AlastairThomson.com

It sounds like it should be the other way round – that complicated things should be harder than simple ones – but that’s not true. Often complicated things are much easier to manage.

Think about this in terms of food.

Go to a supermarket and pick up a pre-prepared meal of almost any kind. What do you find when you look at the ingredients panel?

Probably about 47 different ingredients, most of which you’ve never heard of and, if anything, just make you wish you’d paid more attention in your O Level Chemistry class.

Modified starch? No idea what that is….and modified how, more to the point. Anyway, what’s the problem with using unmodified starch…?

Sodium ascorbate? Dunno. Isn’t sodium something to do with salt? After that, I’m struggling…

Hydrogenated vegetable oil? That sounds like one of those made-up ingredients they talk about in TV commercials for face cream. Why are they putting that in food…I’m going to eat that…!

I could do on, but you get the idea.

Undoubtedly making a supermarket ready meal is complicated, with 47 different ingredients you’ve mostly never heard of to juggle in the right proportions. But – and I can state this with some authority having been a supplier to food manufacturers in a previous role – it’s not really that difficult.

Large scale food manufacturing is done in huge factories, using a lot of automated machinery which has been designed so that people with relatively little training – and ideally who don’t cost too much to employ – can operate them and produce an acceptable result within the tolerances set out by the Quality Control Department.

Your supermarket ready meal is sounding less and less appealing the longer I go on about this, isn’t it…? But let me observe that just having 47 ingredients does not mean a high-quality dinner is being produced here.

Probably the opposite.

(Over)compensating

In case you don’t know, the main reason there are 47 ingredients in your supermarket ready meal is that they are mostly there to make desperately unappetising things vaguely palatable, and to prolong shelf life which helps keep supermarket profits high.

And that’s on top of the usually epic amounts of salt and sugar that get crammed into the typical ready meal.

There aren’t many things human tastebuds respond to more than epic amounts of salt and sugar (especially sugar, in my case). And while there is a biological reason for that, going back to when we lived in caves and hunted woolly mammoths with spears for food, the major food producers, left to their own devices, want to put in enough salt and sugar to keep us hooked on the “high” their products are designed to give us.

Over time, it’s been discovered that many of the ingredients which food producers used in the past to make their food palatable or prolong its shelf life were positively harmful to humans. They were a short-cut which benefited food producers, but at the expense of their customers’ health.

It’s not the only reason there’s an obesity crisis in many Western economies, but food companies using 47 ingredients which are designed to (or have the unfortunate side effect of) making us want to guzzle up more and more of whatever the supermarkets are selling is at least part of the reason.

But that’s a different point, and not one I’m making today, particularly.

However, please note that one of the things all those ingredients do is make something that sounds complicated – making the shelf life of a supermarket ready meal as long as possible – comparatively easy. Just add a few drops of magic ingredient X, and you get another couple of days’ shelf life, no problem at all…!

What real chefs do

Real chefs take an almost diametrically-opposite approach.

Sure, there are some people who use tweezers to pick a single leaf from a special herb kept in climatically-controlled conditions in their basement and pop it on top of a piece of grilled salmon.

But even then, they’re not trying to fool us into thinking that the world’s cheapest piece of chicken is somehow nutritious, like supermarket ready meals try to do. They just believe that the flavour profile of some obscure herb is exactly what that piece of salmon needs to make it perfect. Who knows, they might even be right.

Mostly, though, the chefs other chefs really respect are ones who can make a delicious meal out of just a couple of ingredients.

That takes real skill. Those chefs can’t overcompensate for using poor ingredients or inadequate technique by adding hydrogenated vegetable oils. Their cooking has to stand on its own, the function of ingredients and technique only.

And that’s hard. Much harder than cooking with 47 ingredients. Because there’s nowhere to hide – your ingredients, your technique, your sense of taste has to be absolutely on-point every time.

All the while dealing with the inherent variability of your ingredients – every egg is not exactly the same size as every other egg, for example. For most of us, our level of skill in the kitchen means that will probably be the least of our problems, for a top chef working with just a few ingredients, that can make all the difference.

Take the pasta dish cacio e pepe, for example.

A literal translation (sorry to any Italians reading this) is “cheese and pepper”. Cacio e pepe is a dish from the Lazio region of Italy (although not every Italian sees it quite like that…long story…) and it is literally made from pasta – often spaghetti – cheese, and pepper.

How hard can that be, you might think – there are only three ingredients.

Alternatively, you might be wondering how cheese, pepper, and spaghetti together even produces a creamy sauce and tastes nice. This dish is a million miles away from the tomato-based pasta sauces most people in the UK and further afield would probably think of as “Italian cooking”.

Well, it turns out cacio e pepe is delicious.

But making it well – which I can’t, by the way – is really, really hard because you can’t just lob in another tin of tomatoes if it all goes horribly wrong and still end up with something semi-edible.

What I can tell you, though, is that there are a few key parts of this process:

1. First probably a fairly well-known one – al dente pasta. Very few of us can reliably produce that (I manage it about 1 in 10 tries, personally), but getting that right is key.
2. Then the pepper isn’t just any old pepper you’ve got in the back of a cupboard – it’s fresh peppercorns, warmed gently in a dry pan, and then crushed with a pestle and mortar until the consistency is just right.
3. Next, it’s the cheese. It’s got to be pecorino romano. Got to be. Never use parmesan which, despite being broadly similar in concept, is an entirely different product – using parmesan in cacio e pepe for an Italian seems to be like putting Coca-Cola into 25 year-old malt whiskey is for a Scottish person. So don’t do it unless you’re deliberately trying to cause offence.

But that’s not the hard bit. Although none of this so far is easy.

The really hard bit is ending up with a smooth, creamy sauce which evenly coats the spaghetti instead of a plateful of dry pasta sprinkled with unappetising lumps of dry, un-melted cheese.

Doing that requires an enormous amount of skill on the part of the chef to manage the temperature, make sure the sauce has the right consistency, and the flavours come through as they should.

If you’re not a professional chef, making cacio e pepe is one of the hardest things there is to cook. Yet it only needs three ingredients.

Making a supermarket ready meal, by contrast, is easy. Just load all the ingredients up in a massive hopper and the machinery will take care of it all, relying on the salt, sugar, flavour enhancers, and goodness knows what else to produce something that’s relatively palatable despite being fundamentally beige and unexciting.

KPIs should be like cacio e pepe

Most KPI reports are like a supermarket ready meal’s ingredient panel. Throw enough things in there to make consumers believe they are eating a relatively palatable product and you’re probably OK for another month.

It’s much harder when you’ve only got a handful of KPIs, because there is nowhere to hide.

With 47 different KPIs, you can almost certainly find something that’s going in the right direction to use as a counter when you’re taken to task for something that’s going in the wrong direction.

With just three KPIs, you’re getting the unvarnished truth.

Every time a see a KPI report with 47 different things on it, I know I’m dealing with a business which is at least unfocused, because it’s simultaneously pursuing 47 different things of broadly equal importance. That means each one of them is getting worked on for only about an hour a week – hardly a recipe for significant, and rapid improvement in any metric.

In addition, the organisation is probably, on some level, delusional. Because their organisational culture is almost certainly allowing underperformers on 40 of their metrics to boast about the 7 they’re doing OK on, and for that to be perfectly fine within their management practices.

It’s one reason AI leaves me cold. Sure, it might be very clever, and it might be able to produce 47 different reports in the blink of an eye. But, like the ready meal producer using weird and wonderful ingredients to cover up for the fact that their product is fundamentally of pretty low quality, the real reason people want 47 different reports is to divert attention away from the things that really matter when they go badly wrong.

I’ve got a lot more respect for a business that can be run on just three KPIs than I have for one which has an elaborate KPI reporting system spewing out reams of data nobody will ever read in its entirety every week or every month. (Or even worse, one that builds a real-time data dashboard that nobody ever looks at.)

With cooking, adding another 44 ingredients to cacio e pepe won’t make it better – usually quite the opposite.

So it is with KPI reports. More ingredients usually means what’s inside is much worse than the person doing the reporting would like you to think it is.

It’s a smokescreen. Those extra ingredients are probably about as bad for your business health as the extra ingredients in your ready meal are for your physical health.

You don’t need more KPIs. You need the equivalent of a skilled chef to make your business run with just three ingredients, cacio e pepe style.

Don’t mistake “more” for “better”. It rarely is.

The great thing about business is that business is really about people, much as tech folk would like to convince you otherwise.

There’s never been a truer adage than “people buy from people”, for example.

Or the well-worn advice that people are more likely to do business with those they know, like, and trust.

These are entirely human judgements and, like it or not, almost entirely subjective.

It’s fair to say our tech overlords occasionally do a half-decent job of faking all of those things – certainly good enough to fool some of the people all of the time – but, like all fakery, eventually the chickens come home to roost.

There really is no substitute for humanity in business.

You can pretend otherwise, and believe everything is just about 1s and 0s. But you’re only fooling yourself.

And even to the extent it works, it’s vastly more expensive and resource-intensive to fake things than to be human. Which means the fall from grace, when it comes, it much harder and deeper than it would have been otherwise.

If you doubt me, consider the amount of effort Enron put into trying to maintain the illusion of being successful – up to and including committing outright fraud, for which several people went to jail – when they could have done very nicely out of just being honest about their business.

And if you’re not old enough to remember Enron, stay tuned. I’m expecting a similar dynamic to present itself in the course of this autumn’s slate of IPOs for AI companies.

But Enron’s failure was fundamentally a human failure, not a business one. The human failure came first, the business failure followed.

By the same token, extraordinary humans can make great business decisions. Sure, it requires some creativity, some ambition, some hard work. And humans are flawed, of course, but they’re several orders of magnitude less flawed than tech solutions.

Not every decision a human makes is a good one. But when humans make good decisions, a whole new world opens up.

So today, I’d like to highlight three great decisions made by humans as an example of this phenomenon.

As it happens, they are all women. And, as it happens, they are all examples from the music industry. But every one of these are phenomenal decisions – brave, creative, insightful.

Decisions you’d never make based on a spreadsheet.

These decisions came from the heart, and they are decisions only a human could make.

All three are the opposite of logical. There were millions of really good reasons to do the exact opposite, yet these three women zigged when the world was telling them to zag.

I admire them all immensely.

Taylor’s version

The most recent example is from Taylor Swift – someone so famous that I don’t need to explain who she is, even to people who aren’t fans of her music.

There is a little twist to this tale, even if you think you already know it, so stay tuned.

As a teenager, Taylor Swift signed a recording contract in which the rights to her master recordings belonged to her record company. In fairness, that’s not an entirely unprecedented arrangement in the music industry.

In the blink of an eye, though, Taylor Swift went from being a teenager writing songs on the bedroom floor in her family home to being one of the biggest recording artists on the planet.

After a few years she was so successful, Taylor Swift wanted to buy the master recordings so she owned her entire body of work. However, as a result of some murky, and much-disputed, events (which are not essential to the point of this story) that didn’t happen.

Her record company was bought out by a famous music industry manager, and soon after re-sold to a private equity outfit who proceeded, as private equity companies do, to sweat the assets of the business they’d bought.

Determined not to be bested by her record company, Taylor Swift then re-recorded all her early albums in what became known as “Taylor’s Versions”, so that she now owned the masters of her “Taylor’s Version” recordings of her back catalogue.

With a legion of adoring fans, sales of her original recordings plummeted and sales of the Taylor’s Version recordings soared, even among people who had previously bought the originals.

I don’t know what happened in detail as that isn’t publicly available information, but I think it’s safe to say that value of the income-producing assets the private equity firm bought tanked and their investment was no longer the gold-plated opportunity it had initially appeared to be.

There are two little twists to this story.

The better-known one is that, in 2025, some five or six years after the original buyout of her record company went through, Taylor Swift did come to an agreement to buy her original master recordings back.

By taking a stand, thinking creatively, and being determined not to be beaten, Taylor Swift got what she wanted in the end. All credit to her for demonstrating some top-notch human decision-making there.

The other twist, which I only discovered recently, is that American Idol winner Kelly Clarkson had a hand in this too. Back in 2019 she tweeted to Taylor Swift that she ought to re-record all her songs so she would own the masters, and offered her support.

