The question isn’t which AI tool you should be using. It’s what are you trying to achieve.
Barely a week goes by without a new AI tool landing in your inbox, your social feed, or your group chat. Someone you respect — a colleague, a podcast host, a LinkedIn contact you've never actually met — is raving about how it's transformed their workflow. "Just ask ChatGPT." "Have you tried this one?" "I saved three hours on Thursday."
And look, the excitement is completely understandable. Some of these tools genuinely are remarkable. The things they can do, the time they can save, the sheer pace of what's arriving — it's a lot to take in, and ignoring it isn't an option. But here's what doesn't get said often enough: almost every AI tool being pitched to businesses right now is being sold on exactly the same promise. Efficiency. Do more with less. Cut the overhead. Automate the repetitive stuff. Reduce headcount.
And that's where it gets interesting — because efficiency is only one measure of value. And it's not always the most important one.
Why efficiency dominates the conversation
There's a reason for this, and Rory Sutherland (Vice Chairman, Ogilvy UK) puts it rather brilliantly. He points out that there are two fundamentally different schools of thought about the purpose of a business.
#1. Create new sources of value (from the Austrian School of Economics)
What a customer feels, perceives, and experiences is every bit as real and commercially important as what gets manufactured or delivered. Under this view, marketing isn't a cost centre. It's a value creator. The business exists not just to produce things efficiently, but to continually explore and discover new sources of value — almost like a living, breathing organism adapting to its environment.
#2. Deliver value for the lowest cost (from the Chicago School of Economics)
This is the one that dominates most publicly listed companies today. It assumes customers already know what they want, can price it precisely, and therefore the only sensible goal of a business is to deliver that thing at the lowest possible cost.
The problem is a measurement one. The second model fits beautifully on a spreadsheet. It's measurable, quantifiable, and very easy to report to a board. The first model deals in abstract things like trust, perception, and human connection — which are real, but notoriously difficult to put a number on.
So when an AI company needs to recoup the extraordinary sums poured into building these tools, what's the easiest sell? Cost reduction. Headcount savings. It's the pitch that lands in a boardroom, and it's the pitch that's been working. That's not cynicism — it's just how the numbers work. But it's only ever half the story.
When efficiency quietly destroys value - examples
A consultant walks into a hotel and asks what they pay the doorman. They define his/her role as opening the door. They replace him/her with an automatic door-opening mechanism, claim the cost savings, and walk away — not remotely accountable for what follows.
What they fail to account for is all the other things the doorman did. S/he recognised and welcomed regular guests by name. S/he lent the hotel a quiet sense of security and status. S/he was, for many guests, their first and last impression of the place.
Five years later, the room rates have fallen, the most loyal guests have drifted away, and someone else is left to clean up the mess. The consultant who captured the saving is long gone.
Sound familiar?
The supermarket self-checkout is perhaps the most universally recognised example. In the 1980s self-service checkouts were rolled out in USA as a convenient option for customers with a small basket and in a bit of a hurry. In time they became the default — because once the finance team noticed that customers doing their own scanning was cheaper than employing a till operator, the option became more or less an obligation.
What followed was a rather spectacular rise in shoplifting — including, as Sutherland notes with some amusement, customers scanning avocados and claiming they were onions. Beyond the theft, doing a full family shop through a self-checkout is genuinely unpleasant (especially at Christmas time!!!). The efficiencies were captured on a spreadsheet. The customer frustration, the lost loyalty, and the inventory losses didn't appear on the same one.
The human element is almost always undervalued
There's a reason this pattern keeps repeating, and it's not because business owners are dim! It's because the things that are easiest to measure get treated as the most important — even when vastly more valuable things are sitting right next to them, just harder to quantify.
Royal Mail in the UK is another good example. They spent a fortune improving operational efficiency and on-time delivery, and it had no measurable effect on brand perception whatsoever. When they looked more closely, they found there was no correlation at all between the reliability of the service in an area and how much people liked the brand. What actually determined whether someone felt warmly toward Royal Mail was whether they liked their postie. That's it. A friendly, human presence — impossible to put in a spreadsheet, and worth more to the business than any logistics improvement they'd made.
So what does this mean for your business?
The businesses most likely to see through the efficiency-only pitch, interestingly, tend to be family-owned, smaller or operating in close-knit communities . Not because they're more virtuous, but because they’ve been part of the community for generations. They've been on the till. They know what those customers value — and it's not always what you'd expect. They understand the worth of long-term loyalty and what it does to the bottom line, while still keeping a sensible eye on costs.
None of this is an argument against AI tools or new technology. It's an argument for being deliberate about how and why you use them. The question for every business owner right now isn't "which AI tools should we be using?" It's "what are we actually trying to achieve — and does this tool help us get there, or does it just make the spreadsheet look better in the short term?"
A framework for making better decisions
Before you implement anything — AI or otherwise — it's worth working through five honest questions. I've put these into a checklist you can keep to hand [download at the bottom of the blog], but the thinking behind each is worth spelling out.
#1. Understand what you're actually replacing — not what the job description says, but what the role or process genuinely does, including all the informal, invisible things it contributes. The doorman didn't just open the door.
#2. Account for the value you can't easily measure. Is there a human element to this interaction that customers value, even if they've never articulated it? Would removing it affect retention or word-of-mouth over the next year or two? If you're not certain, that uncertainty is itself worth investigating before you proceed.
#3. Make sure the savings are real — and complete. Have you factored in what goes wrong when a function is removed or degraded? And critically: is the person claiming the cost saving also accountable for the outcomes over the following two or three years? If not, you may be looking at a Doorman Fallacy in the making.
#4. Ask whether different might be better than better. As more businesses automate, keeping genuine human touchpoints could become a real point of difference — not a cost, but a competitive advantage. Are you offering new tools as an option first, or making them the default from day one?
#5. Ask whether you're using AI to improve experience or simply to reduce cost. The most thoughtful applications free up your people to spend more time being genuinely useful to customers — not less. In twelve months' time, will your customers feel better served, or just more processed?
The businesses that will come out ahead aren't the ones who automate fastest. They're the ones honest enough to ask what they'd be giving up — and clear-eyed enough to follow that question all the way through.
As Sutherland puts it, the real value of thinking like a marketer is looking at your business from your customer's standpoint, not just the cost model. That's not a soft skill. That's a survival skill.
The tools are genuinely exciting. Just make sure they're working for your customers — not only your spreadsheet.
