contrast that with the “stand there” approach, and taking the time to come up with a “solve”.

Maybe it turns out that the problem isn’t with your haulage contractor at all. They’re doing their best in difficult circumstances.

Perhaps the problem is that the goods they need to deliver don’t get to the loading dock on time – if it’s a 6 hour drive to the customer and the products only land on the loading dock 2 hours before they’re due to be with the customer, no haulage contractor in the world is going to be able to solve that problem for you. All they can do is their best, in difficult circumstances.

Tracking back a little further, it might turn out that the reason products get to the loading dock late is that the machines in the factory are having to run at a slower speed because the maintenance which was due to happen a few months ago had been postponed because your Finance Department had forbidden any extra spend in the run-up to the financial year-end.

Now we have “stood there” for long enough to understand the real problem, the “solve” is obvious. And once implemented, you won’t need to spend a penny on extra express delivery costs.

Or you can keep spending £000s forever on express delivery services just because someone enjoys the rush of adrenaline as they run through the factory to hand an urgent package across to the express courier who’s waiting, Le Mans style, to jump in their van and screech out of your car park on their way to make your delivery.

That’s the benefit of “standing there” instead of “doing something”.

Things are going too well

The “stand there” mantra applies just as much when things are going well.

Maybe you run a successful independent business, which has supplied a range of specialist stores with your products for generations.

Then, one day, a big retailer asks if you could supply them.

It’s a well-known brand. Supplying them would be good for your reputation, and give some valuable social proof to use with sales prospects among the specialist stores you target.

And more sales has to be a good thing, right?

Well, not necessarily.

If you do the occasional one-off order for a major retailer, probably nothing will change in your business. But big retailers like to have enough of a share of your business that they can crack the whip and squeeze your prices down – if they threaten to take away 60% of your business overnight, odds are you’re going to do pretty much whatever they tell you to do.

Now, there’s nothing wrong with supplying major retailers. And there’s nothing wrong with supplying a network of independent stores. It’s just that those are two different business models, with entirely different operating principles, and they don’t often make easy bedfellows…especially by the time your big retailer client has taken up 60% of your productive capacity.

What usually happens is that over time most organisations prioritise the big retailer client – “they’re 60% of our business, for goodness’ sake, we don’t want to lose that, so do whatever they say!” becomes your primary operating principle.

Meanwhile your traditional clients get side-lined, feel underappreciated, and ultimately go elsewhere to recapture the top-notch service you used to give them before that big retailer cuckoo landed in your nest.

As your traditional clients drift away, that 60% share of your output taken by the big retailer drifts upwards to 70 or 80%, perhaps. A few of your other clients will hang around, perhaps due to inertia, or some personal connection with you. Although it’s unlikely you’ll land another big retailer as a client because retail is an intensely competitive market and the last thing any retailer wants to be is an afterthought in another big retailer’s major supplier.

So ultimately your business becomes a production unit for the big retailer, operating on wafer-thin margins and entirely dependent on the big retailer continuing to buy from you as they take up 60, 70 or 80% of your output.

It’s not great secret that this is how major retailers work. A business I used to run supplied lots of businesses which had relationships with major retailers like this, and they were rarely masters of their own fate.

Admittedly “more sales” is a good thing. It’s something most businesses, rightly, strive for.

But what if, knowing the retailer’s likely end-game, you just “stood there” when the retailer asked if you could supply more than their initial small trial order…and perhaps ultimately form a long-term partnership with them, resulting in £millions of future revenue for your business.

“Doing something” is the easy way out. It gives the appearance of activity. In the short-term, at least, it appears to be eminently sensible and helps the business grow sales revenues faster than dozens of specialist retail clients ever would.

“Standing there” is much harder. But sometimes it’s the right thing to do, in the end.

Do something or stand there?

That being the case, how do you know when to do something and when to stand there?

Well, of course, every situation is a one-off, so I can’t give you a cookie-cutter approach that’s guaranteed to work every time. But there are a few things you can look out for, which might give you some clues.

1. Put out the fires

If your building is on fire, grab a fire extinguisher (if it’s safe to do so) and call the fire brigade. Don’t agonise about taking action to prevent a disaster. Stay safe, but take the action. Once the fire is out you can think about other things, but when the building is on fire, do something, even if it’s just to get out to a place of safety.

“Standing there” is unlikely to be a sensible option.

2. Repeatability

If, once in a blue moon, you have to arrange an express delivery for a client, just do it, and don’t spend a lot of time agonising about it. There’s almost certainly no sensible RoI on you spending hours of your time to save £25 a couple of times a year. Just spend the money and get the delivery made on time.

On the other hand, if you’re organising express deliveries several times a day and spending £000s a year, you’re almost certainly not putting out a (metaphorical) fire. You’ve almost certainly got a deeper problem you need to solve – you can’t just keep applying fixes for ever.

3. Business Model

Be especially careful if you’re tempted to do something that gets to the heart of your business model. Even if your new idea is a good one, and perfectly sensible and rational on some level, “standing there” for a little while to consider the new idea in the context of your business as a whole is nearly always time well-spent.

If you currently serve 300 small clients, think very carefully about whether adding one enormous client to that mix is going to work with your current business model. Even if you think it will, it probably won’t. In thousands of almost invisible ways, your business will be set up to service your current client base, much of which won’t sit well with the demands of mega-corp.

It’s not that serving small customers is right and serving large customers is wrong. Both are perfectly viable business models. They’re just very different.

It’s not that you can’t run both a world-class swimming team and a top Formula 1 motorsports team at the same time. However you’re unlikely to use the same people, systems, and facilities to do both.

So “standing there” for a while, and getting clarity on what you really want to achieve, is usually a good idea.

4. Systems impact

Some things in your business are systems, others are just activities (although a remarkable number of people who carry out activities would like you to believe they’re running systems, because that sounds cooler and more scientific).

Imagine you run a factory making beef burgers. You have a series of operations which need to be conducted one after another in a particular way to get the result you want – a perfectly prepared beef burger at the end of the line.

That’s a system.

An activity is something like writing up a policy for how many days of your staff’s annual holiday allowance they can carry forward from one year to the next.

While you need both systems and activities to run your business, you can change the annual leave carry-forward policy any time you like. Your staff might like what you do, or not like what you do, but there are essentially no other moving parts to this process. Writing a policy is an activity, however important and worthwhile, not a system.

However, choosing to replace the machine which mixes onions into your beef burgers half-way down your production line means there will be some impact on your production system – both during installation and on an ongoing basis.

If people don’t like your holiday carry-forward policy, you can write a new one next week.

Once you scrap your current onion-mixing machine, you’re either stuck with all the problems of the new machine for the next 10 years, or you’re spending £millions to buy a different one to get beef burger production back on track.

In either case, it’s better to “stand there” than just rush to “do something”.

5. External impact

The bigger the likely impact outside your business, the more “standing there” while you make sure you understand all the knock-on consequences of your decisions makes sense.

If your plans include a massive price hike for your existing customers, or the dismantling of a decades-long customer loyalty scheme (as one major airline did recently, to entirely predictable howls of anger from their customers), or a decision to save money on commercial waste disposal and dump factory waste into a local river instead…or anything with a likely big impact to your customers, your local community, or any regulatory bodies you need to interface with…then you should probably “stand there” for a while before taking actions you can’t easily roll back from.

Of course, sometimes taking unpopular action is necessary and unavoidable. But often organisations stumble into major issues entirely of their own making because they didn’t “stand there” for long enough to think through the likely consequences of their decisions.

Even worse, organisations in these situations have been known to blame their customers for being stupid, or tell everyone that people are “out to get them” without fully understanding the issues they’ve caused. That’s usually the nail in the coffin of an organisation’s credibility.

As I think Warren Buffet said, your reputation takes decades to build, but it can be destroyed in just a couple of minutes.

Almost certainly the cost of the blow-back against those organisations is several orders of magnitude greater than the benefits they hoped to gain from their original decision. They tend to be highly RoI-negative decisions that many organisations who go through that process never recover from.

So “stand there” for long enough to make sure you’ve worked all those aspects through. Don’t just “do something”.

The conundrum

The tricky bit is that sometimes taking action gets you a result, and sometimes doing nothing get you a better result.

Sometimes taking action, like pursuing a wildly over-ambitious growth plan, send the business into bankruptcy. Sometimes not taking action and chasing the latest, greatest idea is what saves your business while all the lemmings head off the edge of the cliff together after drinking too much of the same Kool-Aid.

How do you know whether you should be “doing something” or “standing there”?

Well, not to put too fine a point on it, that’s where experience, judgement, and leadership come into play.

None of us are infallible, me included. But I’m definitely better at making those calls than I was 30 years ago, when I thought the world was an entirely logical place and a spreadsheet could give you the answers to everything you wanted to know.

Leadership is a messy place. Running a business is complex.

There are no hard-and-fast rules because if there was, everyone would be doing everything the same way by now.

But at least if you start with the five points above, you’ll get some way down the path and, hopefully, slightly better results in the end.

Knowing when to “do something” and when to “stand there” is a key leadership skill.

Don’t confuse taking action with getting you the results you want.

Sometimes taking action just gets you results you most definitely do not want…the problem is it’s likely to be a few months before the chickens come home to roost, and by then it might be too late to change course.

So, when deciding whether to “do something” or “stand there”, choose wisely.

The finances of many organisations today…perhaps including yours…are organised much like the Victorians organised their factories.

Back in the late 1800s and early 1900s, science had advanced to the point where large-scale industrial production had become possible. Hundreds, thousands, or millions of items could be churned out in a factory, replacing the traditional hand-crafted production methods used for generations beforehand.

For Victorians, science, engineering, and rational thinking was key. All they thought they needed to understand their business was a sufficiently deep knowledge of its physical and mechanical components.

If science didn’t explain it, then it wasn’t worth knowing.

The only problem with this model is that the world is a profoundly irrational place, wrapped up in an occasionally rational-seeming cloak.

To understand the world as it really is, science is about as much help as trying to work out how tall someone is when all you know about them is the colour of their hair.

And that’s particularly true the bigger the problem you’re trying to solve.

Science can tell you how many malnourished people there are in the world.

Thousands of irrational thoughts compounding on top of one another keeps several billion people in a state of perpetual hunger even though the science tells us we just need to give everyone 2000 calories a day, along with a mix of vitamins and minerals, to put an end to hunger for ever.

The problem here isn’t the lack of scientific knowledge. We have as much of that as we need. It’s that the weight of irrationality in the world is vastly more powerful than the results any purely rational, scientific process can deliver.

Sometimes organisations try to counter this effect by trying to make unstructured, subjective data into something that sounds scientific so it can be managed using science (or, perhaps more accurately, the appearance of science).

An example

A great example of this is the NPS, or Net Promoter Score.

That concept is largely responsible for all those customer satisfaction surveys you get when you buy things nowadays, especially online.

There’s a 10-point scale where 1 represents a terrible experience and 10 represents a perfect one. I’m sure you know the sort of thing I’m talking about.

Personally, I’m a big fan of the concept behind the Net Promoter Score, which is that if more people rate your service as excellent than rate it as terrible, you’re probably providing a good service to the bulk of your customers and are therefore likely to get more referrals and recommendations from your current clients, growing your business in the process.

As a concept, that’s hard to argue with.

However, the reality is that people have learned over the years that a score of 7 out of 10 is seen by the company running the survey as “everything was OK”, so unless you feel particularly joyous or deeply upset by your experience, you’re unlikely to rate the experience any higher or lower than somewhere in the 6-8 region.

A small number of outliers apart, what does that tell us? Virtually nothing.

Although I can pretty much guarantee someone in those organisations is doing a lot of spreadsheet gymnastics to fill in a monthly report that says the mean of the responses is about 7, as is the median, and the standard deviation of responses is +/-1 meaning that the majority of responses fell somewhere between 6 and 8.

It’s very rational. It’s very scientific. It involves numbers and spreadsheets, so this must be good information…right…?

It looks rational, but is it really?

