In 1970, a young economist called George Akerlof wrote a paper about used cars.
He sent it to three journals. Two of them said the subject was trivial. The third went even further and asserted that if Akerlof was right, economics itself would need rethinking…In 1970, nobody wanted to hear it.
In 2001 he won the Nobel Prize.
The argument was disarmingly simple. If buyers can’t tell good from bad, they pay somewhere in the middle. The good sellers decide that price isn’t worth it and leave.
What remain are lemons - dud cars in this instance.
Akerlof’s point wasn’t really about cars. It was about what happens to any market when one side knows something the other doesn’t.
I spent twenty years working on exactly this problem without knowing it had a name.
Brand strategy - the discipline I eventually built a career around - exists because of information asymmetry. Not because anyone in the branding world has ever called it that. But the entire point of a brand is to be a signal. A way for buyers to sort quality from noise when they can’t open the bonnet before they sign on the dotted line or hand over cash.
The less visible the quality, the harder the brand has to work. Luxury goods. Professional services. B2B procurement. Every positioning exercise I’ve ever run was, underneath, an attempt to make the invisible - visible. To give the buyer something to trust before they’d experienced the product...
Which is why the current AI market is doing my head in.
Most businesses start sensibly. Buy a team plan. Give people access. See what happens. And things do happen - drafts get written faster, research gets easier, someone in finance builds something clever with a spreadsheet. But at some point, the question changes from ‘should we use AI?‘ to ‘how do we make this actually work for us?‘
That’s when you walk onto the “car lot”. Not choosing between the big AI providers - that part’s straightforward. Choosing between the hundreds of consultancies, platforms, and integration partners who all promise to be the answer.
Every AI tool looks amazing. Every sales site offers incredible AI tools promising enhanced productivity, research, and content generation…
Every platform has a transformation story to “streamline client workflows, automate analysis, and generate insights“. If you’re a finance partner, a managing director, or a procurement lead trying to work out where to put a serious chunk of budget, you are standing in Akerlof’s used car lot. You just can’t tell which ones are lemons.
And the market is behaving exactly as he predicted. Buyers who can’t evaluate quality do one of two things: Overpay for something that underdelivers, or stall and buy nothing. Both are rational. Both are happening everywhere. And the vendors who genuinely solve problems are being pitched alongside tools that will hallucinate your compliance documentation and call it insight. How would we know?
This isn’t a technology problem. It’s a market problem. Akerlof showed us the only way out fifty years ago: credible signals that the seller can’t fake. Warranties. Guarantees. Demonstrated value. Anything that shifts the risk from the buyer to the seller.
In the AI market, the equivalent should be proof. Proof in your context, with your problems, against your data.
We’ve just rebuilt our AI readiness calculator to be more honest and more useful. Because the fastest way to improve a lemon market is to give the buyers better information. Rebalance the asymmetry.
Akerlof’s paper was about used cars and it described everything. A market where bad drives out good isn’t a failure of the sellers.
It’s the absence of any signal a buyer can trust.
The Contour · Newsletter
Want this in your inbox?
The Contour goes out roughly every ten days. Published here first, then on LinkedIn and Substack.