That’s not to say that Taylor Swift didn’t have the same idea independently – I have no idea one way or the other. And to give Kelly Clarkson full credit as a human being, she has since acknowledged in interviews that Taylor Swift would probably have come up with the same idea herself anyway, so she isn’t looking for any credit.

Wherever the idea came from, though, Taylor Swift executed it to perfection and saw through a course of action that very few people would have taken.

As business moves go, what Taylor Swift did is one of the best ever.

All because she thought like a human, not a machine. And because she said “no” when it would have been so much easier just to say “yes”, and take whatever crumbs were on offer from her record company.

Dolly Parton

Dolly Parton is legendary in the world of country music – somewhat less so outside it. She is, however, one of the best-selling musical artists of all time. Along the way she has picked up 11 Grammy awards and written some tremendous songs – I can’t listen to “Coat of Many Colours” without finding a tear in my eye, just for starters.

Dolly Parton started performing as a child, had some success writing songs for other people along the way, and some modest success in her own right as a singer.

In her early 20s, Dolly was invited to join Porter Wagoner’s immensely popular weekly country music TV show, which gained her considerable public recognition. But she was very much playing second fiddle to an established star and her solo career took a back seat for a while even though a series of duets with Wagoner did well in the country music charts.

She did keep writing songs, though. Standards like “Jolene” were written about this time.

Eventually Dolly Parton decided life as the second act in a duet wasn’t really what she wanted, and spilt with Porter Wagoner, even though I’m sure she could have had a very decent long-term career had she stayed.

That was human decision number one. The logic might well have said “stay”. The heart said “go”.

Human decision number two came along shortly after.

In 1975, Dolly Parton wrote and recorded a song called “I Will Always Love You” about her split with Porter Wagoner, which went to Number One in the country music charts.

Particularly if you’re a Brit, you almost certainly don’t know that version of the song.

But, wherever you are in the world, I can guarantee you know Whitney Houston’s cover for her 1992 film “The Bodyguard”. That version spent 14 weeks at the top of the Billboard Hot 100 in and sold over 25 million copies worldwide.

What is not so well known is that back in the mid-1970s, soon after Dolly Parton wrote “I Will Always Love You”, she was approached by Elvis Presley’s manager to see if she would let Elvis record her song.

Now, at the time, there were probably few better outcomes for a songwriter than having Elvis record one of your songs. And, to be fair, I think he’d have made a tremendous version of Dolly’s song.

The only problem was, the standard deal when Elvis recorded a song was that he’d get half of the publishing royalties.

Dolly Parton reached the very human decision that she wasn’t prepared to sign away half the rights to the song she’d written and said no.

The easy, logical answer would have been to agree to Elvis’s terms. Dolly Parton took a human decision.

I know this is speculation, and involves 20/20 hindsight, but if Elvis had recorded the song he’d probably have had a decent hit with it. It is, however you look at it, a great song.

But I doubt the producers of “The Bodyguard” would have chosen such a well-known song as the centrepiece of their Whitney Houston film – they needed a great song, but not one that everybody knew already.

So Dolly Parton traded a relatively certain million-seller from Elvis for an iconic performance by Whitney Houston and sales of over 25 million copies. Even though at the time she turned Elvis down she had no way of knowing that, 20 years later, someone would want her song for one of the biggest motion pictures of the 1990s.

Taking a human decision generated so much more for Dolly Parton over the year – critically, musically, and financially – than the logical decision she could have taken in the mid-1970s.

As business moves go, Dolly Parton’s human decision was one of the best.

Andrea True

At some level, you are probably familiar with Taylor Swift and Dolly Parton – however superficially.

With Andrea True, that’s much less likely. Unless you’re of a certain age, in which case you’ll certainly remember her biggest hit record. But more on that in a moment.

Andrea True had a very musical mother, and had studied music herself, but wanted to become an actor, and moved to New York from her home in Nashville. Along the way she kept her singing career going in pubs and clubs as a way of helping make ends meet.

She didn’t get the results she hoped for in her acting career, but in the mid-1970s she was hired by a real estate business in Jamaica to feature in a series of TV commercials.

After shooting the commercials the Jamaican government had one of their periodic fallings-out with the US government and prevented people taking money out the country. Andrea True would have had to either forfeit her fee or spend the money in Jamaica before going back to New York.

Instead she took a smart, but very human, business decision.

Andrea True contacted a friend in the music industry who had an instrumental track he’d been working on and used the fee from her commercial work it up into a something more like a finished product with the help of some Jamaican musicians. That music could then, perfectly legally, leave the country.

Back in New York, she and her friend, Gregg Diamond, worked on the song and Andrea added her lyrics to the instrumental track. Along the way the song was turned into a disco anthem and the result was the 1976 UK and US Top Five hit “More, More, More”.

This piece of shimmering disco pop was a staple of the dance floors for a large part of the mid-to-late 1970s and quite rightly features in Billboard Magazine’s 2025 list of “100 Best Dance Songs of All Time” at Number 36.

If you’re not familiar with the song, don’t worry, I’ve got you covered – you’ll find a link to it below.

But before we get there, let’s just reflect on Andrea True’s very human decision that there was no way she was going to let the Jamaican government take the money she’d worked for. It took some “outside the box” thinking and a degree of grit and determination that all good ideas take to come to fruition.

Logically, it would have been easier for Andrea True to book herself into one of Jamaica’s nicest hotels and spend a couple of weeks in luxury, thanks to the fee she’d earned from her commercials, after she found out she couldn’t take her money out the country.

Instead, she used her human powers of decision making to turn the potential loss of all her money into a Gold record in the US and an international smash hit.

Well done to Andrea True – that was a great business move, and vastly better than just spending her money on a nice hotel break, thanks to her human, rather than logical, decision.

What does this mean?

There’s no real moral to these stories apart from the observation that a logical decision is often the easy path. Doing what people would expect, being able to express your decision in seemingly-objective terms, conforming to the norms are all the result of logical decisions.

Most of the time, being logical won’t make you broke.

But being logical also means you’ve likely given up a huge percentage of any potential upside.

It would have been logical for Taylor Swift to shrug her shoulders when her record company went against her, and just keep going with her career.

It would have been logical for Dolly Parton to let Elvis record her song and give up half her royalties.

It would have been logical for Andrea True to have a nice holiday in Jamaica instead of recording a global smash hit.

All of those decisions would have been entirely logical. Completely defensible.

And, at the same time, they would all have given up an upside beyond their wildest dreams.

While logic has a role, the sort of human decisions that only humans can make – often seemingly “illogical” at the time those decisions are made – are where all the upside in life is, inside or outside the music industry.

However these three women from the music industry have, between them, made some of the best business decisions of all time.

We could all learn a lot from them.

---

Music mentioned

In case you got this far and aren’t familiar with any of the songs referenced above, here they are:

Taylor Swift “Shake It Off (Taylor’s Version)”

Dolly Parton “Coat of Many Colours”

Dolly Parton “Jolene”

Dolly Parton “I Will Always Love You”

Andrea True Connection “More, More, More”

If you run an organisation of any sort – business, public sector, not-profit, or anything else – you want your organisation to be efficient, right?

You might think this is an odd thing for a CFO to say, but no. You don’t.

What you want is the best return from your organisation’s efforts, whether you think of that in terms of revenue growth, profit margins, or return on assets employed in a business setting, or people helped or missions delivered in the public sector and non-profits.

It’s tempting to think that “more efficiency” will necessarily result in improved outcomes, but that’s not true.

Someone can be very efficient at trying to sell snow to Eskimos, but Eskimos already have plenty of the stuff and don’t need any more of it. Becoming even more efficient in their sales processes isn’t going to help.

Someone else might spend a lot of time, effort, and money trying to sell new cars to people who can’t afford the payments. No matter how efficient they become, they won’t sell many cars.

Yet another person might devote their business to selling premium wagyu beef steaks to a list of members of the Vegetarian Society. No amount of efficiency is going to shift their warehouse full of steaks.

So, if efficiency on its own isn’t enough, what should we aim for?

Outputs beat inputs

I’m glad you asked – you should always aim for effectiveness, not efficiency.

While efficiency, in and of itself, is not a bad thing, it’s a second order question once you’ve developed an effective way of reaching a goal or solving a problem. It’s not where you start.

Both are important for well-run organisations, but they serve entirely different purposes. Getting the two confused, as many people do, is the cause of a large proportion of today’s business and societal ills.

One of the key differences is that effectiveness is an output measure, while efficiency is an input measure.

To take one of the examples above, what I mean by that is the effectiveness of a car sales organisation can only be determined by how many cars they sell to paying customers.

If you work out a way to do that, then by all means see how you can make your sales operation more efficient. Maybe there’s a role for a new CRM system once sales reach a certain level.

But, at the outset, the bottom-line benefits of a CRM will be marginal. The problem here is that the business model is broken. No amount of efficiency will help someone trying to sell products to people who can’t afford to buy them.

Don’t be tricked

Recently I’ve seen people describing, for example, the number of sales calls a salesperson makes in a day, as “an output measure”…which, on the face of it, sounds like a measure of effectiveness by the definition above.

It might be a measure of a salesperson’s personal productivity, but that’s an efficiency measure, not an effectiveness measure.

The key point here for business leaders is that a large number of your people can, with some justification perhaps, claim to be highly efficient. Your job as a business leader is to look beyond that and ask yourself whether, irrespective of the number of calls a salesperson makes in a day, trying to sell new cars to people who can’t afford them is a good idea.

While you’re thinking about that, keep in mind one of my favourite business quotes from Peter Drucker “There is nothing quite so useless as doing with great efficiency something that should not be done at all.”

When you put the focus on your outcomes (effectiveness measure) before you think about inputs (efficiency/productivity measure) you’ll make much better business decisions.

But, to be fair to most people in your organisation, you can only take a view on outcomes at, or close to, board level because that’s the only level where you can see the full end-to-end process which leads to a successful outcome – eg selling a car.

If you employ a room full of telephone salespeople whose job it is to make calls all day long, don’t be surprised if they make a lot of calls, and spend a lot of time trying to convince you that they’re all doing a great job, even if they aren’t making any sales.

“I’ve been working so hard, boss, but people just don’t want to buy a new car at the moment. But hey, I made 400 phone calls this week, so I’m working like crazy to bring those sales in!”

Don’t blame the telesales team for that. The responsibility lies much higher up the organisation. When people there are mixing up the concepts of effectiveness and efficiency, the organisation is sunk, no matter how many calls a day each salesperson makes.

Start with the customer

To run an effective organisation, you have to start with your customer. Their needs, wants, and desires.

Try to do anything which is not addressing at least one of those agendas, and you’re pushing water uphill. That always means your organisation is working much harder, and spending much more money, to get to the same objective than a competitor organisation who thinks about the problem more creatively.

Effectively, organisations who try to power through regardless are using money to compensate for short-changing themselves in the thinking process. That’s always a losing proposition for your bottom line.

It’s like judo – in that sport, you use your opponent’s momentum against them. That’s why a quiet, skinny person with little in the way of visible musculature can defeat someone built like a heavyweight boxer.

In business, creativity performs a similar function.

You can spend a little bit more to generate a brilliant idea or you can spend 10x more in advertising media costs to get some level of recognition across your target audience for an insipid, me-too idea.

All things being equal, I know which one I’d be doing.

But away from the obviously creative areas of your business, the same applies to much more mundane business processes.

You can either solve your customer’s problems with creativity or you can spend vastly more money to get them to do things the way someone in a conference room somewhere decided ought to be done, having never spoken to a customer or used the product themselves.

The classic example

A good example of this in the real world is how difficult many organisations make it for customers to give them money.

There are rare exceptions where there is a legal or regulatory requirement to conduct some checks on your customers before dealing with them, such as the legal and financial services industries’ KYC (Know Your Customer) rules. But most businesses aren’t operating in that arena.

Not so long ago I was dealing with a business (or trying to deal with them, more accurately) which never responded to enquiries via the contact form on their website or to telephone calls made to the number on their website.

Take a quick guess how many sales they made to me?

Admittedly that’s an extreme example, but there are plenty of other examples closer to that end of the spectrum than the “making it easy for your customers” end of it.