If this month’s responses are, on average, “about 7”, and so were last month’s, and so were last year’s, and so on back into recorded history, are you really able to run your business better on the back of that information?

Probably not…

And that’s before we even introduce the idea that tiny, tiny numbers of your customers will ever respond to your NPS survey in the first place.

Organisations obsess about the averages going from 7 to 6.8 and back to 7 again on a month-to-month basis.

They completely overlook the fact that 95%+ of their customers never complete the survey in the first place, so the business has absolutely no idea what those customers are thinking…good, bad, or indifferent.

Of course, science tells us that a representative sample – such as the people who respond to an NPS survey – can give us valuable insights into what the population at large, like an entire customer base, thinks.

Science tells us that, and in the context of a laboratory environment it’s true, within some statistical rules.

You don’t need to eat every cookie from a batch to quality control the output of a bakery. A sampling process, whereby a handful of cookies are randomly taken off the production line for testing, is a perfectly good way to assess the quality of a batch of cookies.

The problem here is that making cookies is a closed system. The bakery is in control of all the ingredients and the production environment. In that context, statistical inference works perfectly well.

But what about if you send a questionnaire to everyone who buys from you and just a couple of percent of your customers bother to send it back. What does that tell you about the mass of people who didn’t respond?

Probably very little.

All that’s happening here is that a thin veneer of science and rational thinking is being wrapped around an entirely irrational, and possibly unjustifiable, assumption that the people who didn’t respond would, on average…had they responded…given responses very much like the people who did respond.

To give just one simple example, it’s more likely than not that busy people respond to NPS questionnaires less often than people with nothing better to do. I remember a stat from a while back which said retired people were more likely to respond to customer satisfaction questionnaires than people of working age.

I can’t quote you the source for that as I’ve forgotten where I read it, but it doesn’t seem entirely unlikely.

That might not be a problem for you if you sell stair-lifts, but if you sell diamante-encrusted cowboy hats for hen parties, this dynamic may be somewhat less helpful.

Embrace the irrational

The main problem here is that organisations are trying to take something fundamentally subjective and irrational and turn it into something with a veneer of rationality wrapped around it.

Now that it looks scientific, we can manage it, organisations say. It can be charted and graphed as much as we like. It can go into RAG-rated KPI reports for the board. The possibilities are endless…

Endless, but not necessarily very helpful.

Some sort of NPS numeric tracking can help make sure you’re not completely screwing everything up. On those terms it can be a useful metric. It’s turning that metric into the be-all and end-all I’ve got more of a problem with.

Concepts like customer service are highly subjective – what you think is a good service and what I think is a good service are two entirely different things.

A company might think that having an always-on AI-powered chatbot available 24 hours a day is going to improve the service they give to clients.

I can’t say this forcefully enough, at least not without using language that would be seen as unprofessional on LinkedIn, but nothing makes me hate dealing with an organisation more than an always-on AI-powered chatbot.

We’re both being irrational here.

The company irrationally thinks that all technology innovations are automatically a good thing, and someone from their marketing or IT department has sold the board on some BS report from a big consulting firm that says “all millennials want to engage with AI nowadays”.

And I’m being irrational in the sense that I’m irrationally expecting the organisations I deal with to service me in the way I’d like to be serviced, when all the available evidence is that most organisations couldn’t care about their customer service in the slightest.

However, of those two points of view, you’re probably better working with the irrational world-view your customers have than with the irrational view your organisation has formed about their customers through a series of “strategy away days”, none of which were attended by a single customer.

After all, your customers have the money. If you want some of it, you’re more likely to get it by stepping into their world rather than forcing them to step into your world.

A simple way to embrace the irrational

Organisations shy away from using “irrational” information because it means acknowledging that they aren’t in control of every aspect of their organisation. On a very fundamental level, many organisations find this a very difficult concept to accept.

But when I’ve been trying to dramatically improve an organisation’s performance, I’ve found that “irrational” information is several orders of magnitude more useful than any more “rational” data.

There are a couple of stages to the process, but don’t worry. This is really simple.

First, you need to accept that you are not the ultimate arbiter of what my experience as a customer ought to be. The only opinion that matters is the opinion your customers have of your organisation, irrespective of whether you agree with them or not.

Secondly, you need to understand that reducing complex questions to a single number for someone to enter on a spreadsheet means you’re asking your customers to make a series of “irrational” trade-offs, so you lose a lot of the subtleties along the way.

To give a simple example, if I rated your service as “not good value”, does that mean – at least in my opinion – your product quality was poor, your prices were too high relative to the quality provided, that I’d have happily paid more if the product was made to a higher standards, or that I’m a natural skinflint and I begrudge every penny I give to someone else, irrespective of the standard of service and quality of product I receive?

If I knew which thought process zapped through your brain as you answered that question, I would get some helpful insights, but just asking a yes/no question or seeking a response on a scale of 1-10 tells me very little.

The third and final stage is to throw out your existing customer satisfaction questionnaires and just ask one question in its place: “What one thing could we do to make your customer experience better?”

Mostly, what you’re looking for here is the outliers.

There will be a few people who wax lyrical about how brilliant their experience of dealing with your organisation was.

There will be a few people who really disliked their experience and who will tell you in great detail everything that went wrong.

Then do something with it…

This doesn’t mean you have to act on every response.

The CEO of Southwest Airlines – the top-rated airline in the US at the time – used to say that the number 1 idea for improvement they had on their customer satisfaction questionnaires was for the airline to serve food on their flights. However he always refused to do that because a fast turnround for their aircraft was Southwest’s secret of success, so factoring in a catering supply element to their turnaround times would have decimated their business model.

So feel free to ignore ideas like that which get to the heart of your business model.

But for most other things, try to amplify the good feedback to encourage more of it.

You will probably be surprised by some of the “little things” people value highly but which you never thought were all that important, or things one member of your staff does, but most don’t, which could easily be incorporated into your standard operating model.

By the same token, do something about the things that haven’t gone well.

Sometimes that’s because there’s a mismatch of expectations which is more your problem than your customer’s. Perhaps your website is full of “top quality” positioning statements, but the reality is you just sell cheap products which aren’t designed to last. Either can the “top quality” messages or sell better products – nothing else is going to fix that problem.

More often, though, it will turn out that some of the “rational” things your organisation does cheeses your customers off in considerable numbers, but you never realised before. In that scenario, stop cheesing your customers off, even if you need to rethink your business model in the process.

I used to deal with an organisation I won’t name here which had a hugely complex sign-in procedure for information which was not exactly top secret material. I suspect if I spent enough time on Google, I’d be able to find the same information freely available.

A sign-on process which takes you to the 7th level of hell and then still insists on texting your mobile to confirm your identity is probably excessive.

All things being equal, they could probably have just texted my mobile and cut out everything else. Having me recall the 7th and 14th letter from my first pet’s name, along with my mother’s maiden name, and the 3rd and 12th letter of my first school did very little to increase the joy I experienced when using their service.

But we can’t do it all…!

When I explain this approach to people, sometimes they tell me they don’t have enough time in the day to take all the responses they receive, read through them, and fix them all.

At some level, I get that. Although that’s usually a giveaway for how much an organisation really cares about its customers – if it can’t be bothered to read through a dozen emails and do something about them, my working assumption is that the true level of interest they have in their customers is somewhere pretty close to zero.

However, at least in the beginning, you don’t need to.

Just pick up one response today and fix it. Any one will do. Select it at random if you like.

Then do the same tomorrow. And the day after. And so on…

After a few weeks, you’ll find that the number of unhappy customers writing in to complain has been magically reduced.

Now, you might find your NPS score hardly changes. After all, only 1-2% of customers will respond anyway, and most people score most things about 7 +/-1 almost regardless of their experience.

So any metrics based around improving NPS scores are likely to be a waste of time. (Unless someone games the system, which isn’t unknown in this area if there’s a big bonus pay-out if the NPS ticks up a bit.)

However there’s a very simple metric you can use instead – how many people have written in to say they’re unhappy about your service this month and have you fixed all the problems they wrote in about. You don’t even need a spreadsheet to track that. Just count them up.

My experience is that a large proportion of the problems people complain about can be fixed almost instantly and at a cost of pretty close to zero. Provided you step into their perspective of their experience and don’t try to force your opinions onto them.

As you fix more and more of those problems, you’ll get fewer and fewer people complaining about them. So you’ll know, even if your NPS score barely budges, that you made an improvement to your customers’ experience of dealing with your organisation.

Yes, some of these things might be “irrational” from your perspective, but if enough of the people who hand over their hard-earned cash to your organisation think something is a problem, don’t waste your energy arguing. Just fix whatever it is and move on.

Don’t think like a Victorian mill-owner, and imagine that enough scientific, rational thought on your part will solve every problem.

In an irrational world, you’ve got to engage with other people’s opinions instead of imagining you have all the answers.

Organisations which get this, and act on the “irrational” feedback they receive, find it remarkably easy to grow their revenues and their bottom line.

Organisations which think getting their NPS from 7.0 to 7.2 automatically means they’re doing a great job tend to spend a lot of time wondering why they aren’t growing as fast as they think they ought to.

Odds are, they’re not being irrational enough.

It’s natural to want to do the very best at whatever it is you do. I know I do and, if you’re reading this, I suspect you do too.

We’d rather be right than wrong. We’d rather be perfect than imperfect. We’d rather deliver 110% than 80%.

As a sentiment, that’s very noble.

And for any personal interests or hobbies you might have, feel free to be as perfect as you like playing chess, carving pieces of wood, or learning a musical instrument.

For business, though, that’s not necessarily what we want at all.

What we want is an output that makes economic sense in the context of the input.

A sentiment I’ve put into a rule I call my Exponential Cost Curve rule.

The exponential cost curve

Why the exponential cost curve matters is that the last few percent of perfection on any project can often cost more than the preceding 95% of the project.

Early on, in most projects, the benefit of doing something – however modest – rather than nothing generally has a highly positive RoI. You’re not spending much, but you’re seeing improvements come through.

But at some point, those costs and benefits equalise, so you get back out pretty much what you put in. Now, there are reasons why you might still want to do that, but few organisations realise when they’re in this zone because most financial decision-making isn’t set up to make this clear.

However the trouble with not knowing when you’re in the “flatlining” zone is that you’re not aware that, just around the corner, is the zone where continuing to spend money will generate a negative incremental return.

That is, not only will you not get any additional benefit from whatever you do in this zone, you will build in diseconomies and disincentives so great that they start to take away from the value you generated up to that point.

Every extra pound you spend reduces the return from the project by £10 or £20.

That’s why I call it my Exponential Cost Curve – taken to extremes, organisations burn through cash in this zone and end up with a worse project in almost every way than they would have had if they hadn’t been quite so keen on achieving some sort of mythical perfection.

Especially common in tech projects

The place I’ve found the Exponential Cost Curve kicking in the most in recent years is in large tech projects.

That’s for two main reasons.

Firstly, and most commonly, because the tech co which developed the solution didn’t understand the real world as well as it thought it did, they built in a range of features which don’t make as much sense in the real world as they do in a Silicon Valley board room.

The trouble is that if you want to use this software at all, you can’t just use the high RoI bits, where you’re doing something rather than nothing and thereby seeing a big benefit from a modest cost.

The software provider has bundled everything up together in a single package and you’ve generally only got a choice between buying all or it or none of it. So to get the “low hanging fruit”, you end up buying an expensive system you might only really want 10% of, but it’s the only way of solving your problem.

So you pay for the 90% you won’t want or need as well, meaning that most of the benefit of whatever you’re trying to do goes into the bank account of the tech co that sold you the solution in the first place, rather than into your bank account.

That’s bad enough, but the second reason the Exponential Cost Curve gets triggered in tech projects is a little more insidious.

Someone – often whoever specified the original project – puts a proposal forward to the board that essentially says, “look at all the exciting things we could do with the 90% of this software’s capabilities that we’re not using”.

