Silvia Adlesic told a story this week that looks like fashion and is actually strategy.

A brand called Arakii, with a Scandinavian identity, a Japanese name, and a website that listed its production countries in this order: Portugal, Italy, Turkey, China. A Swedish blogger sat down and went through all 233 products on the site. The result: 88% of the clothing was made in China. A 90% polyester dress, made in China, carrying a price tag of 19,000 kronor. The brand had been growing like a rocket, from just over one million kronor in 2022 to 57 million in 2025, with a target of 100 million for this year.

The willingness to pay was there, crystal clear. Thousands of customers looked at that price and said yes.

Then they looked at the label.

The thesis Adlesic draws from it, leaning on a Nobel laureate, is the one I want you to take, because it isn’t about clothes: willingness to pay measures what the customer believes, not what you built. Change the information and the number changes with it. And in the AI era, that belief gets corrected faster than ever.

What the customer pays measures what they believe, not what you have

Willingness to pay (WTP) is one of the most used concepts in commercial strategy, and it’s almost always misunderstood. It shows up in surveys, in studies, in pricing decks as if it were a fixed number, a stable ceiling the customer carries in.

It isn’t. It’s a function of what the buyer believes they’re buying. George Akerlof proved it in 1970 with his famous market for lemons: when the seller knows more than the buyer about real quality, the price floats free of value. The buyer pays based on belief, and when that belief is corrected, willingness to pay collapses. Akerlof was writing about used cars. The mechanism is identical for a polyester dress sold under a Scandinavian identity with a Japanese name and a barely implied Mediterranean origin.

The marketing research sharpens it further. When a brand implies a prestigious origin that doesn’t match reality, willingness to pay rises. When the real origin is disclosed, it falls. The premium was never attached to the product. It was attached to the assumption. The order of the countries on that website, Portugal first and China last, wasn’t a formatting accident. It was the pricing architecture.

And to be clear: the problem isn’t China. There’s excellent manufacturing there at every quality tier. The problem is the distance between what the price implies and what the label confirms.

Information asymmetry has an expiration date

Here’s the part that changed, and it’s what turns this from an anecdote into a business risk.

For decades, that distance between what you imply and what you confirm could last years. Full information, the real origin, the composition, the cost structure, traveled slowly. The customer found out late, or never. The belief premium could hold for a long time.

Not anymore. A single data-driven blog post reaches a national press cycle in 48 hours, as it did with Arakii. And AI put a verification tool in everyone’s hands: origin, composition, and claims can all be checked in minutes now. The question stopped being whether full information will reach your customer. It’s when. Adlesic says it in a line worth framing: a pricing strategy that depends on information asymmetry isn’t a strategy, it’s a countdown.

This doesn’t mean charging for a story is wrong. Part of what the customer pays for is experience, identity, aesthetic, the feeling a brand creates, and that’s legitimate pricing territory. The problem begins when the story is used to imply a substance that doesn’t exist: European craft, premium materials, sustainable production. What’s anchored to something verifiable is durable. What’s anchored to atmosphere needs constant maintenance, and AI just raised the cost of that maintenance.

And the same applies to the AI you’re buying

Here I flip the argument, because the same mechanism runs in reverse when you’re the buyer. And almost nobody sees it.

The ROI promises AI vendors make you are, almost always, a willingness to pay built on belief. “A customer burns 100 billion tokens a month.” “8x productivity.” “AI-native.” They’re origin stories, just like Arakii’s “made in Portugal.” They imply a substance, a result, that often doesn’t come with the number to back it. And you pay for the belief, with real budget, until someone shows you the output.

The good news is that the same transparency that will strip Arakii’s price is the one you can use to strip what’s sold to you. This week I wrote the exact other side of this coin: when Anthropic measured itself and published the number, it did the opposite of the vendor who only flexes how much it consumes. Against any AI claim, the question that breaks the asymmetry is the same one that breaks an inflated price: show me what’s verifiable. I wrote separately about the concrete metric you should ask for: cost per shipped feature instead of tokens per month. Don’t believe the implied origin. Ask for the label.

What we do about it at IQ Source

Our business is, in large part, closing that distance between what’s believed and what can be checked. On both sides.

When a company asks us to assess where to put AI, we don’t hand over a pretty story with an inflated ROI number so you’ll sign. AI Maestro produces verifiable findings: a map of the real processes of your operation, not the org-chart version, an AI Opportunity Score process by process, and a Go/No-Go gate decided on what can be checked, not on atmosphere. It’s the opposite of a belief premium. It’s a label you can inspect.

And the same discipline works outward. If your price, or your value proposition, rests on something the customer believes but can’t verify, that’s no longer a price. It is, in Adlesic’s words, a loan against their trust. And loans get called. Before you close the week, ask your commercial team one question: what part of what we charge is anchored to something the customer can verify, and what part is anchored to something that only needs them not to look? That second part is your belief premium, and AI just put an expiration date on it.

Base your AI decisions on what’s verifiable, not on belief