A while back, in order to buy a relatively humdrum item for my house, I had to complete a form requiring all manner of demographic information before I could set up an account and…only then…be allowed to buy from this business.

On this occasion I persisted – partly out of morbid curiosity, partly because I was under very strict instructions to buy exactly that item from exactly that store.

Of course, I knew exactly what they were doing. I was essentially pre-seeding their automated marketing campaigns for them, so they could spam me forever more with personalised emails.

I don’t know what their drop-out rate was, but any process requiring a customer to input all manner of personal information is going to have a higher drop-out rate than a process which doesn’t. So straight away they are giving up sales they might otherwise have had, which, off-hand, doesn’t seem like the smartest strategy to boost a business’s bottom line.

The process brought to mind a question old-school direct marketer Dan Kennedy used to ask – are you finding a customer to make a sale or are you making a sale to find a customer?

To horribly paraphrase Dan’s very worthwhile teaching, his recommendation was generally to make it easy for someone to buy at least some sort of entry-level product which you could use to establish a base to grow from, rather than to spend a huge amount of time, energy, and money to find a perfect customer and then try to sell them something.

In other words, make it easy for customers to give you their money.

Once you had a customer, you could build the relationship out further, but at least you had established a relationship, built up a degree of trust, and demonstrated some added value along the way before you tried to sell them more products. All the subsequent sales were easier because of the positive experience you had established in the process of making the first sale.

While that approach might not be appropriate in every sales scenario – if you sell nuclear power plants, you probably can’t sell an entry-level product first and then try to persuade your customers to buy a nuclear power plant on the back of that – it works well in a remarkable number of settings.

It’s not just about sales

This article isn’t just about making sales, though. It’s about enhancing your bottom line. The same principle applies in every other area of your operations too.

You can, for example, buy cheaper raw materials or sub-assemblies for your products than whatever you buy now. However it’s likely that strategy will lead to more quality complaints, more returned products, and higher costs in your customer service department while they apologise for the substandard products which were delivered to your customers.

Generally, the best return for your bottom line comes from buying good quality raw materials and sub-assemblies which don’t result in quality complaints, refunds, and inbound customer service calls.

And don’t be fooled into putting all your customer service online and think you’re saving money. That does nothing to persuade customers your products aren’t shoddy. All you’re doing is making it less and less likely they will ever buy from you again – if there’s one thing worse than buying a shoddy product, it’s buying a shoddy product from a company where nobody cares enough to put their customers’ problems right.

Equally, you can track 1,000 data points and still run a business that its customers have very little time for, or you can track the only three things your customers get obsessive about and make sure you deliver those 100% of the time.

Whatever the cost of all that data management and tracking might be, you can save most or all of it if you just focus on what your customers care about.

Don’t mistake your notion of what makes you more efficient trump the things that make you effective in your customers’ eyes. While efficiency delivered in the context of doing things which make you effective is no bad thing, just wildly pursuing efficiency on its own terms more often takes away from your bottom line than adds to it.

I know that sounds like it shouldn’t be true, but it is – no matter how much third-rate accountants might try to convince you otherwise.

Being effective is about building value in your customers’ eves.

Being efficient is about impressing whoever sits above you in the organisational hierarchy, irrespective of – and often in spite of – the impact it has on your customers.

Whenever you have a choice about delivering on an internal agenda – efficiency, for example, or any one of a raft of internal policies and procedures – or becoming more effective in your customers’ eyes, always go with the latter.

Not a single customer gives a fig for your internal policies and procedures. All they care about is how well you serve them.

And, in the end, the valuation of your business and the state of your bottom line depends entirely on how positively your customers feel about the service they receive.

If in doubt, prioritise that. It’s almost always the best way to boost your bottom line.

You may not have heard of the bus stop problem before. Possibly because I’ve just invented the concept.

However, bear with me as you might spot some things you or your company could do differently, boosting your bottom line in the process.

To explain what the bus stop problem is, though, I need to take you back to 1966, when a song called, surprisingly enough, “Bus Stop” was a big hit for The Hollies.

Nowadays, “Bus Stop” isn’t the Hollies’ song most people would instantly think of. That’s much more likely to be either “He Ain’t Heavy, He’s My Brother” or “The Air That I Breathe” from the early 1970s.

But, in its day, “Bus Stop” was a decent hit, reaching Number 5 on both the UK and US charts.

“Bus Stop” was written by a 20-year old Graham Gouldman, later of 10cc, and is quite a sweet story of a chance meeting turning into true love. It starts like this:

“Bus stop, wet day, she’s there, I say

Please share my umbrella

Bus stop, bus goes, she stays, love grows

Under my umbrella”

As well as being a very neatly written set of lyrics, the harmonies on “Bus Stop” are every bit as tremendous as you’d expect from The Hollies.

But the other key passage is probably this:

“Came the sun the ice was melting

No more sheltering now

Nice to think that that umbrella

Led me to a vow”

You can probably imagine how the rest of the story goes, although there’s a link to it below if you’re not familiar with this classic piece of 1960s pop.

But, in summary: a chance meeting at a bus stop turned into romance and, ultimately, marriage.

The bus stop problem

Unless you have a heart of stone, I think we can agree “Bus Stop” tells a sweet story.

But if you have an MBA from an Ivy League college, or if you work in tech, this whole episode will instantly strike you as inefficient and capable of significant process improvement.

First of all, the fact that The Hollies’ lead singer, Allan Clarke tells the story of just wandering around aimlessly and bumping into a girl entirely by chance suggests a serious lack of strategic thinking on his part.

Surely, our friends with Ivy League MBAs or who work in the tech industry would argue, this is a ridiculous way to find a girlfriend. What you need to do is set a clear strategic objective and then develop a series of SOPs to achieve your goal, which you can track by collecting and reporting data against KPIs on a daily, weekly, and monthly basis.

So let’s look at how they might go about that.

Since The Hollies were from Manchester, let’s use that location to flesh out their strategy.

A bit of research tells me there are about 12,000 bus stops in Manchester, which is a big number. If someone working in tech was going to try to find a girlfriend at each one of those, and only visited one of them each day, it would take them about 32 years to visit them all.

Clearly, going at that pace is for losers when you’re used to “moving fast and breaking things”? Obviously, they need to increase activity levels.

In the UK bus stops are about 300-400 metres apart on average, further research tells me. So if our lovelorn tech pal was prepared to walk, say, 5 miles a day, they could visit about 20 bus stops a day, on average. That would reduce the elapsed time down to just a couple of years, which sounds more do-able.

But wait!

Putting aside for the moment how creepy an idea this is (very on-brand for the tech industry, mind you…) a strategy of approaching random women at bus stops in Manchester with an umbrella only works on days when it’s raining. Inviting people to shelter under your umbrella on days when it’s not raining is either going to get you sectioned or arrested.

Historic weather data says it rains on 140-150 days a year in Manchester, which means that for about 200 days a year, the umbrella idea won’t work.

So we’re back up to 4 years of searching, or thereabouts, if our tech pal walks 5 miles a day with their umbrella, visiting an average of 20 bus stops a day only on days when it’s raining.

To complicate matters further, I suspect that some of those rainy days will be at weekends, when people might be less inclined to use the bus because they don’t need to get to work. They might choose to stay at home on a rainy Saturday instead of hanging around bus stops hoping that someone with an umbrella to share might come along and stand next to them.

All things being equal, I wouldn’t be surprised if our tech pal’s hit rate is a lot lower on wet weekend days, reducing their chances of success.

So maybe a better strategy is for them to keep their search to weekdays only, in the hope of catching legions of commuters either on their way into work or their way back home again. That’s when the maximum number of people are likely to be found somewhere near a bus stop, improving the efficiency of this process dramatically.

Taking weekdays only, and assuming Manchester’s 140-150 days of rainfall occur evenly across the days of the week, that means there are 52 weeks in the year, and 5 days a week while people are at work to play with. Approximately 40% of those working days, or two days a week, will be rainy, and therefore viable candidates for our tech pal’s umbrella strategy.

That, in turn, gives us 104 days a year (2 working days a week for 52 weeks of the year) which, at 20 bus stops a day, means it will take about 6 years for our tech pal to get around them all.

Which sounds like a long time.

Hyperscaling

Thankfully, the tech industry has a way of dealing with this. You take a fundamentally stupid idea and throw money and resources at it, in the hope that you get lucky and something vaguely like the result you were hoping for pops out the other end quickly enough for you to declare victory before the VC money runs out.

This is called hyperscaling.

Since walking is inefficient, maybe our tech pal should get a taxi from one bus stop to another, minimising their walking time. That means they can double the number of bus stops they visit on days when it’s raining.

According to Manchester City Council, a 10 mile taxi ride (i.e. double our original 5 mile target) costs about £30. So if our tech pal runs this strategy on each of the 104 weekdays a year when it’s likely to be raining, that would cost about £3,000 a year.

But wait. That fare for a 10-mile journey isn’t quite right. It significantly understates the true cost.

Believe me, this won’t be the first time a business plan based on data that’s complete baloney has floated across a VC’s desk, but maybe our tech pal should try to make sure their plan reflects at least some approximation of reality.

So let’s think about this again.

On the assumption that buses in Manchester come along every 20 minutes on average (no idea if that’s the right number or not – it’s just a guess), then our lovelorn tech pal needs to hang around each bus stop for at least 10 minutes or so before each bus is due as that’s when they are most likely to find other people waiting for a bus.

With that strategy, they would need to keep the taxi waiting for 10 minutes at each bus stop at a cost of about £4 per bus stop, again according to the Manchester taxi rate card. Multiply that by 40 bus stops a day, now that we’ve upped the coverage to 10 miles a day, and that’s an extra £160 per day, or almost £17,000 a year, just in waiting time.

Add that to the taxi fare, and assuming our tech friend doesn’t take a taxi to the first bus stop of the day or one back home again at the end of their adventures each day, and they need to reckon on spending about £20,000 a year in their quest to find a girlfriend.

However there’s another problem with this model.

It assumes the bus timetables work in such a way that our tech friend can get around them all in 10-minute intervals, which is unlikely to be true. We probably ought to double the waiting time allowance to allow for some inefficiencies where our preferred taxi route and the Manchester bus timetabling system do not allow maximum efficiency in terms of our taxi expenses.

So let’s call that a £40,000 a year cost over, potentially, a three year period at our increased 10 miles/40 bus stops rate on rainy days.

Which is a potential £120,000 over a three-year period if our tech pal needs to get round every bus stop in Manchester before someone agrees to shelter under their umbrella.

It could be less than that, of course. Maybe at the very first bus stop our tech pal bumps into the love of their life and they’ve done the whole thing for under a tenner.

But the odds are against that.

The odds

Leaving aside the fact that this whole thing is a bit odd, the odds of approaching a random human at a bus stop on a wet day and them agreeing to shelter under your umbrella are infinitesimally small just for starters.

The odds of that person then turning out to be the love of your life are so small that, even across a pretty big number like 12,000 bus stops, you’d need a microscope to see the chances of success.

Now, in sweet songs from the 1960s, and Jennifer Aniston romcoms, that doesn’t matter much. There, the concept is being used purely as a creative device. We’re not trying to bring it to fruition ourselves.

But in the real world – something that doesn’t often feature in tech industry business plans – the “approach random humans at bus stops” strategy isn’t very likely to pay off.

If you work in the tech sector, this won’t trouble you much, because people working there are mostly delusional. Or on drugs, which amounts to much the same thing, in practice,

In tech, the prevailing belief is that provided you throw enough resources at any problem, success is pretty much assured – all you need is the data to demonstrate there’s the potential of a positive outcome and diligent activity tracking.

To be able to do that, our tech pal probably needs to build some custom mapping software and have a way to track which of Manchester’s 12,000 bus stops they have already visited.

Some route planning software is probably also required, to make sure they’re being as efficient as possible as they navigate your way from one bus stop to another and minimise their taxi costs. It’s the least their investors expect.

They’ll also need some report writing software in order to keep the investors happy and to enable them to prepare pretty charts and graphs to demonstrate how hard they’re working towards the end goal.

And, of course, a computer to run this on, together with a budget for tokens because if they’re insane enough to try this strategy, they’re probably also insane enough to use AI to help them.

I don’t know what all this would cost, but let’s add in £10,000 a year on top of our existing costs to give us some idea.