Usually the software provider or their systems integration partners are happy to produce all sorts of data showing how much more efficient the business will be if only they start doing things in that 90% space. They can smell a fresh batch of licence fees or a considerably longer customer lifetime value at 1000 paces with a project like this.

That’s particularly the case when the supposed benefits are to do with implementing more structure and control within the software environment and the services which rely on that software.

It’s not a bad thing

Of course, a degree of structure and control is not a bad thing. It’s very hard to run a successful business in an environment of complete anarchy.

But structure and control is something you can have too much of. When you build in inflexibility and rigidity in the name of “control”, the laws of unintended consequences start to flex their muscles because they know they’ll be called into action sometime soon.

To give a real-world example, in my days running a large call centre, we had (as most call centres both then and now do) a call scripting system, which prompted the call centre agent with the questions they needed to ask the customer in an order which someone had programmed the software to ask them.

The agent had to click a “next” button to move to the next question.

The basis for doing this was that it would allow us to control calls better and ensure a more consistent call duration which made resource planning easier to manage. I’m sure there was also some theoretical conversations about how a tighter control on call durations would make sure agents weren’t engaging in idle chit-chat with customers, which in turn meant we would be more cost effective as a business.

While all of that is, of course, entirely plausible in theory, the business had reckoned without one important consideration.

Customers tended not to call up primed to answer questions in the order our software prompted the call centre agent to ask them.

Often, for example, they would start explaining what the problem was – which might have been question 10 in the scripting tool – instead of giving us their customer reference number, which was question 1.

So the agent had to wait until the customer paused for breath before interjecting and saying something like “OK, I understand Mr/Ms Customer, but before we go any further, can I just get a couple of details from you?”

The agent then went back to their script and started at question 1, while doing their best to remember what the customer had already told them about their problem (question 10) which would be coming along 9 questions later.

Inevitably there was more discussion about what the problem was when this conversation got to question 10, a bit of recapping of the story the customer had told previously, and some confirmation of understanding from the call centre agent.

The problem is, all of this took a lot of time in a business where time spent on the phone was directly related to our salary bill, which was far and away the biggest cost the business had.

It was too much control

While the motivation of the people who put the call scripting system in originally was well-intentioned, it came from a place where more control was seen as a necessary element of delivering greater efficiency despite, as it turned out, that not being the case at all.

So, at a time when the call centre industry was hardwiring business processes into software solutions, our operations director, Paul, had a brilliant insight and took us in the opposite direction.

He realised that, while we needed the answers to those 10 questions to enable us to solve a customer problem, the order in which we got the answers was irrelevant, as long as we got them all.

So he re-engineered our call centre agents’ desktops so they could skip from one question to another easily. You could think of this like tabs on an Excel spreadsheet, where clicking a tab instantly takes you to a different part of the spreadsheet.

Now, if a customer started at question 10, it didn’t matter. While they were giving us the answer to question 10, the agent would key in all the details needed for that question. Then they’d let the conversation flow naturally until all the questions were answered.

Think about this for a moment.

We gave up control.

Rather than the conversation being directed and controlled by the call centre agent (or, more accurately, being dictated by the software on their desktop), it was being directed and controlled by the customer.

But get this – on average, calls were much shorter because there was less backtracking, less repeating of information which had already been given once before, when the agent wasn’t on “the right question” to capture the information, and less interrupting by the agent to get the conversation back to flowing in an orderly manner from question 1 to question 10 if the customer had gone off-track anywhere in the process.

Some quick maths

Thanks to this realisation, our average call duration went down almost 20%.

In very approximate terms, we therefore needed 20% fewer agent-minutes to handle the call volumes. And, in theory at least, that meant we could reduce our salary costs by 20% too.

Flipping that maths around, what that means is that implementing the call scripting software to “improve efficiency” and “introduce better control” cost the business 20% more to run than not having that software.

When you have an annual salary bill running into the tens of millions each year, a 20% saving on that is not to be sniffed at.

And that’s what I mean about the Exponential Cost Curve. We used the rest of our CRM system and it worked just fine, but the call scripting software was in that zone where, despite the theoretically attractive business case, it cost us more to run than the benefits it brought to the business.

We had reached the point where a relatively small additional spend to bring in and integrate an extra piece of software had a huge negative RoI, not just on that element of our investment, but on the business as a whole.

That’s what an Exponential Cost Curve does to your business.

And the thing is, it’s usually hard to spot in advance. But the common factors I’ve come across are either in the desire to achieve some sort of perfection (however defined) or as part of the business wanting to exert control over the nth level of detail in its operations – especially when real-life human beings are involved, as they tend to be more random than tech companies think they are.

The hidden benefits

For the business I ran, there was a very real hidden benefit in switching off the rigid approach to customer service – customers enjoyed their interactions with us a lot more. We offered a fast, customer-centric process which sorted out their problem on the spot 95%+ of the time.

By the standards of most people’s customer experiences, then as now, this itself is pretty exceptional.

We won a slew of industry awards on the back of this idea, and others we implemented over the years. And that made it easier for us to attract more business. When you offer a really good service, word has a way of getting out about that.

And if all your competitors are offering a drab, robotic service, frankly you don’t need to try all that hard to be vastly better than your competition. So we really stood out from the crowd.

Now, you may not run a call centre, but I can virtually guarantee you that someone inside your business…perhaps even you…is currently in the middle of a project to either achieve some theoretical level of perfection and/or to improve control, and thereby efficiency, in your business.

I’m not trying to discourage that. We all want to make our businesses as good as they can be.

But what I am saying is that the Exponential Cost Curve – when incremental spend makes the whole business worse – is a very real concept. And it’s particularly real when either perfection or control is being pursued to the nth degree, because it’s in those last couple of percent of implementing total perfection and/or control that you’re more likely to come across the Exponential Cost Curve.

It was very obvious to us in a business where salary costs were such a significant factor.

But it’s not always that obvious.

Maybe you’re just absorbing much higher technology costs than you need to.

Or maybe the inflexibility you introduce into your systems is either increasing business risk to the extent that it’s going to backfire on your business badly one day, or it’s cheesing off your customers to such an extent that they take their business elsewhere.

Rigidity in systems and processes, when carried to extremes, often increases costs dramatically, no matter what the people selling you technology systems designed to implement that rigidity tell you.

Your challenge, should you choose to accept it, is to curb your enthusiasm for the last few % of perfection or control. Often that’s where the biggest negative RoIs are hiding.

It has 288 words, although the word “column” only appears once. Those 288 words are spread across 32 lines. The letter “a” appears 65 times, but the letter “f” only appears 13 times.

So tell me…

How much do you like this song?

“Wait, wait…!” I hear you cry. “But I don’t know anything about this song. All you’ve done is give me some numbers and statistics.”

And yet in a business setting, once we have 5 or 6 datapoints, people are generally quite comfortable making decisions.

That’s especially true for people who inhabit head offices several hundred miles away from the front line of business operations. Things are so much simpler there – that’s where they “manage by the numbers” all the time, perhaps comparing statistics between different subsidiary companies and using those statistics to decide which subsidiaries to keep and which should be sold.

But what if I produced something 320 words long, used the word “column” twice, and spread those 320 words across 38 lines. And imagine I maintained the letter “a” count, but boosted the letter “f” to 19 appearances.

Well, what I’ve written must be better, right?

Across pretty much all those metrics I’ve achieved substantially more than the first song did, haven’t I?

So it stands to reason you must like my song better.

That’s what the numbers say. And numbers never lie…(or so people often tell me…)

The prejudice of perception

OK. I’ll make it a little easier for you to decide if you like this song.

But the minute I give you the next piece of information, you’re likely to jump to a conclusion.

Your preconceived ideas about different musical styles will overlay everything else you might have thought up to this point. It will no longer matter if there are 288 letters in the song or how many times the letter “a” appears.

You might not instantly decide if you like the song when I tell you this. But a lot of people will take an instant dislike to it just based on the next piece of information I give you.

Are you ready?

It’s a country song.

What do you think about the song now?

After all, you have lots of data points about the song and you now know the context for those numbers. That’s more than enough for a decision one way or the other, isn’t it?

Well, if you’re a Brit, odds are you’ll already have decided you don’t like the song because country music is not very popular in the UK. If you’re from the US or Canada, it’s slightly more likely that you’ll be prepared to give it a chance still, but “likers” are probably still in a minority overall.

And that’s the prejudice of perception – something that, whether you like it or not, comes along for the ride any time you see some metrics.

Whether it’s right or not, or fair or not, your mind takes shortcuts to assess whether something is good or bad, even if you can’t articulate a logical process by which your brain reached that conclusion.

“It’s country music” is enough for most people to decide they wouldn’t like the song, even though the average Brit doesn’t listen to enough country music to be able to make a rational decision one way or the other.

Mind-bending time

In the interests of full disclosure, I sold you a bit of a dummy there. But I did it on purpose to make a point about perception.

It would have been more accurate of me to say that the song I’m writing about here started out as a country song. However, it became a massive crossover hit in the pop charts.

And what if I told you that, by the early 2000s, this song became the second most-played song on US radio stations, beaten only by the Beatles’ recording of “Yesterday”.

Now do you think you might like it?

After all, radio stations have no incentive to play records their listeners don’t like. If listeners get bored and tune away before the next commercial break, that costs the radio station money in reduced ad revenues.

So millions of people must enjoy listening to this “country song”. Presumably even people who, like you, probably, “don’t like country music”.

How does that work?

Well, in decision-making terms, we’ve introduced some social proof into the equation. Social proof is very powerful in swaying people’s decisions in the direction the person sharing that information wants to sway them.

And, probably, with that knowledge under your belt, you’re now feeling much more positive about that 288-word song. After all, millions of people can’t be wrong…right?

Let’s layer in something else

Now it’s likely you’re feeling much more positive about the song, let’s layer in something else.

The person who had the big crossover hit with this song didn’t write it. The original singer/songwriter had a very modest level of success with his version, but the person who took the original version made the tiniest of tweaks before re-releasing it. That’s all it took to turn the song into a million-seller.

However the person who had the big hit was pretty much unknown by the general public when they released their cover version.

They were an experienced, and highly-regarded, musician in their own right. But they worked as an uncredited studio musician who beavered away in the shadows to produce records which made other people famous. Generally, they didn’t even get a mention on the liner notes.

So, this song didn’t become a million-seller because a famous name muscled their way into the process, short-circuiting the hard work needed to sell millions of records.

I wonder how you feel about the song now.

Odds are, a lot more positive.

You’ll recognise, given this background, that the song and the performance must have been exceptional to get any traction at all against a sea of indifference from the record-buying public. For a little-known song to be turned into a global hit record from those inauspicious beginnings, it must have been something special.

So let’s knock your confidence again

Now that you’re feeling much more positive about this mystery record, let me introduce just one other piece of information.

This is also likely to polarise opinions…so, fair warning. All the positive feelings you’ve been building up might be about to take a knock.

Here it is…

This 288-word masterpiece features a banjo throughout.

Now, I quite like a banjo, but it’s an instrument most people have strong opinions about. Mostly negative ones.

It’s something of an acquired taste, I grant you. But if you check your thinking process so far, you’ll probably find that, starting from a low base, your increasing sense that the song I’m describing is likely to be good has probably taken a big knock in the last couple of paragraphs.

However none of the original data has changed.

The song still has 288 words. It still has only one instance of the word “column” and 13 instances of the letter “f”.

Everything has remained exactly as it was in the very first line of this article. All I’ve done is play with your perceptions about whether this mystery song is likely to be any good or not.

And I’ve done this to illustrate the problem of just “managing by the numbers”.

Of course numbers are important

I’m not suggesting for a moment that numbers aren’t important. I’ve built a career based on compiling and managing numbers.

But I find, in corporate settings, people make two common mistakes in the way they look at the numbers on their reporting dashboards or KPI reports.

Firstly, numbers are used far too superficially in many review processes. There’s no possible way of knowing whether a 288-word song is any good or not without some additional information.

The one thing we know for sure, though, is that more words doesn’t necessarily make a better song. Yet in a lot of organisations, the only objective that matters is to make whatever the numbers were last year into bigger numbers this year.