We’re now up to an investment of £150,000 for our tech pal to stand a chance of finding a girlfriend.

The key point here – in fact pretty much the point of this entire article – is that spending £150,000 in return for an infinitesimally small chance of success isn’t smart. Just because you can identify some data, work it into a plan, and throw financial and other resources at it, that doesn’t mean this is a good idea.

And, like most of the tech industry’s ideas lately, the whole concept is creepy as hell. Just writing out the plan above makes my skin crawl.

Furthermore, even with all that data – and more that you’ll pick up from your route-mapping software and monthly reporting decks – this has done absolutely nothing to increase the odds that someone you approach will agree to shelter under your umbrella.

You’re dealing with odds no better than The Hollies’ chance of walking up to a random bus stop and offering a gesture of kindness to someone standing there in the rain.

While The Hollies’ chances of ending up with “…a sweet romance / Beginning in a queue…” are low, they didn’t spend £150,000 trying to find a sweet romance. They just went to catch a bus one day when it was raining and happened to have an umbrella with them.

In fact, The Hollies’ chances are probably better than our tech friend’s because you’re very likely to end up on a register, or in jail, if you spend 3 years propositioning women at every bus stop in Manchester every day it rains.

Drucker’s view

A lot of tech is like this fictional example.

I prefer the late, great Peter Drucker’s view – “There is nothing quite so useless as doing with great efficiency something that should not be done at all.”

The mere fact that you can compile a superficially plausible proposition, backed with data, and with some impressive-looking graphs and charts, does not mean that you’re doing anything worthwhile. If you work in tech, you probably think those two are synonymous. But they’re not.

You may be pursuing your own personal fantasies, but frankly in this example our tech pal is just being an obsessive creep who intrudes into other people’s lives in pursuit of their own personal gratification. (In fairness, you could say that about pretty much every tech CEO, so this is a critique of the industry as a whole, not a particular individual.)

Much like AI, this “bus stop project” doesn’t deserve to succeed because of the societal harm it causes and the monumental waste of money required to stand a one-in-several-billion chance of the plan actually working.

Along the way, you’ve creeped out most of the women in Manchester and you’ve probably seen the inside of a police station holding cell on more than one occasion.

This will come as a surprise to Ivy League MBAs and people working in tech, but no matter how excited you are about the outcome, how many stats you have showing what a good idea it is, and how beautiful your project plan and progress reports are, there are some things which are better not done at all.

The tech person in our scenario has burned through £150,000 in cash, made people feel unsafe in pursuit of their own personal gratification, and almost certainly still not achieved their objective, even though a lot of data analytics and graphs have been compiled along the way.

So next time someone comes to you with an investment proposal, consider the financial and the non-financial implications of their idea.

Surprisingly often, the best course of action for your bottom line is to do nothing. It genuinely can be cheaper to pay someone to do nothing than to let them make a jumbo sized dent in your cash flow for no bottom line return.

While doing nothing might not seem quite as satisfying as taking action, a lot of activity for zero outcome, no matter how pretty your charts and graphs are, is a monumental waste of time and money.

Too few organisations realise that activity alone as a metric is profoundly harmful to their bottom line. It’s entirely possible to be really busy whilst having a negative impact on the bottom line, as we’ve seen above.

One of the fastest ways to create a positive bottom line impact is to take Peter Drucker’s advice and just to stop people doing pointless tasks. That includes closing down crazy ideas with a low chance of success before they take up too much time and money.

Sadly, in too many organisations, that’s a lot rarer than it probably ought to be.

---

Finally, as promised, here in all their glory, are The Hollies, singing “Bus Stop”…

If you know me even slightly, you’ll know I’m a big music fan. So much so that a colleague introduced me to someone recently as “a huge music nerd” – which I took as great compliment, given the equally-huge musical nerdiness of the person making the introduction.

But beyond my occasionally nerdy tendencies when it comes to music, I firmly believe that music (and the Arts more generally) can teach us a lot about business.

Today’s article is about unsung heroes – the people without whom some of the best songs ever written would not exist. Yet their names aren’t (huge music nerds aside) on everybody’s lips – specifically the people who wrote the words to the songs we all know and love.

Think about it – we can hum along with a melody, of course. But we can only sing along if there are some lyrics alongside the music.

And even if we don’t get the words quite right (quick Boomer reference: “My ears are alight”…IYKYK) we still sing along regardless. Catchy tunes are always welcome, but it’s generally the words that make great songs great.

However, the names of people who wrote the words we sing along with are often all-but invisible, with the person who composed the music or the person who gave the iconic performance of the song are generally much better-known than the humble wordsmith who wrote the words we sing with.

As a simple test, with some very simple lyrics, try imagining “YMCA” without the words. It’s an iconic, hook-laden, joyous, danceable track that gets everyone up on the floor to do the dance moves. It’s still a great tune. But nobody is singing along…and they’re probably not even getting up from their chairs to dance.

And that’s just four letters of the alphabet. Imagine what a whole song can do when a great lyricist is involved, with the whole alphabet at their disposal.

Quick business aside – I’ll keep this brief. Almost anything worth doing in business is a multi-dimensional activity. It’s one reason most tech solutions fail – their programming is based entirely on surface-level logic, which assumes humans handle one activity at a time sequentially when, in fact, we handle dozens, if not hundreds, of simultaneous inputs in parallel, each of which we apply in a range of intensities which vary according to how we feel, minute-by-minute.

That’s something the world’s most powerful computers are unable to do. Or, in my view, will ever be able to do.

So it is with music. It’s not just the music, important though that is. It’s the words. It’s the musicians. It’s the backing singers. It’s the indefinable groove that great players bring to their playing – something that goes so far beyond reproducing the notes on a page, ordinary mortals can’t quite figure out how they’ve done it.

And so it is with your organisation. Whether you realise it or not, your organisation runs largely – perhaps almost entirely – on a range of unsung heroes without whom everything would fall apart. People whose names don’t make it into the board minutes, and who may be almost invisible on your organisation chart. Yet they are the glue which holds your organisation together.

If this article inspires you at all, I’d urge you to make it your business to find the unsung heroes in your organisation and make sure they know you couldn’t do it without them.

Lorenz Hart

This article was inspired by a Sammy Davis Jr performance I heard recently. He was singing “My Funny Valentine” – composed by Richard Rogers, with lyrics by my all-time favourite lyricist, Lorenz Hart.

Sammy Davis Jr, an immense talent in his own right, introduced his performance as “a song by Richard Rogers, one of our finest composers”. The second part of that is absolutely true – Richard Rogers is one of the finest composers who ever lived.

But the first part is not. Beautiful music though it is, Lorenz Hart wrote the words we sing along with, not Richard Rogers:

My funny Valentine

Sweet comic Valentine

You make me smile with my heart

The best-known song Lorenz Hart wrote the lyrics for is probably “The Lady Is A Tramp” – most famously performed by Frank Sinatra, but also recorded by just about everyone who is anyone. (I have a particular soft spot for the Tony Bennett/Lady Gaga version though).

Without an unsung hero in the shape of lyricist Lorenz Hart, we wouldn’t start singing along as soon as we hear “She gets too hungry for dinner at eight…” before we go through a series of Lorenz Hart’s bitingly-humorous lyrics which take the mickey out of New York’s “polite society” in the 1920s and 1930s.

(Fun fact: both “The Lady Is a Tramp” and “My Funny Valentine” – two songs almost as different from one another as it’s humanly possible to write – both come from the same musical: “Babes In Arms”. Rogers and Hart were a songwriting team par excellence, before Lorenz Hart’s untimely early death in 1943.)

Hal David

Not long after listening to Sammy Davis Jr’s introduction to “My Funny Valentine” I was watching a TV show based on, and I quote, “the songs of Burt Bacharach”.

While it’s absolutely true that Burt Bacharach wrote the music for some of the greatest songs of the 20th century, he didn’t write the words. Almost without exception, every Burt Bacharach song you can sing along to had its lyrics written by Hal David.

The Burt Bacharach/Hal David catalogue is so extensive I could probably write 50 articles just about that, but includes Dionne Warwick’s “Walk On By”, Dusty Springfield’s “The Look Of Love”, and The Carpenters’ “Close To You”.

My favourite Hal David lyric is from another Dionne Warwick song, though, “Do You Know The Way To San Jose”:

In a week, maybe two, they’ll make you a star

But weeks turn into years, how quick they pass

And all the stars who never were

Are parking cars and pumping gas

Hal David’s wonderful storytelling chimes very much with the unsung heroes theme. He is writing about someone who went to Hollywood in the hope of becoming a movie star, but ended up moving back home to a dead-end job in a (then) quiet town in order to survive after the studio system spat them out.

How many people with immense talent might be hiding in your organisation, who most people only ever see as doing the equivalent of parking cars and pumping gas.

Bernie Taupin

Pretty much every Elton John song you know had its lyrics written by Bernie Taupin. Yet you could walk past Bernie Taupin in the street and not know who he is.

The chances of you walking past Elton John in the street and not knowing who he is are pretty much zero.

Bernie Taupin is the unsung hero behind Elton John’s breakthrough hit “Your Song” – written when Bernie was still a teenager. Those tender lyrics have stood the test of time more than just about any other song from that era.

He is also responsible for the lyrics of “Candle In The Wind”, originally about the tragedy of Marilyn Monroe’s life and, many years later, sensitively rewritten for the funeral of Princess Diana. That metaphor of a candle in the wind is a particularly haunting one for both those women. Almost 30 years later, the 1997 version of “Candle In The Wind” is still the UK’s best-selling single of all time.

There are very few more poignant lines in popular music than:

Goodbye, Norma Jean

Though I never knew you at all

You had the grace to hold yourself

While those around you crawled

Bernie Taupin wrote those lines while he was in his early 20s – words with a maturity and artistic sensitivity that most people 30 years older who have been writing songs for 30 years would struggle to match.

In Elton John’s case, we know how pivotal Bernie Taupin’s lyrics are. To give Elton John credit, he has often explained that their songwriting process is for Bernie Taupin to hand him a completed set of lyrics, to which Elton sets the music.

So, in a very real sense, without Bernie Taupin, we would never have had the songs most people describe as “Elton John songs”.

Purpose

The purpose of this article is two-fold (beyond giving me a thinly-disguised excuse to write about a nerdy fascination of mine):

1. It’s to highlight that every successful endeavour has its unsung heroes. Their names might not be the ones in lights, but the results wouldn’t have been anything like so successful without them.
2. And secondly, it’s to encourage you to seek out the unsung heroes in your organisation and nurture them. No organisation becomes great without great people, and often the greatest people you employ are hiding in plain sight, parking cars and pumping gas.

I get a lot of sales pitches at the moment from people telling me they can transform my business with AI.

Hardly a day goes by that I don’t see some tech CEO high on ketamine enthuse about robots taking over the world.

And there is no shortage of bots lining up to tell social media users that if they don’t fully give themselves over to AI, they have no future and might as well go and live in a cave now.

I have major problems with all these arguments – mostly because they are all baloney – but there’s one particular problem the world’s tech enthusiasts complete miss. And that’s the fact that any business not being run by amateurs or idiots is going to need a business case, with a bottom-line RoI for whatever you want them to do that’s different to what they’re doing now.

Now it’s possible to lie about the bottom-line benefits. And based on what I’ve seen, that’s mostly the strategy of tech cos nowadays. But even then, the suppliers will be found out eventually, giving them two problems: firstly, they’ll be forever tarnished as liars; and secondly nobody who isn’t an idiot is going to buy from them a second time, even if they claim to have fixed whatever caused the problem the first time.

When I say “lie about the benefits”, that’s sometimes…often, perhaps…a deliberate outright lie.

But quite a lot of the time, it’s because people who think they’re smart imagine they understand things they don’t, and so think what they’re doing makes sense when it doesn’t.

Problem 1: Poor risk/reward ratio

By way of example, let’s look at a field I know well – accounting. Although exactly the same principles apply in every other field too, based on discussions I’ve had with a wide range of experts over the last couple of years.

Firstly, let me say, the AI pitchmen often tell me that their AI system can make it much more efficient to analyse data and convert all the things I do manually into some magic reporting machine which delivers instant results any time I want.

As tech pitches go, that isn’t the most insane one I’ve ever had. And on some superficial level, if you don’t understand what a real CFO does, it sounds like it’s delivering a benefit on some level.

Not a bottom-line benefit, exactly, because no cost has been reduced in this process and no revenue gained. But I acknowledge there’s the hint of a benefit in there.

What that means in practice is that this tech solution provider wants me to give them money and, in the process of doing so, my costs won’t reduce and my revenues won’t increase.

That locks in a negative RoI, to at least the extent of whatever this solution provider wants to charge me. I’m struggling to see the business case here.

The much bigger problem for tech providers is what do they think I do now?

I mean, if I thought about it really hard, I could maybe think of a report I might need for something that I don’t get at present, but by definition (if I haven’t already worked out a way to do it) the impact of knowing that information on the business is marginal at best. We already know that because, if it wasn’t, there would already be a report with that information on it, however inefficient it was to compile.

I mean, if I’ve already got a modern accounting system (even relatively basic packages like Sage and Xero, never mind NetSuite or Microsoft Dynamics) I can produce pretty much any report I want with the in-built reporting tools.

I can set up a live, real-time link to Excel to do any analysis I want. And if I want to do something really fancy, I can always hook up Power BI and produce as many pretty tables and charts as it’s possible to imagine.

So, if I can already get all the reports I need or want, more or less instantly, with the systems I already have, by definition no AI tech solution can provide better reporting than I already have.

As a side note, AI systems are incredibly vulnerable to bad actors, so there is a very real business risk to putting anything on an AI system given how many people can successfully hack their way into AI systems, whether they’ve been coded in Silicon Valley or someone’s back bedroom.

So, not only do I not get anything I can’t already get now. I introduce a whole new set of risks and vulnerabilities into my organisation at the same time. You can understand why I don’t see this as an attractive risk/reward ratio in the context of what we have already established is a negative bottom-line benefit.

Problem 2: Low incremental RoI

The second problem all these AI services have is that they are having to generate an RoI over and above the cost of using them in a field where a huge amount of optimisation has already taken place.

To listen to a tech co’s sales pitch, you’d imagine accountants still used quill pens and handwritten ledger books. Relative to those, of course some new tech would deliver dramatic efficiencies.

The only problem is that’s what everyone already did in the 1980s.

When I was a very, very junior accountant straight out of university, we still had a small number of clients who kept handwritten ledger books (albeit the quill pens were no longer in use). But most of our clients were on a computer system of some sort and those that weren’t soon were.

By the early 1990s everyone was on a computerised accounting system, pretty much, and the days of accounting departments with a couple of dozen ledger clerks, cashiers, and accounts assistants to post all the accounting entries had long gone.

And over time, computerised accounting and reporting systems have got better and better, driving more change and greater efficiencies inside most businesses.

An accounting department which might have been 20-people strong in the 1980s is probably being run by 4 people today – perhaps sales ledger, purchase ledger, payroll, plus someone in charge. That’s about all you need unless you run a massive international business (which most companies are not). And a small business can probably get by with just a couple of people.

So whatever whizz-bang tech solution that comes along isn’t making a business case by comparing itself against an operating model which hasn’t existed since the 1980s. All that does is demonstrate how little most tech cos understand the real world of accounting today and just makes the person delivering the sales pitch look out-of-touch and ridiculous.

While it takes some work, of course, showing a positive RoI based on making cost-savings against a staff base of 20 isn’t the world’s hardest thing to do. But, after 40 years of optimisation, an argument which requires you to fire 3 of the remaining 4 people to generate a positive RoI is unlikely to fill anyone with enthusiasm.

Problem 3: It’s not all the same

Because tech folk don’t understand the real world, they think everyone in an accounting department is a fungible unit of resource – that is, any one of them can be interchanged with any other one of them.

Even thinking that shows how naïve and out-of-touch tech folk are.

The reality is, if you have a four person team in the structure outlined above (sales ledger, purchase ledger, payroll and someone in charge), you almost certainly have four entirely different personality types and skillsets, none of which you can eliminate entirely from your business.

A good sales ledger person is likely to be more like a salesperson than an accountant. They have to sell, persuade, cajole, and encourage payments out of their customers in a way that a tech solution of just sending one Dunning letter after another could never hope to do.

I’ve only worked with a small number of truly exceptional credit controllers over the years and the people who do this well are an incredibly rare breed. However, good credit controllers accelerate your cash flow like nothing else on earth. You would need to be insane to get rid of one of those, if you’re lucky enough to have one.

Your purchase ledger person will be very focused on transactional processing, but they play a key role in managing your expenses. Particularly in making sure your expenses are correctly categorised – if they are wrong at this stage, any downstream reporting will be incorrect, so they need a level of focus and attention to detail which most people outside accounts departments don’t have.

They also have to deal with a barrage of suppliers wanting to be paid, queries internally and externally, balancing supplier statements to the purchase ledger accounts and more.

This sounds like the sort of thing tech would be good at, but it isn’t particularly as there is a lot of fine judgement involved. To give one simple example, buying a pen would have an entirely different accounting treatment depending on whether that pen was something for your warehouse team to sign delivery notes with, an antique Montblanc pen you’ve bought as an investment, or one of a batch of pens you intend to sell in the normal course of business as a stationery supplier.

Machines aren’t going to guess the right account code between those scenarios just based on the knowledge that a pen has been bought any time soon.

If you have a dedicated payroll person, you’re running an organisation with a large and complex payroll. That requires deep expertise in HMRC’s ever-growing and ever-more-complex rules and regulations which nobody can pick up instantly, and tech solutions can’t understand the nuances, and oddly inconsistent rules, involved in our tax system.

They will also be creating a range of legally-required returns to a variety of official bodies and fending off queries from your staff who don’t understand their payslip or think they’ve been underpaid or overpaid.

Because of the legal and financial consequences of getting this wrong, if you outsource this area of your business to an AI robot, I hope you enjoy prison food.

Finally, your “person in charge” is compiling all the reports, doing the monthly accounts, liaising with the auditors and dozens of other things. Anyone who thinks this is a purely mechanical process either doesn’t understand nearly enough about how accounts are prepared or they’ve worked with a lot of third-rate accountants.

While of course, some of this is relatively mechanical, it’s often in the process of doing the mechanical work that you notice anomalies which require further investigation before you can be sure the accounts are right. A tech tool which just reports the numbers as they find them is of very little use at this stage in the process – human insight is what you really need at this stage.

Which, by definition, is something you can’t get from something, er, a non-human.

So, when accounts departments have shrunk to perhaps 20% or less of the number of people they would have had before computerisation came in, any new tech service has to justify an RoI from removing people who are, to a large extent, un-removable because you need them, and their respective skillsets, within your business.

And if you don’t have an accounts department which looks like that, whoever has been running it clearly hasn’t been doing a great job over the years. In this scenario, you don’t need a new tool. You just need to make sure you’re getting the value out of the tech you already have and your costs will drop away instantly, without any further investment in new tech.

All the benefit of using the tech you already have better flows straight to your bottom line.

All the areas where you bring in a new system are taking away from your bottom line and, in the absence of offsetting cost-savings, represent a drain on your bottom line, not an addition to it.

In a nut-shell

I realise this looks different if you’re some Silicon Valley wunderkind who doesn’t know what they’re doing, but every AI pitch for “transforming my accounting system” looks the same.

They provide zero functionality over and above what I already have with my existing tools.

They offer no realistic cost-saving opportunities because we’re not in the 1980s any more and I’ve got an accounts team of just 4, not one of 20.

If I really pushed it, I might get that down to 3 (at the cost of some significant loss of valuable skills), but then a tech solution has to cost significantly less than one full-time person’s salary, or there’s no RoI on the investment.

I genuinely believe you’d be insane to get rid of a £30k salary to spend £100k on a tech solution, but tech people don’t understand how accounting departments really work, so they think spending £100k on tech “will make you more efficient”.

It doesn’t. At best it’s a naïve misunderstanding. At worst, it’s an outright lie.

And given that one of my fundamental operating principles is not to deal with naïve people, idiots, or liars in any field, that makes the tech co’s sales pitches hard to take seriously.

It’s not that I’m against tech or don’t think there are further efficiencies out there somewhere. I spent most of my career from the 1980s onwards implementing new systems and driving efficiencies to improve bottom line performance. So I’m not against the principle of this at all.

But on a practical level, AI solutions which claim to transform accounting departments just don’t.

Their “solutions” are inefficient, packed full of additional risks, generate no real-world cost savings anywhere which isn’t still using quill pens, and delivers no meaningful features which aren’t already available to an averagely-competent accounts team through their existing systems.

One day, in light of evidence to the contrary, I might change my opinion. But right now, despite ketamine-fuelled tech CEOs’ claims, there is no compelling case to replace any accounting system with an AI one.

And an even less compelling case to add it in as an extra layer on top of systems which can already do everything you need and which you’re already paying for.

If you really care about your bottom line, think twice about you fall for the tech cos’ sales pitches. They are mostly not true – your only job is to decide whether they are deliberately lying or if they just don’t know what they’re talking about.

No, this isn’t a song by a Village People tribute band. (Although I kind of wish it was.)

Nor is it a reference to the art and science of macroeconomics. I work with a fine group of people who are vastly more qualified to write on that topic than I am. So I’ll leave that to the experts.

Rather, the article title refers to those tricky little things in Microsoft Excel called “macros” which are essentially a way of automating a sequence of activities so that something which used to take hours to do manually can be done with just a couple of mouse clicks.

Before we go any further, however, don’t despair – I’m not going to teach you how to write Excel macros or run through a list of my favourite macros of all time. This article is all about efficiency gains and automation (bet you’re wishing it was about my favourite macros of all time now…).

In the old days…

I consider myself fortunate because I’m just old enough to, at the very start of my career, have experienced what it was like to keep a set of accounts entirely manually. Handwritten ledgers. Manual extended trial balances. Reconciliations written out longhand.

I didn’t do a lot of it as computers were fast making inroads into the world of accounting. A year or two after I started work, keeping your accounts by hand seemed anachronistic, but once upon a time…and not even all that long ago…it was how everybody kept their accounts.

Now, accounting is an easy thing to computerise. It’s all about facts, numbers, and surface-level logic – a perfect match for the digital world because that’s exactly how a computer runs. And still do – while they’re a lot flashier and generally more reliable than early PCs, the underpinnings of a modern computer are exactly the same as the prehistoric models we used in the late 1980s and early 1990s.

The reason I consider myself fortunate to have kept a set of accounts manually is because it gave me a perspective most accountants today don’t have – because they never worked in a manual bookkeeping environment, they have no baseline for comparison.

It also gave me a level of understanding of the nuts and bolts of the accounting process which are mostly invisible to users of modern computerised accounting systems, where many things just take place “as if by magic” in the background. In the old days, I’d have had to enter all of that manually somewhere.

While I’m not pretending this has been a skill I’ve used very often in the last 30 years or so, nonetheless there have been a few pivotal moments where I’ve worked out the answer to a particularly thorny problem faster because I understood exactly what was going on in the background from when I had to make those entries manually.

Occasionally, that got businesses I’ve worked for out of some sticky corners.

Then came computers

When computers came along, they fairly quickly got to grips with accounting thanks to the fact-based, surface-level logic required to prepare a set of accounts.

While modern packages like Sage and Xero are much easier to use than those early ones, and considerably more helpful than their historic equivalents, they all do essentially the same things as a manual set of accounts used to.

All the debits still go on the left and all the credits still go on the right – that’s been at the heart of double-entry bookkeeping for over 500 years.

But what computers did bring to accounting was massive efficiency gains. A Finance Department which might have had 20 people in the 1980s would probably have 4 or 5 people in it now.

The cost savings for finance teams have been massive. A whole swathe of accounts assistants, ledger clerks, and document processors were no longer necessary.

However the point came where it was much more challenging to make further cost savings. After you take 75% of the people out of your accounting department, it becomes less likely that you’ll save another 75% of the remaining staff complement moving forward.

So finance professionals shifted away from automating the “grunt work” of ledger processing, because that had already largely been done. The opportunities for further cost savings were very limited, if they existed at all, so the RoI on pursuing every fine little detail was low.

They did, however, recognise that one area of accounting which had been largely untouched up to that point was report production.