That’s unlikely to be your route to successful world domination.

Secondly, the measures we use tend to be too inward-looking. The numbers I gave at the start of this article are equivalent to the data spat out by the average corporate information management system – true, as far as they go, but shorn of so much context that they aren’t really much help in managing the business.

When I told you that the song was a million-seller, and the second most-played song of all-time on US radio, your perception of whether or not it was likely to be a good song shifted dramatically.

And while I apologise for bringing the mood down a little by mentioning the banjo at the end, that was just to illustrate that more internally-generated information alone is probably not helping you make a decision about how good a song is.

Yet, when organisations manage metrics and dashboards, they tend to look for more and more internally-generated information – with a huge investment in time and money, it has to be said – when comparatively small amounts of information from outside your business might give you all you need to know about how well something is doing.

We don’t spend nearly enough time tracking down external perspectives, and far too much time chasing down the tiniest, and largely irrelevant, details of internally-generated information.

Oh, and before you say it, doing things like tracking NPS scores or sending out customer surveys – while better than doing nothing at all – are not really the answer.

NPS has its uses, but the way the process is used inside many businesses is to turn some small external data points into internal data which, shorn of context and meaning, is probably not much help in the decision-making process, over and above what you already know.

The challenge is to go out much wider for information without forcing it to fit into a KPI dashboard-friendly format.

The search for meaning

Any set of metrics, when done well, are about conveying meaning. They’re not just about conveying numbers.

When you understand the meaning behind the numbers, you need comparatively few metrics to run your business.

And that’s not easy.

For example, the song I’m talking about is, to me, a desperately sad song about a love that should have been, but never was.

But I know someone who thinks the song is an inspirational tale of a love that a person carried in their heart their whole life through, no matter the slings and arrows that were thrown in their direction along the way.

To be fair, both meanings are entirely consistent with the lyrics. But whether you take one view or the other will determine which emotions come to you when you hear this song.

And the same is true of your metrics and KPIs. On their own, they’re fairly meaningless unless you know what you want to do with them, and the objective you’re trying to reach.

Most FTSE100 chief executives wouldn’t think that being rude to their customers on social media was a positive for their business.

Yet Ryanair have turned that into an art form and their business has grown despite…or perhaps because of…having a very different attitude to their customer satisfaction statistics than most other public companies.

What a metric means is much more important than what the number is.

Quals often beats quants

Unless you’re running an infinitely repeatable process, like mixing together a handful of chemicals to make a shampoo, or building a new car on an assembly line, quantitative information can only take you so far.

Quantitative information like the 288 words, or the number of times the letter “a” appears in a song have very little bearing on a “soft metric” like “did you enjoy the song or not?”

Mostly, what we want to know is whether or not customers enjoyed dealing with us.

A few metrics along the way might provide helpful information for controlling our own internal processes.

But ultimately no amount of KPIs can definitively answer the question “did you enjoy dealing with our business?”…even though that’s the only question that really matters.

Far too often I’ve seen businesses in real trouble convince themselves there was nothing to worry about because all their quantitively-based KPIs were pretty much where they wanted them to be.

One time, I did some work for an organisation where customers described the atmosphere as “unwelcoming and occasionally hostile”.

No amount of green-rated quantitative KPI scorecards can counteract that qualitative perception of an organisation.

Yet, the people who complained were seen as “troublemakers” or “people who we wouldn’t have wanted to deal with anyway” because this organisation was so focused on its numbers it lost sight of what really mattered. This qualitative information was dismissed as just anecdotal and not consistent with data on the KPI scorecards, so therefore irrelevant.

Eventually, this quants-only approach to performance management drove the organisation to its knees.

Things might have been different if they’d applied an adage from Jeff Bezos: “When the data and the anecdotes disagree, the anecdotes are usually right”.

Or, as I might put it, numbers alone only get you so far. Beyond that point, you probably don’t need more numbers. You probably need more qualitative information, external to your organisation, because that’s usually where the real information lies.

---

I know I’ve kept you waiting far too long for this, but if you are still wondering what song I’m referring to it’s “Gentle On My Mind”. Originally written and performed by John Hartford, and made into a million-seller by Glen Campbell.

And if you’re wondering what the little tweak was that Glen Campbell made to achieve million-selling status, it’s this…he just played it a little bit faster.

If you need a reminder of the song, or want to reacquaint yourself with the joys of some banjo playing, you can find it here. (And the original version is here, if you want to compare and contrast.)

As the business world has become more and more metric-focused over the years, it’s becoming ever-more apparent that just having more metrics at your disposal does not make businesses better.

I say this with the greatest of respect to numbers and metrics. After all, I’ve had a 30+ year career on the back of being good with numbers and metrics.

But, in the hands of people who take them too literally, and apply them formulaically without really understanding what’s going on, numbers and metrics can be dangerous and lead to poor management decisions.

This problem has compounded as today’s computerised monitoring and tracking systems can tell you pretty much anything you want to know about your business in an instant.

You can set up systems to ping you with a daily report on anything you want. Weekly meetings with your direct reports probably start with a trawl through their KPIs. Packs for your monthly board meetings are filled with more numbers than ever.

But is business getting harder or easier on the back of all these metrics?

Well, I don’t often come across people who think business is easier to manage today than it was a couple of decades ago.

And I suspect a large part of this is that businesses today try to manage complex outcomes using methods which only work for simple processes.

Sometimes it’s fine

The way most organisations manage their metrics is fine for single-dimensional issues. It’s just not a good way to manage anything more strategic.

A single-dimensional issue is one where all the elements are under your control and you can assess performance with a couple of simple yes/no questions.

Let me give an example. Some years ago I worked with a business which despatched packages to their clients overnight. On-time delivery was important to their clients’ business model, so the business I worked for used a well-known, and not-inexpensive, delivery service which promised to get packages delivered to our clients by 10am the following day if they were available for collection from our factory by 6pm the evening before.

Now, for the courier company, I’m sure there were enormous practical challenges to be overcome to deliver on that promise. But for us, it was really simple.

Since this particular business “absolutely, positively” promised to deliver by 10am, the only question we needed to know the answer to was “did they?”.

To be fully transparent, on very rare occasions, they didn’t. But we had something like a 99.8%+ on-time delivery track record and that’s what we told any new clients coming on board with us.

The key point is that on-time delivery was a simple metric to track for this business. An important metric for our clients, for sure, but a simple one for us to track and manage.

Nobody in that business spent a lot of time or energy worrying about whether deliveries were going to get to their intended destination on time. They just did. And we tracked the deliveries on a weekly basis to make sure we stayed pretty close to 100%.

But for us, there were very few moving parts to manage. As long as the boxes were on the loading dock by 6pm, they’d be collected by the courier company and delivered to our client the following morning.

That’s a single-dimensional metric. All we needed to know is “did the parcel arrive by 10am?” which we knew from the signed paperwork we got back from the courier company with their invoices as proof of delivery.

(To be fair, on the rare occasions deliveries weren’t made by 10am, we normally had someone from our client on the phone shouting at us long before the paperwork turned up from the courier company.)

In this context, “single-dimensional” doesn’t mean “unimportant”. On-time delivery was very important to our clients.

It just means that the elements were under our control.

Deciding whether or not the box got to the loading dock by 6pm was a simple yes/no question. As was “did it get to the client by 10am the next day?”

We could probably have tracked several dozen more metrics if we’d put our mind to it, but the reality was that none of those other metrics would have done anything other than corroborate what we already knew – that our client got their deliveries on-time very nearly 100% of the time.

That, in turn, meant that the time, cost and effort which would go into tracking any other metrics in this area would not add any more value to what we already knew about our clients’ on-time delivery experience.

So we didn’t track any other metrics.

We had everything we needed, and nothing we didn’t need.

It was also the lowest-cost way to assess our on-time delivery performance. Anything else we tracked would make the process more expensive and convey no additional useful information in the process.

Multi-dimensional issues

As a way to manage single-dimensional metrics the approach above is fine.

The problem is that most of the issues you spend your time dealing with in a leadership role are not as simple to manage as our on-time delivery performance was.

And the “right answer” isn’t always obvious. At least, not at first glance.

Imagine your call centre is going crazy. All the calls are backed up, and your customers are furious.

All your call centre’s performance metrics look terrible.

If this was a single-dimensional issue, you’d probably just fire whoever was looking after your call centre and bring in someone who knew what they were doing.

However, call centre metrics alone are unlikely to give you the full picture in this scenario.

Assuming you haven’t hired a complete idiot to run your call centre, it’s much more likely that something else has gone badly wrong in your business. But it’s pretty unlikely that any of the metrics you currently collect across your business will tell you it is.

Perhaps the marketing department launched a new national campaign and forgot to mention it to the call centre, leading to understaffing relative to demand in the call centre, even if the staffing levels were perfectly sensible based on the activity the call centre management team were aware of.

Perhaps the last batch of product you manufactured was terrible, so all your customers are calling to complain and demand their money back. From refund requests being so infrequent they didn’t have a major impact on call centre resourcing, hours of staff time are now being spent to process all the refunds and product replacements. This means the regular customer call-load is being neglected, resulting in them calling in to complain…and so on, and so on…

Perhaps the IT department’s recent software upgrade has made all the call centre’s CRM system run much slower than it used to. IT focused on the technology aspects but never really tested the impact on a call centre agent’s desktop “because the supplier said we needed to upgrade to the latest version of their software”.

Except now a 2-3 minute average call has turned into a 10-12 minute average call which means nobody’s calls are being answered in the usual timeframe.

The point here, though, is that none of the metrics you’re collecting for your call centre – or anywhere else in your business, probably – will give you the insight you need to work out what’s happened in any of the scenarios above.

Twice the number of metrics won’t tell you any more than you already know. You really are in deep into diminishing returns territory if you layer more metrics into your call centre reporting.

Finding the signal amongst the noise

The skill more organisations need more of is finding the signal amongst the noise. The one thing that really matters out of the thousands of pieces of information coming your way.

Originally an engineering term, the concept of “signal vs noise” is popular among financial traders where it’s all about how you find the one true nugget of information in the middle of information overload from all your data platforms and news feeds.

Yet this concept is not applied to the metrics organisations use to manage themselves as often as it might be.

By collecting more and more datapoints, all most organisations are doing is increasing the amount of noise and making the signal even harder to spot. Oh, and spending a lot more money in the process into the bargain.

Some people might claim that “AI will fix that”. I’m doubtful – and if it fixes the problem at all, it’ll be years before it does this well.

Like all computer systems, AI works on a purely logical level. If this, then that. If A, do this, if B do that.

Sure, AI is a flashier interface, but all IT-based systems have to work entirely logically because no-one has found any other way to make computers work yet. The supposedly smart things AI does is just because some people have worked out a way to apply a logical process to produce a seemingly “beyond logic” result.

But whether or not you think AI will help, it suffers from exactly the same issue as most human reviewers of metrics.

It starts with looking at an organisation as a series of discrete silos, and concentrating any review process on looking deeper into those silos. That’s the logical approach, after all…drilling down into all the available information.

“Problem in the call centre? Let’s call up even more call centre metrics and see if we can work out what happened.”

More likely, the real problem is somewhere else in the business and it’s just the call centre’s misfortune to be “downstream” of whatever that problem was. To solve the problem you need to look more widely, not more narrowly, and you’ll probably find that the metrics you collect, whether in the call centre or anywhere else, on a daily/weekly/monthly basis will only get you so far.

The signal you’re looking for is unlikely to be found within an avalanche of purely call centre-based statistics.

Your mission…should you choose to accept it…

The objective for any performance management system should be to have as few metrics as possible. Too many metrics, especially if they’re just corroborating information some other metric already provides, is a waste of time, effort, and money.

Just because your IT system tracks a feature automatically, and can produce a report on it, doesn’t mean you should track, measure, and manage your business using it.

Every metric over and above the minimum required increases the cost of running your business. It doesn’t make your business better, or easier to manage – usually the opposite, in fact.