Not reports as in P&Ls and balance sheets – they come off the computerised accounting system just fine already. But the sort of summary reports you might give to the board, perhaps, or a department head – people who don’t need all the details that a professional accountant might need, just a summarised version.

Until fairly recently, that sort of reporting was largely manual still, and involved lots of spreadsheets and keying data into cells so that information could be manipulated inside Microsoft Excel and formatted appropriately for its intended purpose.

That’s where macros came in.

Although macros had existed in some form inside Excel for quite a while beforehand, it’s only been in the last 20 years or so, as Microsoft introduced the version of Excel we still use today, pretty much, that macros have been more widely used.

In the early days, they were clumsy and hard to use unless you had a background in computer programming. Microsoft, to give them credit, have made macros vastly easier to use in the last 20 years or so.

Now most finance departments use macros to automate some or all of the processes for extracting data and compiling regular daily/weekly/monthly reports.

Here’s the funny thing

Here’s the funny thing, though.

Macros were originally billed as making finance departments more efficient. Tasks which might have taken a day beforehand could be done in an hour. I’m sure some people were salivating at all the cost savings which could be generated from introducing macros into their financial reporting processes.

Yet their actual impact on efficiency and cost saving has been more or less zero.

“Wait!” I hear you say. “How come introducing a technology which makes compiling reports more efficient hasn’t allowed us to make cost savings in our finance department?”

(Don’t feel too bad if that was your immediate thought. Most accountants would have thought he same when macros were brought in to financial reporting systems.)

There are two main reasons for this:

Reason 1

If you have a finance department of 4 people (down from 20 people 30-odd years ago), it’s extremely unlikely that you have any single person who spends all their time on financial reporting.

It’s probably 10% of everyone’s job – simple reports being done by a junior person, board reports being done by a senior person, etc – and you’re unlikely to get all 4 of those people to switch to a contract where you only pay them 90% of their previous salary even if you do automate all of the reporting (which I doubt).

So what probably happens is that you keep all 4 people on full-time salaries, meaning a cost saving of £0.

Reason 2

Because writing reports became “easy and simple” people started demanding a lot more of them. The volume of reporting on often trivial details in organisations today would have been considered genuinely jaw-dropping 30+ years ago.

Never mind a finance team of 20, you’d have needed a finance team of 100 to get through what would be considered a normal reporting load nowadays.

It’s a separate topic for another day, but far too high a proportion of the reporting done inside most organisations is either recording trivialities which don’t need to be recorded, or is so widely ignored that it’s a complete waste of time and effort creating the report in the first place.

The combined effect of either or both of those is that, despite introducing lots of ways of making finance teams “more efficient” in the last 30 years, there has often been little or no positive bottom line impact as a result of all this extra efficiency.

Productivity

People often mistake productivity for efficiency and assume the bottom-line results are self-evident. Well, except in rare situations, they’re not.

You could, for example, say that the accounts person who is churning out 100 reports a month instead of 1 is being 100x more efficient than they were in report production.

While that maths is superficially tempting (and, to be fair, a lot of software sales have been made on the back of exactly that maths), if the extra 99 reports are on trivial matters and/or are rarely read by the people who get them that might be seen as more productive from the perspective of the person producing the reports.

However it isn’t more efficient, as far as the business is concerned, because there is no bottom-line impact.

In fact, there has probably been a negative bottom-line impact – not only do you still have the salary you had before, you also have extra software costs you didn’t have before, and no offsetting efficiency gains to factor in. So the bottom-line performance is actually worse than it was before you became “more productive”.

It’s like all those people on Twitter who claim AI makes them 10x more productive in writing software.

In the very narrow sense that they are using technology to write more lines of code, that’s correct. But it’s highly likely that they are writing code to carry out functions which have little or no real-world bottom line utility, in which case how productive they are (in terms of lines of code written) is irrelevant.

From a more strategic viewpoint, it’s all been a complete waste of time.

The questions to ask

Given what I do for a living I get pitched all the time by people who want to make X, Y, or Z more efficient in my business.

I’ve got to say, at the risk of appearing like a jaded old curmudgeon, I can tell at a glance that their claims are mostly garbage.

Most of them are, on the very kindest interpretation, productivity improvements which will have zero real-world impact on the bottom line, like the accountants using macros for reporting above. They do more, perhaps, but it makes no difference to the cost base because we will still be employing as many accountants as we were before, so there is no bottom line impact (apart from the negative one due to the cost of additional software licences).

Alternatively the claimed theoretical “efficiency” turns out to make absolutely zero difference to my life, never mind my cost-base, and is completely pointless.

My favourite example of this at the moment are those services which claim to save you time by getting AI to summarise documents so you don’t need to read them.

I get two sorts of documents.

1. Documents I can read in 10 seconds or less – it takes AI longer to summarise those than it takes me to just read them (I know – I’ve timed them). And anyway who needs a summary of a 3-para document? Just read the darned thing, for goodness’ sake.
2. Documents, often legal in nature, which run to 30 or 40 pages. The last thing in the world I want are those documents summarising as there are far too many nuances which need careful consideration – a summary tool would skip over those details because they don’t look important enough, even though often the success or failure of an entire agreement turns on those details (or lack thereof) in a contract.

So given that 95% or more of the documents I receive are not suitable for a summary tool, that means the cost saving any system brings must deliver a positive RoI on the 5% of the documents I receive which could be considered candidates for a service like this.

Whilst that isn’t completely impossible, it’s a tough benchmark to reach and, in practice, none of the tools software companies are desperate to sell me come close to it.

So when you get a pitch from someone promising to make your business more efficient by automating some task or other, ask yourself these questions:

• Never mind automating that task – does it need to be done at all? If not, your lowest cost outcome is to eliminate that task completely, not to automate it.
• Is it more productive or more efficient? Efficiency is good because that translates into bottom line impact. Productivity isn’t a bad thing, but if it’s just people doing a pointless task faster, you’re kidding yourself if you think that’s going to transform your business.
• Where exactly are the cost savings? Unless you know precisely who you’re firing, or exactly which existing service you’re going to cancel your subscription for when you bring in a new service, you’ll never make any cost savings – you just end up with the same cost base you have now, with the added cost of some new software or service on top.

What people selling productivity and efficiency solutions forget is that the last 30 years has seen a huge increase in productivity for most organisations. That 20 person finance department of 30 years ago is now a department of 4 or 5 people, so generating incremental cost savings isn’t impossible, but it’s an order of magnitude harder than it used to be.

That’s largely because people selling productivity and efficiency solutions are just selling a product – they don’t understand how real businesses operate.

They fantasise that summarising all my emails will save me 2 hours a day when the reality is I wouldn’t get 95% of my emails summarised anyway, so at most they might save me 5 minutes a day and the amount of time, energy and hassle I’m prepared to put up with to save 5 minutes a day is pretty much zero.

I’m certainly not giving anybody’s vibe-coded AI app access to all my emails in order to save me 5 minutes a day – I’m just going to spend that “extra” 5 minutes.

It turns out that the biggest efficiency gains you can make to deliver a positive bottom line impact are cutting out pointless, low value-adding tasks, not automating them – all that does is paper over the cracks of a fundamentally broken system.

And being clear on exactly what costs you are, in the real world, going to eliminate by bringing on board some system for improving productivity and efficiency. If you keep all the same costs, and add in an extra layer of new costs for software on top, you’re losing money, not making money, on that transaction.

To paraphrase the Village People, every accountant wants to be a macro man (or lady). You need to be clear that “doing more stuff” translates into bottom line benefit, or the best decision you can take for your bottom line is – paradoxically – not trying to be “more productive”.

Don’t fall into the traps set by software companies in particular – productivity and efficiency are not the same thing – productivity is a measure out output (100 reports vs 1 report), efficiency is a measure of bottom line impact.

While software companies hope you don’t spot that slight distinction, I promise that it’s better for your bottom line if you do.

Bigger. Better. Faster. Stronger. Cheaper.

No, not that’s not a Kanye West song (although it nearly is…). Today’s article is about what makes us human…and why that should matter for your business.

Given a chance, humans like to -er things. Take something that exists now, and make it into something more than it is now, on some dimension.

Got a car with a top speed of 30mph…how do we get it to 40mph?

Got a machine that churns out 100 widgets a day…how do we get that to 200?

Got a product that costs £100…how do we make much the same thing for £50?

Bigger, better, faster, stronger, cheaper – anything we can do on any of those dimensions means a better product we can sell more of, right?

And we also get a marketing advantage – at least in the short-term while our competitors work out how to copy what we’ve done. Given the choice between trying to sell a product that potential customers see as indistinguishable from every other product on the market and trying to sell a product that’s different from every other product on the market in at least one of those dimensions…I’m taking the product that’s different every day of the week.

Not least because now you’ve given your customers some logical basis to post-rationalise their decision, and head off a bit of buyers’ remorse.

While it’s fashionable to think otherwise, we all buy everything emotionally first and then post-rationalise why we bought it with logic. So you can’t just sell the dream – you have to back it up with logic.

“No, dear, I didn’t waste money buying that fancy car. I bought it because it will get me home faster so I get to spend more time with you.”

Please note I didn’t say the post-rationalisation was actually objectively rational. Just that it provides a framework for the little lies we like to tell ourselves ever since the Victorians rediscovered the ancient Greek philosophers and decided that logic was the purest form of decision-making.

Which is weird, because logic often throws up the very worst decisions imaginable. But then the Victorians – and the ancient Greeks for that matter – never seemed to have any trouble with, for example, supporting slavery or treating women as second-class citizens.

Once you accept the (offensive and wildly incorrect, let me say) position that straight white men are superior to the rest of humanity, then logically slavery and oppression are perfectly justifiable.

There is no purity in logic. It’s just one of the circuits we can use in our decision-making. Which is why wise people recognise the best answers sometimes come from a source other than pure logic.

That said, if you want to sell something to someone, it’s a lot easier if you can sell them emotionally first, and then drop in a logical basis for them to post-justify their decision.

And although I’d recommend only doing that ethically, all of history’s biggest scam artists – up to and including some of the people selling AI solutions today – aren’t especially troubled by ethics.

That’s the pitfall with logic…once you’ve decided logically that stealing as much money as possible from as many people as possible to enrich yourself is a good thing, then logically you should steal from as many people as possible as fast as possible because that takes you closer to your end-goal of living the rest of your life in comfort in a country which doesn’t have any extradition treaties.

And, to be absolutely fair to the ancient Greek philosophers – Aristotle, specifically, in this case – even they recognised that logic wasn’t the be-all and end-all.

I realise if you run a tech business, or work as a politician, it will come as news to you that Aristotle talked about how a blend of ethos, logos, and pathos – broadly, ethics, logic and emotions – were essential to good communication, and by extension to life itself.

The “logic” bit was only one element out of three. Not the only basis for interacting with humanity and making decisions.

Change management

One place all these factors often bump into one another, like a multiple car pile-up on a motorway, is when it’s time to change something inside a business.

You know the sort of thing – sales aren’t where we need them to be, profitability has taken a hit, our costs are too high relative our competition. So it’s time for change.

The wisest words I ever heard on the subject of change management were from an early boss of mine, who used to say, “Changing things is easy. It’s knowing what to leave the hell alone that’s difficult.”

His view was that any idiot can throw everything up in the air and make whatever they were doing before different.

Often that’s why people who are described as “having a bias for action” are so highly prized in organisations, because they throw everything up in the air and make changes everywhere.

To the uninitiated, that looks like a real go-ahead guy or gal…someone we ought to be giving a promotion to.

More often, in my experience, it’s a sign of someone you should be demoting, if anything, because they don’t understand the situation they are changing well enough to make more intelligent decisions.

To channel my old boss’s sentiment – they don’t know what to leave the hell alone.

You see, change management (despite the name) isn’t actually about changing things willy-nilly.

Its about getting a better answer – typically, in my case, an improvement to a business’s bottom line, but the concept applies equally to making more sales or engineering a more reliable product. Once you get your head around that concept, it becomes more obvious that getting a better answer might, at least in principle, be a blend of things you do differently and things that stay the same.

That’s why the term “change management” sets my teeth on edge. “Business Improvement” would be a much better term, but we are where we are in a world that thinks change for its own sake must necessarily be a good thing.

But there’s a flip side to this as well.

As Albert Einstein supposedly said, according to a number of sources on the internet, the definition of insanity is doing the same thing over and over again while expecting different results.

So if you want to improve your business results, leaving everything the same is not likely to be a successful strategy. It’s much more likely that you’ll want to change some things, but not others.

Change everything

One of the many reasons I know AI will never work is that it comes from a “throw everything up in the air and change everything” place.

And that includes changing the things that work perfectly well now which, ironically, usually comes with significant downsides to your business in terms of reduced revenues or increased costs – the precise opposite of what you hoped to get from your AI Transformation Project.

To understand this, you have to put yourself in your customers shoes, first and foremost. You can change anything you like inside your business, but if it’s a “meh” to customers – or, worse, a distinct turn-off – then you’ve wasted your time and money, and ended up in a worse position than where you started.

That’s not smart, but it’s what every AI transformation specialist is doing every day of the week at the moment.

Let’s look at an example.

It’s partly a generational thing, I accept, but I like being able to pick up the phone to suppliers to get problems sorted out on the spot, when possible.

I do not like having to raise a ticket on some soulless interface which promises me that someone will get back to me within the net 48-72 hours. By then my business might be irreparably damaged.

Now, I get it that a soulless interface for raising a ticket is a lot cheaper to operate than having someone available to take an inbound call within a reasonable period of time. I used to run a large call centre operation, so I know how the economics works.

However people working in call centres are not usually paid life-changing amounts of money, so the actual cost of having someone take a call – assuming they are kept reasonably busy – is only a couple of quid a call.

If I’m spending £000s with your business and you’re not prepared to reinvest a couple of quid to pick up the phone every now and again to help me solve my problem, then that’s a sure sign of a supplier I don’t want to deal with if I can possibly help it.

“Oh, but Gen Z prefer interacting via digital methods”, people selling tech solutions often say.

That might even be true (although I doubt it). But not many Gen Z people run £million budgets where they need a key supplier to respond to their queries in less than 48-72 hours. Odds are most of the people who do would prefer to talk to someone on the phone and get an instant answer.

So you have to start where your customers are, not where you would like them to be, or where you were told they would be in a fever dream inspired by the glossy slide deck some sharp-suited salesman took you through the other week.

I can’t say this strongly enough – you will lose business from me if I can’t pick up the phone, speak to someone, and get my problem solved in less than 48-72 hours.

By definition if I’m calling you, it’s not because I’ve run out of things to spend my day on. Its because there’s a major issue and we need support right now.

So if you’re making me raise tickets or chat with some AI tool, I can save you a lot of time trying to pitch me. I’m not interested.

It’s not insanity

That doesn’t mean you need to keep things the same.

When I ran a large call centre, we made no end of changes to our service inside the call centre which made us more resilient, more efficient, and helped us provide a better service to customers.

The difference is that, outside the call centre, you’d never know. Or if you noticed anything at all, you’d just be impressed by the speed and efficiency with which we solved your problem.

We didn’t focus on implementing a change from a customer perspective. We focused on leaving the things the customers liked the same.

But what was going on under the bonnet? That was fair game.

So, if I was running an AI transformation, I’d leave the element people liked the same (ie, picking up the phone to an actual human, if that was my preference, and getting my problem solved) and I’d use technology to enable my call centre agents to service clients better.

I might, for example, make it easier to access the details of a customer account with a single click rather than making the call centre agent tab through several different screens on a piece of helpdesk software to find the details they were looking for.

The “making information easier to access bit” – sure, make changes to your systems. The “people pick up the phone and answer your calls” bit – leave that alone.

Generally the problem with change management as a discipline is that its shorn of any responsibility for, or even concern about, the end-result beyond “change everything”.

That’s partly by design, because otherwise AI transformation people (and their predecessors in areas like digital-first transformations, TQM, and Taylorism) need to find as big a cost base as possible to start with so their fees sound like a bargain by comparison.

If you want to sell a service costing £500k, you ideally want a £5m budget to hack away at, because that way you should be able to find enough ways of saving 10%+ of that £5m budget, thereby justifying your fees. (Mind you, I could probably do that right now without an AI transformation, but that’s not my point here.)

It’s not because that’s necessarily the right answer – it’s entirely about making eyewatering fees look like they’re good value, in the context of the budget they’re starting with.

Frankly, saving £500k out of a £1m business is just as valuable. It’s still £500k. It still lands on the bottom line the same way.

But that pitch is much more high-risk for the change management and transformation community, because that means they’ll have to come up with some really exceptional ideas. The odds of them finding something you don’t already know yourself to the tune of making £500k savings out of a £1million budget are very much against them.

That’s why, even though the right answer from a customer perspective is to leave their ability to call up and speak to a human intact, the change management and AI transformation folks need to get to as big a starting point as possible, because then their £500k fee looks like a bargain by comparison.

“Give us £500k and we’ll sack everyone in your call centre, which costs £10m a year to operate and replace it with an AI-powered customer service system costing just £3m a year” sounds like a bargain, but it’s very likely to be the worst decision you could take.

If you sell low-cost B2C products, that might be OK. Necessary, even, in the face of steep competition. And if someone is spending £50 a year with you, that doesn’t leave much of a margin for high levels of customer service or a high-cost infrastructure.

In B2B – and especially if you deal with senior people – that same strategy is likely to be pretty dumb, no matter what the tech folk tell you about people preferring to interface with suppliers via an app on their phone.

Switching costs

Part of the reason you should leave some things “the hell alone”, in my old boss’s words, are that even if the proposed strategy is reasonable enough on some level, the cost or risks of getting there are out of proportion with the benefits.

This is another way the AI magic beans crowd demonstrate they don’t really understand what they are doing. That’s why they are dangerous.

But, despite my long-held scepticism about AI, the same thing applies to any strategy which boils down to “we need to do X everywhere”.

To quote Abraham Maslow – or at least the version of him that pops up on internet quote sites – “When all you’ve got is a hammer, everything looks like a nail.”

In change management if, before you’ve even started a transformation process, you’ve already decided the only acceptable answer is “AI”, then the likelihood of you reaching the best possible solution is precisely zero.

Not just because I think AI is over-hyped (although I do). But if you’re reading this article in a few years’ time after the AI bubble has burst, this applies equally to anything anyone ever tries to convince you is the single magic answer you need to deploy everywhere in your business.

There are many reasons why any “everything everywhere” strategy is bonkers, but it boils down to the fact that when all you’ve got is a hammer, bashing everything on the head as if they were a nail isn’t always the best solution.

If its the only solution you’ve got, you’ll end up doing plenty of things that backfire on you and your business.

Often that’s because, in a “change everything” world, people forget about switching costs.

What I mean by that is, notwithstanding the theoretical attractions of an idea, making it happen in practice would be so expensive, that the smartest thing to do for your bottom line is not to make the change and leave things as they are.

The best example of that I’ve seen was on a visit to Nissan’s Sunderland car plant a few years ago.

Volume car manufacturing is, as I’m sure you know, fiercely price-driven and hugely competitive. Car factories are full of robots because that means a higher line speed and more efficient production.

But Nissan didn’t have a “robots everywhere” strategy. They had a “lowest cost manufacturing strategy”.

So, after going through a number of robotic processes, the production line snaked through an area where 6 or 8 people are all over the car, putting in the seats, making sure the trim is in the right place, doing the final fit on the windscreen, and so on, before going back into a mostly robot-led phase.

I’m sure it’s theoretically possible to get robots to do all those things. And I’m prepared to bet that the people at Nissan know exactly how to make that happen.

But they wanted the lowest cost solution. Not the most robotic solution.

With that objective, there were just parts of their production process where it was more efficient and cost effective to get a human to do the work than to spend tens of millions on more robots.

The people at Nissan understood the brief.

They weren’t people armed only with hammers walking around looking for nails to bash. Their focus was on low-cost production and they understood that sometimes low-cost meant “robots” and sometimes it didn’t.

When they were building their shiny new factory – one of the world’s most efficient car production facilities, rivalling even Japanese levels of efficiency – and robots started to make their way into vehicle manufacturing, the people at Nissan had a very clear idea of what they were going to change and what they were “going to leave the hell alone”.

Turns out there is no such thing as one answer that works for everything, everywhere, all at once.

Rather, there are times when robots in your factory, AI software, or any other “magic solution” you get presented with is the smart way forward.

And there are times when doing that is spectacularly dumb.

I’ve always found that approaches grounded in dogma or ideology, like “AI should be everywhere” dramatically underperform against solutions which blend different elements together to come up with the best practical solution, one which works for your business and your customers equally.

Or, as my old boss David might say, when you want to make changes, one of the most important things to identify are the things you really ought to leave exactly as they are.

Vibe coding is very much the thing of the moment.

Wherever you turn, there seems to be no shortage of people suggesting that banks vibe code apps to help people transfer millions of dollars via bitcoin to some far off-country in order to save on the money transfer fee.

Or that hospitals should decide which patients’ lives should be saved based on the recommendations of a vibe coded app instead of getting a qualified doctor to take a look.

Or taking legal advice from a vibe coded app when you’re being sued for $millions – whether or not that claim has any legal validity.

By and large, the people proposing those strategies are idiots.

Vibe coding is fine for really simple stuff for your own amusement with no downside risk, like animating a photo of a much-loved, but long-departed, pet.

But it’s a really terrible way of handling any of the tasks outlined above because they all require nuance, the ability to manage complexity, and an appreciation of risks that go beyond the formulaic surface-level logic that computer programmes (vibe coded or otherwise) run on.

What’s extra-puzzling, though, is how little appreciation vibe coders have for a mature decision-making process.

By this point in the mid-2020s, pretty much everything that can be put into a simple formula already has been. Large-scale, personalised automation was commonplace in direct marketing, for example, long before vibe coding came along.

And critical business systems, like accounting systems, for example, are highly automated nowadays. Yet, to make sense of what the vibe coders on Twitter are saying you’d have to imagine we still prepare accounts with a quill pen, an abacus, and handwritten ledger books.

Modern accounting systems run reliable, repeatable processes at scale in a way that a vibe coded app is very unlikely to be able to do. That’s because people like Sage and Xero have spent several decades doing their best to make robust, reliable, dependable accounting systems available to pretty much everyone in the world for just a few bucks a month.

Why reinvent a wheel that doesn’t need reinventing, and introduce a whole new level of risk into the process into the bargain? This hardly seems rational.

Yet vibe coders think that transferring $millions around the word, handling complex legal or tax questions, and deciding whether or not someone in hospital should be given life-saving treatment, is a perfectly reasonable thing for vibe coded apps to do.

Even though pretty much none of the people who are happy to be cavalier with your financial information, or making potentially life-altering tax or legal decisions for you, thinks that vibe coding an aeroplane or a car would be a sensible course of action.

That’s because, in their naivety, building a car or an aeroplane seems “difficult” to vibe coders, but weighing up all the factors to make life and death decisions, or making complex legal or tax filings, is “easy” because they think those decisions are made now by following through an if/then decision tree which they can replicate with the aid of some crappy AI chatbot.

I mean, that’s not how they are made at all. But if, like most tech folk, all you see when you look at a situation is formulaic, surface-level logic, that’s an easy trap to fall into. After all, that’s all you need to make a computer programme work.

It’s just not how skilled professionals in any field work in the real world actually operate – at least not in totality. Of course there is some surface-level logic involved. But the ultimate decisions are not exclusively based on surface-level logic, and never can be.

Vibe coding a Cadillac

Let me say again that vibe coding anything non-trivial is a terrible idea.

But, just for the sake of argument, let’s imagine vibe coding a car might be something we’d want to do.

How would that work?

Well, thankfully we don’t need to imagine how it might work, because back in the early 1970s a guy called Wayne Kemp wrote a song about doing precisely that. It was called “One Piece At A Time”, and was a big hit for Johnny Cash in 1976.

Wayne Kemp’s song is the story of a country boy who moves to Detroit to work in a car plant and, envious of the shiny new cars passing him every day, forms a plan to steal a new car for himself.

Obviously, he couldn’t just drive one off the assembly line, as he’d be caught and fired. So he decides to steal a car “one piece at a time” to build for himself at home.