And you need to be rigorous about the RoI on any metric.

Many times I’ve seen organisations able to get 95% of the information they’d ideally like pretty simply, easily, and inexpensively. They then spend six or seven figures a year to get that up to perhaps 97% or 98%, accepting that 100% is unachievable as a goal for a whole host of technical and practical reasons.

In reality, are they going to make any different decisions with 97% or 98% of the information they’d ideally like vs the decisions they would have made with 95% of it?

I’m not saying it’s impossible, but I’m deeply sceptical.

More likely, if the business just spent a fraction of the extra cash they spent to get an extra percent or two closer to factor in, and budget for, a little extra risk management in their decisions and processes, that would give a better bottom-line outcome for the business, and avoid an over-proliferation of metrics which add very little to what was already there, but at huge expense.

If you want to get your business off to a strong start for the new year, take a look at the metrics your organisation collects and go through them with a fine-tooth comb.

When you really challenge the assumption about more data necessarily being a good thing, you might find your organisation tracks a remarkable number of metrics which add significant cost and complexity for remarkably little bottom-line benefit…if any at all.

Then ask yourself: if anyone had presented a budget proposal to you for collecting that metric now you know the true RoI, would you have green-it the project?

You might be surprised by how often the answer, in the cold light of day, is “no”.

Most organisations like to optimise things. Which is fair enough – why wouldn’t organisations want to be the best at what they do?

But there’s an interesting dynamic at work here. Often the root cause of an organisation’s implosion is because they’ve been a bit too keen on the optimisation front.

That’s generally for one of two reasons.

Either they try to optimise everything or they pick one thing to organise, which turns out to be the wrong choice.

Optimising everything is a fool’s errand

Of course, there’s a superficial rationale for optimising everything. Sometimes it’s one of those well-meaning activities that “seems obvious”.

At other times, an organisation tries too hard to beat the competition and uses optimising as a blunt instrument to do that in every aspect of their operations, instead of thinking things through properly and establishing some meaningful differentiation with their competitors.

The reason trying to optimise everything is a fool’s errand is because pretty much every organisation is trying to accomplish a range of goals which pull the organisation in entirely different directions.

Here’s a simple example.

Your business gets a service call from a client because the equipment you installed broke down. Your service contract promises fixes “normally” within 48 hours, but doesn’t make a firm commitment to do that.

It generally takes two hours for an engineer to get the equipment up and running again, but this client is two hours’ drive from you which means it’s likely that you’ll incur a full day’s costs for your service engineer, but only be able to charge for 2 hours of their time.

Normally, you try to get a number of engineer visits in the same area on the same day to minimise travel time and maximise your ability to charge clients for your engineers’ time.

So the question is, do you send an engineer or not?

And the answer is…well, it depends on what you’re optimising.

If you’re optimising for service, you send an engineer the minute you put the phone down.

If you’re optimising for short-term profitability, you wait for a day or two in case another customer in the same broad geographical area has a problem they can fix in the same trip, and then reluctantly send an engineer anyway if one of those happy opportunities doesn’t come along.

But note this: you can’t optimise for both of these outcomes simultaneously. Doing one guarantees you can’t do the other.

A lot of objectives in the corporate world are like this. Organisations are often a series of silos each incentivised on their own metrics, so they spend large parts of their day fighting with other silos over whose targets are going to take a hit every time there’s a choice to be made.

In the example above, the Finance Department and the Service and Maintenance Department are each going to see this situation rather differently. The CFO will be incentivised on bottom line profits. The Head of Service and Maintenance on response times and client satisfaction scores.

(NB: I deliberately said “short-term profitability” above – in the long run, you will almost certainly experience higher profitability by serving your customers better, even if it costs you a little more in the short-term.)

Optimising everything is pretty much impossible. More likely, all you’re doing is setting up your senior people to spend their time scrapping with one another to protect their bonuses than pursuing the organisation’s objectives.

And the opportunity costs of two senior people scrapping it out are likely to be vastly higher than any benefit the business could ever gain from optimising either of those two perfectly reasonable objectives.

The “one objective” fallacy

Another common approach is to have one overarching objective that everyone is focused on, to the exclusion of pretty much everything else.

Now, you won’t normally work out what that is by looking at an organisation’s website or studying their annual report. Those are more outlets for corporate PR than helpful information in matters like these.

Rather, you have to work it out by watching what organisations do.

Now, I have no idea if this is true or not, but my reading of the situation is that recently one of the world’s largest and most profitable companies has put its very existence under threat by appearing to optimise for one single objective to the exclusion of everything else.

You’ll have heard of this business. It’s called Google.

By any standards, Google has been an exceptional business over the last 20 years or so. In a remarkably short period of time, it’s gone from being an idea in an academic paper written by Google co-founders Larry Page and Sergey Brin to becoming one of the largest and most powerful companies in the world.

Along the way, their business has entered everyday language. Just 30 years ago, nobody Googled anything, because the term didn’t exist, but now “googling” is a word in the dictionary. And, at its peak, Google accounted for well over 90% of all web searches.

So I’m not dunking on Google. They’re an exceptional business.

However, over time, Google would appear to have optimised for a single measure – ad revenue from search results.

In the early days this was pretty harmless. Nobody minded an ad or two in return for a fast answer to their questions from a reasonably reliable source.

But that was so profitable, Google put more and more ads in their search results. From a position on the right-hand sidebar, they started appearing at the top of the search results.

Then more and more ads appeared at the top of the search results, often driving organic posts “below the fold” so you had to scroll down to see them. But people kept clicking on the paid ads, so more and more businesses ended up giving money to Google on a regular basis.

In the short-term, that was great business for Google. They had no meaningful competition and oodles of cash poured into their bank account every day of the week with very little effort on their part. Great times…if you were Google…

If you weren’t Google, however, it wasn’t so great.

People trying to sell you stuff crowded out the information you were looking for. Information you would have found easily on a Google search five or ten years previously became progressively harder and harder to find.

So the average person’s search experience got considerably worse.

Roll the clock forward a few years and in more recent times Google has found itself under pressure from a huge number of AI companies who claim (although, frankly, I’m sceptical about most of them) to give you instant answers to all your questions.

For the average consumer, this means you don’t need to trawl through 10-20 search results of distinctly varying quality, alongside a similar number of ads, to find what you’re looking for.

So why would anyone Google something for a poor experience when they could ask ChatGPT/Perplexity, etc the same information and get an instant answer without any ads (for now, at least).

Let me say, I have no insight into the thinking and strategy of Google on this topic. But as an outsider, it seems to me Google have behaved like a business which was trying to maximise ad revenue at the expense of just about everything else would behave. So my working assumption is that’s probably exactly what they’ve been doing.

Fast learners

To be fair to Google, they are fast learners.

They developed at lightning speed…or released something they’d had for years but kept secret in order to protect their ad revenue, depending on how much of a conspiracy theorist you are…an AI product of their own, Gemini. Which is actually quite good.

Now a Google search shows results from Gemini first, then some ads, then the organic results.

The only difference is that now Google are scrapping it out with Microsoft’s CoPilot (which is also quite good), ChatGPT, Perplexity, Claude and whatever else is the AI flavour of the day.

They’ve gone from having an impregnable position in online search to having to fight over a market they used to call their own.

While I don’t think Google is likely to go out of business any time soon, they will almost certainly never again have the share of search and the endless waterfall of cash into their bank account they had in the pre-AI days.

Some people might think that’s a good thing. After all, the principle of creative destruction is hardwired into capitalism. If you don’t keep up with your market, sooner or later, you’ll become an irrelevance. That’s how the system is supposed to work.

But imagine this…

What if Google had optimised its business for search experience rather than ad revenue?

Sure, it might have forgone a bit of ad revenue in the short-run, but it would have meant that there would be very little opportunity for the AI companies to encourage people to cut Google out of the equation altogether and go straight to them for a simpler, faster, better answer to their questions.

Odds are Google would still have well over 90% of the global web search market.

And…if some people are to be believed and Gemini had existed “secretly” for quite some time before being released to the public…had Google released their own AI-powered search experience while owning 90%+ of the search traffic, would ChatGPT and other AI companies have got off to such a rapid start?

I suspect not.

The AI upstarts would have been pushing hard to solve a problem pretty much nobody had, because a Google search experience with fewer ads and built-in AI would have given the average person everything they needed.

By prioritising short-term profitability, Google has probably put a big dent in its longer-term profitability.

Maybe one day we’ll even speak of Google the way us old-timers speak of AltaVista or Yahoo.

Google blew them out the water with much better search results and a much better user experience. Could someone else do the same to them?

Well, I don’t know.

But whereas five years ago I’d probably have said that was impossible, now I’m not so sure. By at least appearing to optimise a single, short-term objective, namely ad revenue, it’s possible that Google have sown the seeds of their own demise in the longer term.

What to do instead

So, if it’s not a good idea to optimise everything, and it’s equally not a good idea to become obsessive about a single measure, what should you do instead?

Well, there are two answers to that question.

There’s a simple one and a more complicated one.

The simple answer is to optimise the experience you give to your customers. The better you serve them, the less likely they are to go elsewhere and the more likely they are to give you whatever you want (within reason) to keep serving them in the way they’ve become used to you serving them.

Yes, you need to keep an eye on your costs and manage your business professionally, but as long as you raise your time horizon beyond just this week, this month, or this quarter, that’s not as difficult as most people seem to think it is.

And you get big benefits from this approach, which are not usually brought into consideration when someone gets the spreadsheets out. For example, you are much more likely to get a stream of pretty much free, high-converting referrals from existing customers if you provide a great service vs a poor-to-average service, although that never shows up in an RoI calculation.

The more complicated answer is to have a hierarchy of principles that guide your decisions.

You might say, for example, that you always put customer service first – and do whatever it takes to wow them – unless it’s going to cost more than $1,000, in which case you need to get a manager to sign off your proposed actions in advance.

You could take that a step further and add in that, in addition to the $1,000 limit, a decision can’t result in any of your service engineers working more than an average of 40 hours a week, measured on a rolling monthly basis, without also getting approval from a manager.

Now, you don’t want to have too many layers to this or it becomes unmanageable.

But it’s entirely possible to have a three or four-layer hierarchy that governs your decisions…or at least 90%+ of them…without requiring an enormous management overhead to keep track of everything and/or to referee fights between different members of your senior management team who are each incentivised differently, and in a mutually-exclusive way.

The key, however, in the words of Michael Porter, is to recognise that strategy is about choices.

What you decide to do…and often what you either decide not to do or ignore completely…ends up determining what sort of business you end up with.

There’s nothing wrong with maximising profitability. But if you over-maximise it, you’re probably just opening up a space for a competitor to step into.

Away from the tech world, I can’t help noticing that every major UK supermarket now uses “price-checked with Aldi” as a shorthand way of saying “you couldn’t buy this cheaper anywhere”.

Aldi (and Lidl) only exist in the UK grocery market today because the established operators optimised their performance in a pretty uniform manner (one reason I’m deeply wary of adopting “industry best practice”).

The more they became like one another, and the higher they jacked up their prices, the bigger the market the major supermarkets opened up for their “upstart” competitors.

From pretty much zero a few years ago, Aldi and Lidl between them now have almost 20% of the UK grocery market.

At a time when household budgets are under pressure, and the cost of living is a frequent headline in our daily newspapers, it turns out that two German companies who optimised for low cost over whatever Tesco and Sainsbury were optimising for was a winning strategy.

For a while Google, Tesco, and Sainsbury looked like they were impregnable.

But in over-optimising to compete with one another on a broadly similar basis, all the major supermarkets did was open up a fair chunk of a market they’d previously had to themselves to a couple of upstart competitors.

So be careful what you optimise for.

You might win a battle. But you could increase your chances of losing the war.

In a business of any size, you’ll find no shortage of processes. There’s a process to get a budget to hire a new member of staff. A process for submitting an expense claim. A process for annual appraisals. A process for…well, you get my point. There’s a lot of them.