His theory was “I’d get it one piece at a time / And it wouldn’t cost me a dime”.

With the aid of some friends and an extra-large lunchbox: “The first day I got me a fuel pump / And the next day I got me an engine and a trunk / Then I got me a transmission and all the chrome”.

This is the vibe coder’s dream. Starting somewhere, then scrappily iterating their way through a range of tasks, navigating through a series of twists and turns, then magically ending up with a valuable product at the end – like an app. Or, in Johnny Cash’s case, a new Cadillac.

Now, Johnny Cash reckoned that his finished automobile would be worth a hundred grand, which isn’t a bad rate of return for a product where the parts cost him nothing (putting aside the moral and legal issues involved in theft for a moment).

Nowadays vibe coded apps are, so their backers claim, worth $billions. So there’s a similar level of delusion at work here.

One important blind spot

The conceptual problem vibe coders and people stealing Cadillacs one piece at a time share is that they assume a static external environment.

But whether that’s assembling stolen Cadillacs or vibe coding money transfer systems, that is unlikely to be a good assumption.

As Johnny Cash discovered when he came to put his “hundred grand Cadillac” together: “The transmission was a fifty-three / And the motor turned out to be a seventy-three / And when we tried to put in the bolts all the holes had gone”.

This is theory meeting the real word, like when vibe coders discover that most countries tightly regulate their financial services industry, mostly to stop the sort of people who become vibe coders from stealing customers’ money, and their “ever-so-smart app” doesn’t have the necessary security protocols built in.

So after complaining like little kids about the fact that their vibe coded app isn’t approved by the financial services regulators as some sort of conspiracy against them – it isn’t, it just means a bunch of amateurs didn’t take the time to understand the landscape into which they were launching their app – they set about making the app fit the regulations they have only just found out about.

Of course, because they still don’t really understand what they’re doing, they hack their way through, a bit like Johnny Cash.

After realising that all the holes for the bolts had gone, Johnny had a plan: “So we drilled it out so that it would fit / And with a little bit of help from an adapter kit / We had that engine running just like a song”.

At this point, the vibe coders imagine their work is done, but all they’ve managed to create is a Frankenstein-style monster which does nothing but attract the ridicule of people who actually know what they’re doing. Whether that’s car designers or actual lawyers, doctors, and accountants.

“Now the headlight was another sight / We had two on the left and one on the right / But when we pulled out the switch all three of them came on.”

The end result

The end result of vibe coding is a mess, whether you’re vibe coding a banking app or vibe coding a car.

Johnny Cash ended up with a “’49, ’50. ’51, ’52, ’53, ’54, ’55, ’56, ’57, ’58, ’59 automobile”.

The fact that your vibe coded app might have a shiny user interface doesn’t take away from the fact that, under the hood, it’s no less worthy of ridicule than Johnny Cash’s Cadillac.

It’s not that incumbents shouldn’t be disrupted – that’s what capitalism is supposed to be all about.

It’s that a bunch of naïve amateurs are likely to do more harm than good if they just mess around without understanding what’s really going on, the structure of the industry, and any potential legal or regulatory guardrails.

It’s also likely that vibe coded software is bloated and inelegant because it’s been built piecemeal, one bit on top of another, like Johnny Cash’s Cadillac, rather than being approached with a clear brief from the outset and a drive to create the most efficient way of getting the job done.

There’s a reason people don’t vibe code cars.

The end result is a ridiculous abomination of an automobile, with only one tail-fin, and two headlights on the left, but only one on the right.

If you think you’ll run an efficient, cost-effective business with systems assembled “one piece at a time” like Johnny Cash’s Cadillac, I’ve got news for you. Just go out and buy a Cadillac – in the long-run, doing things properly right out the starting gate is always more cost effective than bumbling around for a few years, hoping you might get your steam-powered, Frankenstein monster systems working to the same level as something already designed for the job.

We live in an increasingly complex world, but what often seems to be missing in company decision-making is an acceptance that we need to make choices.

Often, there’s a sense of “we can do everything” when faced with major decisions. But the reality is that nobody can cover every option. Everything is about the choices and trade-offs you make for your business.

Now, there’s a very fertile market in people who want to persuade you that you really can have everything you want. Those people are mostly con artists, or work in AI (but I repeat myself).

The reality is you can’t, though.

A relatively small number of strategic decisions set the template for your business. Once you’ve set that template, you’re pretty much betting the company on that working out.

For example, if you decide that your overarching objective is to minimise customer service costs, then you’ll probably fire all your call centre staff and replace them with AI bots.

The minute you make that decision, though, you effectively decide to run a purely transactional business where customer loyalty no longer matters.

Now, that’s a perfectly reasonable business decision, if you go into it with your eyes open, but now you’re locked into a very specific business model, whether you like it or not.

It means you’ll have to shift your marketing to an entirely “they do it for X, but we can do it for 10% less than X” model.

In the short-term, it’s entirely possible to run an outwardly successful business like that.

At least, until Amazon decides that you’re in a market they’d like to be in. Then you don’t have a business any more, because they’ll out-compete you on price every day of the week without breaking sweat.

Or until the Google ads or Facebook ads which drove traffic to your “we can do it for 10% less” offers dries up because your competitors, who all follow the same strategy, keep bidding up the click costs to the point where it becomes uneconomic to sell your products that way.

And it also means that at least a proportion of the people who used to buy from you might decide that AI-powered customer service and the relentless degradation of your products to keep that 10% price differential isn’t to their liking, so they end up taking their business elsewhere.

Remember, you don’t need everyone to do that for you to have a big problem on your hands. Most businesses, were they to lose just 5-10% of their customers, would quickly run out of cash.

So the choice that you make when you decide that the over-riding priority for customer service is to minimise the cost of delivering it means that you’re locked into all sorts of other decisions, many or all of which have the potential to backfire badly on you somewhere down the line.

“Strategy is about choices”

I’ve always loved that line from Michael Porter. Mostly because he’s absolutely right.

When you choose to do something, by default you decide not to do something else.

You can’t both shut down your call centre and also have high levels of personal service delivered by the people working in your call centre. It’s one or the other.

As a side note, it’s also why I get irritated about the word “strategy” popping up in places it has no business being, just so someone can feel important.

I don’t need an “expense claim strategy”. I just need someone to efficiently refund expenses incurred for business purposes by staff members.

So there’s no strategy there, because you have no choice to make – you’re going to give your staff their money back. There are just some practical decisions to be made about how to pay back business-related expenses to the staff who incurred them.

If there isn’t a very fundamental decision to be made, you don’t have a strategy at all because you have no fundamental choices to make. You just need to take care of something on a practical level.

There is a trick to this, though. You have to pitch the choice at the right level or you’re essentially gaming the decision-making process.

“First” businesses

A few years ago, it was fashionable to have a “digital-first” strategy. Now it’s common to have an “AI-first” strategy.

I can’t say this strongly enough, but neither of those decisions is a strategy-level consideration, although people are often tricked into thinking they are.

And that’s because the answer is in the question.

If you’ve already decided that digital is the way forward, from that point onwards you are only choosing from a range or digital options. Every non-digital option has, by definition, been excluded when you set a “digital-first strategy”.

That also means, by definition, if there is a better non-digital strategy, you will completely ignore that in favour of a worse outcome, because the decision has already been made that you are only able to select from the range of digital choices available.

From a business strategy point of view, this is nothing less than insane.

You might think I’m being a little harsh here, but when you set up your decision-making structure in a way which ensures your business can only choose from a range of sub-optimal outcomes, “insane” is the politest word I can think of to describe this.

You might well have a customer service strategy, within which there are some digital elements and some non-digital elements, for example, to reflect the areas in which each of those options is stronger.

But to have a digital-first strategy, and with that decision, rule out every other potential option – even options which offer a lower cost and better service than you have at the moment – is wild.

However the only reason businesses do that sort of thing is because tech vendors have essentially conned people into thinking they don’t need to make a choice, because…as if by magic…they can achieve perfection through a range of entirely digital options.

What this essentially does is outsource your strategic decision-making process to whichever shiny-suited salesperson has just waked into your office with a glitzy slide-deck in hand.

If you fall for that, good luck to you. You’re going to need it.

Strategic opposites

The great Rory Sutherland has a line about the opposite of a good strategy also being a good strategy.

So, a price-driven, high volume retailer like Asda can be successful. As can a high-price, high-touch business like Fortnum & Mason.

Their strategies are complete opposites, but both strategies make sense. The job of each of those businesses is to make sure all the other decisions they make are consistent with their respective strategies.

If Asda starts selling vintage brandy-infused cherries for making $100 cocktails, they’ve gone off-track. As would Fortnum & Mason selling bumper value family packs of own-brand cornflakes.

The key here is not that, on a product by product basis, both Asda and Fortnum & Mason, couldn’t probably articulate a business case to sell those products in their own right – I’m sure there would be a positive margin to be had.

It’s linking the decisions to do, or not to do, either of those things back to the strategy of being a value-led business or a providing a high-quality experience which is the really important part of the process.

Deciding to be digital-first, or AI-first, is like Fortnum & Mason having already decided to sell family packs of own-brand, value-priced cornflakes. Your only job is to put them on a shelf somewhere in the breakfast cereal aisle.

The key here is consistency – is the experience of your customers consistent with the market you are set up to serve or not?

Neither Asda nor Fortnum & Mason are “wrong”. They’re just different. And each behaves consistently within the strategic framework they’ve set as a value brand or a premium brand.

Both of those businesses are very clear about who they serve and what they offer. Customers make their own decisions about which organisation they want to spend their money with.

But neither Asda nor Fortnum & Mason are trying to serve everyone. Or pretending they could, even if they wanted to.

So why should you?

Logical consistency

Many tech solutions fall down due to a lack of understanding of the difference between Asda and Fortnum & Mason.

It is impossible to fire all the people working in Fortnum & Mason, switch everything to AI, and expect still to have a premium business in a few months’ time.

I don’t care what stats someone shows me about how much people love AI (those stats are all lies, by the way) but only a very small part of the reason people buy things from Fortnum & Mason is to get their hands on a specific product.

You could easily find ways to store and ship products using robots in a warehouse just off the M1 much more cheaply than you can sell using lots of human beings pacing the halls of a huge store located amongst some of the world’s most expensive real estate in the heart of London’s West End.

If lower prices and Amazon-like service were the primary driving factors for Fortnum & Mason shoppers, they wouldn’t have any customers.

So the strategic decision to make isn’t “digital-first”. It’s “what are the best decisions we can make which are consistent with our value-brand, or premium-brand, strategy?”

Some of those decisions may well be digital or AI-based. But not all of them. Perhaps only a minority of them.

Upgrading the tills at Fortnum & Mason might well be a sensible digital decision to speed up reporting, or help manage logistics better, for example. But that digital decision is still entirely inconsistent with the ethos of the business. People just want to swipe their credit cards efficiently and breeze off for tea at the Ritz.

Whereas shutting the store, flogging off the prime real estate, and getting everyone to order online would certainly be consistent with a digital-first strategy.

It would, however, probably end with customers leaving in droves for other fancy high-end retailers, because they’re not just buying a product from Fortnum & Mason. They are buying a high-quality end-to-end customer experience, which includes lots of high-touch intervention in an impossibly glamorous setting, none of which you can replicate digitally.

So if you’re tempted to go all-in on AI, or develop a digital-first strategy, just pause for a moment.

That makes no more sense than deciding to have Ford-first car-buying policy or an Asda-first cocktail ingredients policy.

Just by setting the parameters in a way which means you can only consider digital solutions – even when there are demonstrably better non-digital options – means you’ve likely set your business up for failure.

Keep your strategy decisions at the right level – it should be “how do we serve our target customer group more effectively?”, not “out of a range of sub-optimal decisions, which is the least worst?”.

To my continued astonishment, in an AI-first world, far too many businesses which ought to know better are, consciously or otherwise, taking the second of those two options.

Don’t copy their mistakes. “Digital” or “AI” isn’t a strategy. They are, at best, one of the ways in which you might choose to run a small part of your business.

Thinking otherwise means you’ve locked your business into a “tail wagging the dog” situation.

And that’s not smart. No matter how many statistics a tech salesperson has in their “you really can have everything you want, as if by magic” slide deck.