But what if I told you that processes are often unnecessary, invariably costly and, more often than not, deliver sub-optimal outcomes?

Yes, I know every bit of business education you’ve ever had has emphasised the importance of repeatable processes.

Every consultant you’ve ever hired has insisted on implementing processes to control your business in excruciating detail.

Every business book you read, seminar you attend, and podcast you listen to probably goes on at some length about the importance of running your business on standardised, repeatable processes.

It’s not always crazy

Despite what I’m going to say in a moment, having repeatable, standardised processes in your business is not completely crazy. There are some parts of your business where you absolutely should have a buttoned-down process to manage what happens.

If you’re assembling cars, there’s no need for a debate as each new vehicle comes along the assembly line about whether the wheels go on first or the dashboard goes in first. Once you work out there’s an optimal order to deliver a finished car as fast and as cheaply as possible, then by all means “ink in” those processes as “the way we build a car around here”.

Similarly, if a function that involves a degree of regulatory oversight – especially when that oversight comes with a punitive penalty regime for getting it wrong – then you absolutely should nail down exactly what needs to happen to make sure nothing gets overlooked.

Yes, it’s a pain if you’re buying a house and your solicitor demands to see your passport, a utility bill, your bank statements and excruciating levels of proof about where the money you’re using to buy your house has come from.

But given that your solicitor can lose their licence to practice, and potentially end up in jail if they don’t carry out those activities to a forensic level of accuracy, it’s understandable that law firms have extraordinarily tight procedures governing how lawyers do this.

And in health-and-safety critical environments, there is also a good reason to be stringent about activities like changing out of the overalls you’ve been crawling around the inside of a nuclear reactor in all morning before you nip out to Asda to pick up a sandwich for lunch.

For activities in those categories, having a standardised, repeatable process is generally an excellent idea.

The problem is when that spills over into other areas of your business. That’s when you tend to lock in high levels of cost and achieve sub-optimal outcomes.

And, in most businesses, that’s probably 90% or more of the things your people do every day.

This is problematic in all sorts of ways, but today I’ll just highlight three of the main problems an overly process-driven approach can lead to.

Problem 1 – the set-up costs

The first way in which processes add cost into your business is that someone – usually a relatively senior executive or a high-priced consultant of some sort – spends hours, days or months creating elaborate processes to follow in a given set of circumstances.

And odds are, while they’re doing that, they’re not doing anything else. So your opportunity costs mount up fast.

If your Finance Director is writing out a process for making expense claims, they’re not doing more productive things like working to reduce your business’s cost of capital, accelerating cash flow, or reducing operating costs.

I understand the argument that, since you’re paying your Finance Director a salary, it doesn’t cost you anything “extra” if they write up your expense claim process.

However if you don’t understand the concept of opportunity costs well enough to know that asking senior executives to spend their time on relatively trivial activities is crazy, you probably need to get yourself a much better Finance Director.

At the rates you pay senior staff or external consultants, every time you set up a process of any sort, you’ve probably spent into five figures at least. More in larger organisations where there are a lot more stakeholders and an exponentially higher level of operational complexity.

So you should always ask yourself “does the activity we’re about to develop a process for show a positive payback over the £10k plus we’re likely to spend developing it in the first place?”

Very often, the answer is “no”.

Which means, if you do it anyway, you’ve just spent £10k for a return of less than £10k.

Often, in my experience, for a return of £0. It’s just like rearranging the deckchairs on the Titanic.

Yet, for some people, making dreadful RoI decisions like this has Pringles-like levels of addiction – once they pop, they just can’t stop.

Sadly, just like eating a full can of Pringles on a regular basis, this behaviour will make your business fat and bloated. And, ultimately, broke.

Problem 2 – the repeatability illusion

Assuming you pass the first threshold, and think there might be a positive payback from implementing a process of some sort, the next area organisations trip up on is the “repeatable” part of “repeatable process”.

The fundamental error organisations make is in thinking activities are repeatable when they’re nothing of the sort.

When a business is completely in charge of all the elements of the process, it’s more likely to be genuinely repeatable.

But if, for example, you introduce a random member of the public into the equation, or you sometimes buy this sort of raw material and sometimes buy that sort of raw material, or sometimes deliver to one client specification and sometimes deliver to a different client specification, the likelihood of there being a huge amount of repeatability declines dramatically.

The other important element of repeatability to bear in mind is that an activity needs to be happening pretty frequently in your business for repeatability to even stand a chance of being a major factor.

I’ve often worked with organisations who claim to have a process for compiling their annual budget, for example.

Well, by definition, that’s something most organisations only do once a year.

By the time next year’s budget rolls around, most people will have forgotten what they did a year ago. Some of the people who do next year’s budget won’t have been in the organisation this year. Or perhaps they were, but weren’t at “budget-responsible level” last year.

Those people aren’t repeating anything. They’re doing for the first time

Frequency is also an important element in developing repeatable processes.

For an activity to be worth developing an in-depth process for, someone needs to be doing it several times a day, ideally, or at least several times a week.

At that level of frequency, you’re more likely to be able to generate cost savings which deliver a positive RoI on the cost of developing and managing a process.

You have activities taking place often enough to use proper statistical analysis to identify the most cost-effective way to achieve your ultimate goal.

The volume of activity is likely to be high enough to make a measurable difference to your business operations after you’ve fine-tuned it to the nth degree.

But if you don’t have an activity which you do repeatedly, then using a repeatable process to get the answer you want is not the magic bullet all those podcasts suggest it is.

Problem 3 – an addiction to control

The reason many business processes I’ve come across in my time are implemented are nothing to do with running a cost-effective business They’re entirely about implementing a system of control.

And while sometimes that’s necessary (the lawyer’s proof of funds process or the nuclear power station worker nipping out for lunch), mostly it isn’t.

When processes are put in place purely as a mechanism of control, I can virtually guarantee you that your business is operating at a much higher cost than it needs to, and also that it’s achieving sub-optimal results.

This “high cost / worse results” scenario isn’t one I usually recommend.

I heard a great example of this recently on Giles Edwards‘ excellent Call to Action® podcast with Mark Denton.

Mark described the process-driven environment in today’s advertising agencies, which typically involve lots of meetings.

From the organisation’s point of view, a process means they can control each project tightly. Every one of them starts at A, goes through B, C and D, before eventually popping out at Z with a completed piece of work.

Someone important can have dashboards of what jobs are at which part of the process. No doubt there are project managers (NB extra overhead!) managing all those jobs through. Reports are being written. Dozens of people are involved. Costs are racking up faster than you can count them.

Mark recalled the “glory days” of advertising when an entire campaign might be developed from scratch over a few days by a couple of people who had been let loose on a problem.

Now, I’ve worked in advertising agencies, so I know it doesn’t always work like this. But let’s imagine the inspiration strikes and an art director and a copywriter can develop an entire campaign between them in the course of a few days. The cost to the business is a couple of days’ salaries.

Alternatively, you might adopt a more structured process.

However, by the time you hard-wire in a meeting cadence, a team of people each representing a different stakeholder group, a project manager, a regular reporting system, some managerial oversight, and goodness knows what else, your costs have exploded relative to a couple of people sat next to each other in a studio for a couple of days.

A rigid process also hard-wires in a timescale. You can’t do anything by the end of the week any more because you’ve designed a process that, by definition, will take weeks or months to run through.

You’re also covering a lot more salary costs because a cast of thousands is now involved in the project, instead of two people.

None of that’s good. But we haven’t even got to the worst bit yet.

The worst bit is that the more people you involve in doing anything, the more likely it is you’ll end up with some grey, average result.

Day-long strategy meetings, mandated by some process or other, with a dozen people in them are rarely the places you’ll find high levels of creative insight. Mostly the attendees are just trying to stay awake.

So a rigid process inks in a longer timescale and a higher cost-base, as well as delivering poorer results.

Why would anyone not addicted to control think that was a good set of outcomes?

And yet, that’s a remarkably common way organisations run today. Even organisations which have nothing to do with advertising.

When you prioritise control over results, your costs are higher than they need to be and your results are generally worse.

Maybe those days are gone

For a host of reasons, I doubt the world will be run the way London advertising agencies were run in the 1980s any time soon.

And I’m not suggesting you should, particularly. At least not all the time.

For me, it’s more about being rigorous in identifying the costs and benefits of implementing processes.

No process is “free”. There is always opportunity cost at the very least.

And carried to excess, I’ve seen organisations which ended up building in several layers of management just to manage all the processes they’ve inflicted on themselves, all the time wondering why they find it harder and harder to make money nowadays.

Where you’d happily spend money on the process – such as the lawyer who is keen to keep their licence to practice and stay out of prison – then go right ahead and implement a process.

If every 10 seconds you shave off the production time for a new Nissan puts £millions extra on your bottom line at the end of the year, go right ahead and implement a process for that too.

And if you don’t want to get sued because one of your employees has turned all the sandwiches at Asda radioactive, that’s probably worth a process as well.

Away from those areas, though, keep it simple. And make sure there’s an RoI on every one of your processes.

When I’ve helped organisations dramatically reduce their operating costs, a key part of my approach is just to stop doing things that cost more money to do than they bring in bottom line business benefits.

I realise that might sound simplistic. But equally, it might not surprise you to hear I’ve never found a shortage of activities which cost more to operate than the benefits they bring.

Odds are there are plenty of those same opportunities hiding inside your organisation too.

If you can find some time in your process-driven meeting schedule today, it might be worth a look to see how many you can find.

The best Finance Director I ever worked for – and one of the two best bosses I ever had in my life – taught me an important lesson early in my career.

At the time, like a lot of accountants, I was obsessed by managing the costs in the business I worked for. I’d analyse them incessantly, got competitive prices in for everything we bought in case we could save a few quid by switching suppliers, pushed budget-responsible managers to bring their annual expenditure in under budget, and so on.

My boss, James, acknowledged that there was a role for that, of course, in the life of an accountant.

“But,” he said wisely in a meeting one day, “you have to manage your finances for the business you want to become, not the business you are today.”

I don’t know if there’s such a thing as a Damascene Conversion for accountants, but if there is, that’s what I had in that meeting.

What this means

What James meant was that my approach to managing company finances – an approach most accountants are taught during their professional studies and rewarded for in the workplace – contained an implicit assumption.

The implicit assumption was this: that there is no opportunity for your business to grow in the future, so the only sensible strategy for Finance to follow it to eke out the largest possible return from an essentially fixed revenue and cost base.

Now, if that is what you believe, this strategy has some merit…at least mathematically. (It can be self-defeating in other ways, but that’s a conversation for another day.)

But seen through that lens, why would any sane businessperson ever invest in new machinery, develop new products and services, or explore new marketing and business development strategies to increase sales revenues?

Well, the answer is they wouldn’t. With that lens in play, none of those strategies make any sense as all they do is increase costs in the short term without increasing future revenues.

And if you look at most businesses today, their behaviour is often consistent with that view of the world.

Not because that’s what they say – most businesses spend a lot of time talking about their bright future as a much larger organisation – but because of what they do. The decisions they take today are inconsistent with their claims of building a brighter future.

It’s why people cut corners with their products and services. It’s why you spend an hour on hold to a call centre somewhere. It’s why third-rate AI is being used to replace humans with gang of soulless robots.

The only – and I mean the ONLY – way any of this makes sense is if you don’t believe your products and services are good enough, or you don’t think you’ll ever find more people beyond your current customer base who will want to buy from your company in the future.

In every other scenario, it’s an ultimately self-defeating strategy, albeit with the surface appearance of a mathematically-rigorous piece of decision-making.

What you’re effectively doing, as my old boss James pointed out to me many years ago, is shooting yourself in the foot.

The growth paradox

I’ve yet to meet a business owner who didn’t want to grow their business.

But a remarkable number of businesses have fallen prey to conventional financial strategies which put a greater priority on nickel-and-diming their customers, suppliers, and staff than on taking the decisions which will ultimately grow the business.

Let me give you a tangible example.

The meeting I was in with my former boss James was about whether or not the business I was responsible for should invest in some new machinery.

Some board members were reluctant, pointing out that the business wasn’t all that profitable, and that the increased cost of interest and the additional depreciation charge on the proposed investment would reduce the already slender profits this business made.

Now, to be fair to the board, mathematically they were right.

Or, more accurately, they were mathematically right in the short-term.

But in the long-term, as James pointed out, they were wrong.

That new investment would increase our productive capacity at a relatively modest additional cost, so we would be able to offer more competitive prices to customers, as a result of which we ought to win more business from them.

And we could do all that while still increasing our own profits – any business which can increase its productive capacity by about one-third against a broadly unchanged cost base has plenty of scope for offering more competitive prices to customers while still having the headroom to build its own bottom line.

This was what James meant about making financial decisions which were consistent with the business you wanted to become, not the business you are today.

There was a deliberate trade-off between short-term profitability, but with the downside of longer-term stagnation, or lower profits today, but with the benefit of a much greater opportunity for additional sales and profits in the longer-term.

James sketched out a revised financial plan on the back of the agenda papers for the meeting in a couple of minutes to illustrate the likely impact of supporting the investment decision (he was an amazing accountant) and the board ultimately decided to back the business for growth.

David’s view

David was another early boss of mine – as it happened within the same group of companies as James. Together, James and David were the two best bosses I ever had, and both were foundational in terms of my career.

David wasn’t an accountant, but he was the smartest Operations person I ever worked with.

By the time I met him he was Deputy CEO of a huge PLC. But David told me about a former employer of his, who clearly had worked out a strategy somewhat like James’s for themselves.

This former employer was the biggest business in its sector, and was headquartered in the US, where David had worked for a number of years before joining the group of companies I worked for.

What this US business did when they were making a new investment was that they always invested in 50% more capacity than they needed at the time they made their investment

If they needed to build a new factory on the back of winning a $10 million contract, they would build a factory which could deliver $15 million-worth of capacity.

Then, with the bedrock of the initial $10 million of business to limit their downside risk, they’d set about finding other things to do with the extra $5 million-worth of capacity which they could turn on with little or no extra cost.

I’m sure you don’t need me to tell you this, but if you can put an extra $5 million in revenue through a factory while barely moving its operating costs, that extra $5 million is astonishingly profitable.

Because that extra $5 million was so profitable, it also delivered significant additional cash flow, which means that by the time the now-$15 million factory was full, they would have the cash to invest in a $20 or $25 million factory, and so on.

For years that was pretty much their business strategy. And it took them from a decent enough SME to a huge conglomerate spread across the world.

It only makes sense in the future

This approach, however, only makes sense in the future.

At the point the board makes a decision to build a $15 million-capacity factory when they only have $10 million-worth of work, there’s an argument that the board is insane.

If there happens to be litigious short-term shareholders on the company’s shareholder register, it’s hard to make decisions like this. Because today it’s (arguably) the wrong decision, if all you care about it this month’s numbers or this quarter’s returns.

But, within a relatively short timeframe, spending a little more money today to earn a lot more money in the future makes a lot more financial sense than capping your ambition for the business at whatever you’re billing today and hoping to value-engineer your way to boosting the bottom line.

Eventually you run out of road with that approach. And when you do, stagnation as a me-too commodity supplier is the only way forward.

There are also some little secrets to the process David’s company used to their advantage.

Firstly, building a factory with 50% more capacity doesn’t cost 50% more…provided you build that capacity in at the time you build the original factory. It might cost you 10-20% more for some more bricks and a larger roof, but not 50% more.

All your contractors are already on-site, all the cranes, diggers and whatever else you need is already there. The extra cost of keeping them on site for another day or two too lay a few more bricks is small relative to the cost of building just the $10 million-capacity factory you need today.

For all those reasons, and more, this strategy is likely to save money compared with, for example, adding an extension at a later date to open up the extra capacity you need. That probably will cost 50% more…maybe even 60 or 70% more…to grow capacity by 50% because you’re effectively starting again from scratch.

Secondly, because they built a new factory to service a “launch customer” who had agreed a multi-year supply agreement with, their downside was extremely limited. Even if they never did a single extra piece of business in their new factory, their launch customer would cover their operating costs and deliver a modest profit for that location.

The flip-side of that, however, was if they could funnel some extra business from another source (or sometimes even extra business from the same launch client) through this new factory, their marginal costs were tiny relative to the incremental revenue generated.

In a relatively unglamorous “old economy” B2B business, they could get SaaS-type margins of 70-80% on everything they sold beyond the $10 million launch customer’s contribution. You don’t need to do that very often each month to make a tidy profit at the end of the year.

Thirdly, they were able to explore new market segments relatively cheaply with this extra, almost free, capacity.

They sometimes used the “extra” capacity for product development work, or experimenting with new production methods to increase efficiency and reduce unit costs, or to offer very competitive prices as a way into a part of the market they had not historically served, but where they thought there might be big future opportunities.

It didn’t always work

It’s fair to say these strategies didn’t always work. But they didn’t need to work all that often to be valuable, given the margins they could earn on their “extra” capacity.

If, after 6 or 9 months of trying to break into a new market segment or develop a new product, they weren’t getting the traction they hoped for, this business would just stop doing whatever it was and try something else instead.

Up to that point, it was unlikely they would have lost any money, given the incremental margins they could command, so there was a sense of “nothing ventured, nothing gained”.

There was no internal organisational shame about trying things that didn’t work – they were just regarded as worthwhile experiments which, at the very least, meant the business was not locked into ultimately unprofitable activities for the foreseeable future.

But at its heart, the strategy the business David used to work for came from the same place as the wise advice James gave when I was a boy accountant.

You have to make financial decisions consistent with the business you want to become, not the business you are today.

Today, there are 101 reasons not to spend any money you’re not legally obliged to spend.

There are 101 reasons why you shouldn’t experiment with new ideas today if that means you have to stop doing something that earns you money now in the hope that your new idea will work out better.

There are 101 reasons why you should pursue low risk (in the short term) nickel-and-diming strategies today instead of committing resources to building a bigger, better business in the future, if it means you have to take a hit on this month’s numbers.

If, on the other hand, you want to build a fast-growing, vibrant, successful business in the future, it generally makes more sense to make financial decisions today in the context of the business you want to become, not the business you are now.

I’m still grateful to James and David for, in their own ways, teaching me so many important lessons early in my career. Their lessons have stayed with me my whole working life and been a big part of my career development over the years.

The only sad thing is that a vanishingly small number of organisations I’ve come across in my career pursue strategies like those James and David were pursuing 30+ years ago. And those who don’t, are generally poorer for it.

A long con is where someone spends months or years painstakingly winning a person’s trust in order to rip them off big time.

The payoff has to be really big or the time and effort expended isn’t worth the time it takes. But, with the chance of a big enough payoff, spending months or years painstakingly crafting a long con starts to make economic sense.

Ponzi schemes, pyramid schemes, some of the dodgier investment “systems” are all good examples of a long con.

From Charles Ponzi himself back in the 1920s, to Bernie Madoff, to modern-day crypto heists, all those schemes share the same basic structure: get people to trust you enough to hand over their cash, then choose your moment to disappear with all the cash to somewhere that doesn’t have an extradition treaty.

In the pre-tech era, small-scale frauds were easier because you couldn’t, for example, google someone’s company name to check if they were legitimate or not.

But large scale frauds were extremely difficult because doing everything on manual, paper-based systems meant, at some point, the activity levels were so high that you couldn’t effectively cover your tracks any more.

In the tech era, that’s switched around.

Small scale fraud is harder now – you need to do things like confirm your identity via biometrics on your smartphone before your bank will make a payment.

But large scale fraud is much easier because the problem of scaling your organisation to rip off increasingly larger amounts of money from increasingly larger numbers of people is now comparatively simple. Tech does dishonesty at scale just as easily as it does honesty at scale.

I’m not anti-tech

I’m not completely anti-tech.

I am just old enough to have had clients at the start of my accounting career who still kept their accounting records in leather-bound, hand-written ledgers.

Nowadays Sage, Xero and other accounting systems make keeping a set of accounts vastly simpler and more efficient than it ever was when I was a boy accountant.

And there are considerable efficiency gains. Nowadays an accounts department of three people can do what took perhaps 10 or 12 people at the start of my career.

But there’s one crucial difference between those tech products and AI.

If I had to, I could still produce a set of accounts entirely manually from a pile of invoices, with the aid of a calculator and a pencil.

Equally, I can take whatever is in Sage or Xero and satisfy myself that the accounting entries are accurate and genuine by tracking back to the original third-party evidence.

For accounts preparation, there’s an increased robustness to the process, as well as greater efficiency and, normally, faster turnarounds. At the same time, there’s still in-built checkability, should you ever want to dig deeper. Everything is provable.

Sometimes people point to well-known accounting frauds and use that as evidence this isn’t true. But if you look at accounting frauds, they don’t usually come about because the double-entry bookkeeping was incorrect at a technical level.

They arise because someone in the organisation lied about what the accounting entry was supposed to be for and, often, falsified information to make their explanation look credible.

As in the old days of con-artistry, your odds of getting away with that sort of fraud reduce dramatically the more people need to be involved in the con to make it work.

It’s why an accounts department of any size will have separation of duties as a matter of course, so that two or three people would need to collude to get up to any mischief.

Even highly cynical accountants like me take the view that the more people are involved in a process, the less likely it is that all of them will be dishonest, so the separation of duties becomes an important element of an organisation’s internal controls.

AI isn’t like that

Although much of the language used to promote AI is the same as the language used to promote the early computerised accounting systems – greater efficiency, speeding up mundane tasks, etc – AI isn’t like that at all.

For a start, AI systems are all proprietary models which means we never know how they reach their conclusions, which factors they weigh more highly than others, how they decide which sources to use, and so on. There’s a secret “black box” at the heart of AI technology

Now, AI systems pretty much have to operate like that because, unless there was a proprietary model at its heart, the people who invested $billions in AI start-ups wouldn’t have a route to getting their money back.

The secrecy and opacity built into AI systems is a feature (for their investors) not a bug.

However this secrecy and opacity means that, unlike an accounting system, nobody can replicate the same job manually and guarantee to reach the same answer.

And that’s why AI is ripe territory for the long con. Not necessarily outright fraud, although that’s certainly possible too, but just manipulating the truth in ways that aren’t obvious.

Most of the classic frauds involve some degree of black box decision-making: a secret method, a closely-held recipe for success, a “can’t fail” system.

For fraudsters, that black box element is a feature, not a bug, too.

The last thing they want is someone peering too closely at what’s going on inside their “magic system”. Mainly because the only thing going on in there is the perpetrators syphoning off their investors’ cash.

From an auditor’s point of view, AI is also problematic because you can only do “one way” testing on the results from AI systems.

What I mean by that is that you can take the output of an AI system and, perhaps, track it back to some third-party source document.

However, you can’t take a random third-party source document and track it through the AI system to support the conclusion generated by the black box model, or what we’d call “two way testing” in my auditing days.

That random source document might never have been entered into that model in the first place (although there’s no way of us ever knowing).

And even if it was, the AI system might not have used it, or downgraded its relevance compared to a range of other sources, based on criteria we have no insight into and no opportunity to influence.

The lack of ability to do rigorous two-way testing is the weak spot in every AI model there is.

Which is great news for hucksters, fraudsters, and snake oil salesmen of all kinds.

Start slow, then build up to the payday

All long cons start with the eventual victim getting a win.

Let’s face it, if the victims weren’t suckered in that way, they wouldn’t hang around for long enough for a fraudster to clear out their life savings.

It’s a con as old as time – like the superstar poker player who deliberately loses the first few hands in a game with a wealthy amateur before proposing that they “make it more interesting” by upping the stakes.

So the fact that you can “draw” a camel eating a banana while driving a sports car for free is, for people who have nothing better to do with their time, a benefit of AI. This is an example of the early win that sucks you in.

Ditto the “summarise all the works of Charles Dickens into a single side of A4” and “write your university essay in 5 minutes by using the right prompts” brigade.

They are providing a benefit for free…albeit an ethically questionable one…to get people hooked. Because getting people hooked on a particular AI system is one of the ways those companies will find a way to extract cash from you later.

In fact they’ll have to. When you’re spending a reported $100bn a year to develop AI systems it’s a dead cert someone is going to want a return somewhere down the line – either a direct financial benefit or a benefit in other ways (eg a political result) which becomes a financial benefit later.

It doesn’t always matter

When you’re dealing with things that don’t matter much, I’m pretty sure that some AI technology will prove useful.

Where I’m after something relatively trivial, like “who was Number One in the charts on the 14th of August 1970?” or “what is the formula to give me X result in Excel?” I ask Copilot or Gemini rather than ploughing through 20 search results from Google.

It’s fast and it isn’t the end of the world if Copilot or Gemini tells me the wrong record.

If they give me the wrong formula, it’ll be pretty obvious as I already know the ballpark where I’m expecting the answer to be, it’s just putting it in “Excel speak” I’m not sure about.

So, if I either don’t care all that much, or if I’ve got some other way to sense check the answer, solutions like Copilot and Gemini work well.

But the problem is there is a limited future income source from helping people with trivialities.

To get the big bucks – or to build entire societies in your image – you need to move into the big time and get involved in distinctly untrivial stuff.

The big question

I’m not a complete Luddite. AI is clearly here to stay and I’ll certainly be using it instead of Google search for quick answers to trivial questions for the foreseeable future.

But every time you’re tempted to use an AI solution ask yourself this: if you had to, could you do the same job yourself manually and get exactly the same answer?

If you can, then it might actually be a faster, more efficient way of getting a job done, just like AI’s proponents claim.

If you can’t, then it’s not robust enough to use for important decisions.

While I rarely do a set of accounts by hand these days, I know if Xero didn’t exist tomorrow, I still could.

And, most months, at some point, I get under the skin of a problem I’m trying to solve by drawing out the double-entry, “old school-style”, in a set of T-accounts.

Where you can’t perform the same task manually and get exactly the same answer as your AI tools, you’ve started to inject risk into the process. And the more AI you use, the more risk you inject.

Even something apparently as simple as compiling a report from a set of data can be problematic. From 30-odd years of doing this, I know there are usually dozens of ways to present the same information.

If you let some tech solution choose, you’ve given up editorial control and increased the risk of coming to a sub-optimal conclusion.

My fear for a lot of organisations is that they won’t realise the vulnerabilities they’re building into their systems through the more widespread use of AI until it’s too late.

Even if it’s not strictly a long con, in the sense that AI companies aren’t trying to drain your bank account and disappear to Brazil, there will come a time when your business will be too deeply in to get out of the AI-induced haze.

When I’ve worked with deeply troubled businesses in the past, it’s almost always because they had complete faith in a set of numbers which turned out to be garbage.

Not garbage mathematically – the numbers always added up. But garbage in the sense that they measured the wrong things in the wrong way, and so reached the wrong conclusions because no-one in the business really understood what was going on.

That’s bad enough when humans are involved.

But AI shows all the signs of operating like all the worst people I’ve ever worked with.

They all had blind faith in numbers they never really understood too.

They all drew the wrong conclusions from the data because they lacked the full contextual understanding of the problem they were trying to solve.

And, in some cases, they subtly manipulated the numbers to give them the answers they wanted despite the underlying reality pointing at a different answer altogether.

AI is the technology equivalent of letting third-rate people with an invincibility cloak and an underserved degree of over-confidence in their abilities make all the decisions.

And I don’t need a fancy technology solution to tell me that’s unlikely to be a wise move in the long run.

(PS: in case it’s been bugging you, Gemini tells me that Elvis Presley was top of the UK charts on 14th August 1970 with “The Wonder of You”. )

“Garbage in, garbage out”, or GIGO, was something we were all warned about as young accountants. Sure complex data systems had their uses, but if the quality of the stuff being loaded into them was poor, there was no point using their outputs to drive business decision-making.

One of the reasons people sometimes find me challenging is because, when they present numbers to me, my starting point is never “what do these numbers tell us?”. It’s always “how do we know the data underlying these numbers is correct?”

This is partly from experience. I can’t tell you the number of times I’ve found that the numbers being presented to me are…at the most charitable interpretation possible…only a partial version of the truth. Often what they don’t say is more important than what they do say.

So I spend a lot of my time in meetings wondering what I’m not being told.

It’s not that the numbers are wrong. They are inevitably mathematically accurate, based on the data someone fed into a spreadsheet.

But if the underlying data is suspect, or presents an incomplete view of the problem, it’s easy to end up taking entirely the wrong decision just because the numbers you are presented with would suggest that’s the best course of action.

A quick example

A relatively common example of what I’m talking about comes from salespeople.

Now, sales is the hardest job in any business, and I’ve worked with some great salespeople over the years.

But some salespeople, even quite senior ones, forget that it’s not the company’s objective to make a sale. The company’s objective is to make a profit.

And while you can’t make a profit without a sale, it’s equally quite possible to make lots of sales without ever making a profit.

Years ago, I worked with a salesperson who was seen as very successful, but that’s only because, every time his customers asked for a price reduction, he always gave it to them. The customer showed their appreciation for this by giving him more of their business.

Over time, as his main (very large) client switched more and more of their purchasing to him, his sales numbers skyrocketed.

The only issue here was that all the work we did for this client lost us money. And every time this sales rep increased sales to his mega-client, not only did we lose more money, we also took up space on our production line which could have been used to produce profitable work instead.

Before I got there, all the board presentations by the sales director had been encouraging the business to ride this incredible growth from a large, prestigious client because the fact we were growing so fast with them meant “we had to be doing something right”.

However, nobody had worked out the real costs of servicing this client, they’d just done a superficial calculation or two and pressed ahead. So, sadly, the only thing we were “doing right” with this client was learning how to lose money by servicing them.

Why the company was losing money was a mystery to everyone involved because, until I got there and asked a few awkward questions, they only had some of the information they needed to make good business decisions.

They thought they were making good decisions, but this was really an example of “garbage in, garbage out”. Bad…or, perhaps more fairly, incomplete…information at the front end meant that the decisions they made, based on that data, were inevitably flawed.

Another quick example

Another time I went to work for a business with a number of financial difficulties (yup, I love a challenge…!)

This business had mixed up two important – but entirely separate – financial concepts…the profitability of an individual job or project, and the overall profitability of the business.

Now, these sound like the same thing, but sometimes they’re not.

This particular organisation had a high fixed cost base, but beyond a threshold we could produce two or three times the sales volume with only a tiny increase in cost. Past the threshold, each incremental sale produced about 99% gross margin.

However, supposedly in the name of “rigorous financial control”, this organisation had decided that if a project wasn’t going to achieve a margin of at least £X they weren’t going to do the project at all.

Not completely crazy, you might think. And sometimes you’d be right.

But in this case, most of the projects which came along were, within a small margin above and below, somewhere around £X.

And because £X was an arbitrary limit, if your project was achieving 95% of X, we didn’t take it on, whereas if it achieved 101% of X we did.

This all sounds commendably data-orientated, doesn’t it…?

Except the person making these decisions – and engineer by trade, as it happened, who valued precision above all – had forgotten that, against a fixed cost base, the best way to turn around the organisation’s fortunes in the short term was to take every project which delivered a positive return over our “hard” costs.

It was fine to have £X as an objective. And obviously in the longer term there are other strategies you would want to deploy to enable the business to grow revenues and shrink costs. But in the short-term, turning down jobs which covered some of our fixed costs and threw in a contribution to our profits, however small, was just about the worst approach to take.

Let’s imagine the required £X was £10,000, and we could have sold a project which brought in £9,500. Under his rules, the threshold hadn’t been met, so it was an automatic “no”.

The reality is that £9,500 was still covering a big chunk of our fixed costs, making us a profit (albeit a slightly smaller one than we might have wanted), and providing some positive cash flow.

By turning down a £9,500 job, perhaps £8,000 of fixed costs were now not being covered at all, and also we missed out on £1,500 in extra profit. Less than we would ideally have liked, but still a none-too-shabby contribution on the bottom line.

What’s more, by not taking those £9,500 jobs, our production facilities stood idle for longer and longer each year. All in the name of “rigorous financial control”.

It won’t surprise you to learn that this organisation lost £millions in revenues due to this policy, couldn’t cover its fixed costs, and ended up in severe financial difficulties.

Mercifully, I was long gone by then. It was clear that this organisation would never be able to shift their thinking while that individual remained in charge.

Even as their ship was sinking fast, this organisation remained obsessed about whether individual projects achieved £X, and could never bring themselves to see that there might be another way of looking at their decisions. One which would result in the exact opposite decision.

Of course, it’s not always right to take on projects which don’t achieve whatever targets you’ve set for the business. It’s a complex issue which depends, amongst other things, on your sales and marketing approach, how you manage capacity in your business, and the split between fixed and variable costs on a typical piece of work.

All you need to take away from this story is that there was a way of looking at the numbers which nobody even considered, which would have given a completely different (and correct, in this case, as it happens) conclusion. The organisation was completely blind-sided because it didn’t even consider there might be another way to make decisions apart from “does it achieve £X?”

They didn’t challenge the underlying assumptions they were making rigorously enough. In this case, the underlying assumptions were the “garbage in” bit of the equation, so the “garbage out” in terms of poor decision-making were pretty much inevitable.

So what does this have to do with Miles Davis?

If you don’t know the name, Miles Davis was a famous jazz musician. Many commentators have described him as one of the coolest people who ever lived.

His 1959 album “Kind of Blue” remains an icon of 20th century music and one of the best-selling jazz albums of all time.

But Miles Davis’s outstanding composition and playing skills aren’t the issue here.

Rather, it’s something he said when asked about how to be a great musician.

“Don’t play what’s there; play what’s not there.”

Think of it like this. If you ask your computer to play a set of notes in sequence by reading a musical score, you’ll get each specific note played accurately enough. But you’re probably going to get bored of listening to that pretty quickly.

For two reasons.

Firstly, music, like business, is not just about the mechanical reproduction of a pre-determined outcome – whether that’s to play a B flat, or to build a new 3-series BMW.

While it’s especially true for jazz musicians like Miles Davis, particularly when playing live, all musicians improvise around the notes on the score, if for no other reason than to stop themselves getting bored playing the same tunes over and over again.

Even in classical music, a genre where compliance with the composer’s specific instructions is prized much more highly than in most other musical disciplines, musicians play ever so slightly differently to one another.

When a reviewer writes “so-and-so gave the best performance of Brahms I’ve ever seen”, it isn’t because this is the first time they’ve heard the notes played in the right order.

It’s because the performer they were listening to played “something that wasn’t there”.

The second way musicians play “something that isn’t there” is in how they put emotion into their playing.

Using the same musical note, you could be trying to convey happiness or sadness, get people up to dance or bring them solace in the midst of grief, send people off to war or celebrate the outbreak of peace.

The only difference between those very different scenarios comes down to how well a skilled musician conveys the emotions they want to evoke through their playing.

How you play a musical note is “what’s not there”. But you’ll never see how to play a note on a musical score. All you’ll see is what to play.

So accounting is cool after all… 😉

Although you might have trouble believing it, this is how accounting can be cool too.

Sure, you can just “follow the notes on the page” and you’ll produce something competent…if a little dull. That’s “playing what’s there”.

Or you can “play what’s not there”…work out what the other side of the argument might be and see if there’s any data for that…think about what’s not being talked about, like the story of the unprofitable, but fast-growing, customer above…or figure out what ingrained thinking is holding the organisation back instead of propelling it forward, like the guy with the fixed project target of £X.

That, for me, is the exciting bit about accounting. That’s how I’ve added value to the organisations I’ve worked for over the years, in both financial and non-financial roles.

And while I might not be quite as cool as Miles Davis, as accounting goes, this is about as cool as it gets.