A founder sent me a 42-slide deck last spring. Import volumes by HS code for three years, search volume trends, a competitor map, a bottom-up TAM of €340 million, a top-down cross-check within 8% of it. Genuinely good work. He asked me whether the market was real.
I asked him a different question: how many people in that country, who personally control a budget line, had he spoken to about buying this specific thing at a specific price? The answer was zero. He had spoken to two distributors at a trade fair, which is not the same thing, and to a consultant who had produced most of the deck.
That gap is the most expensive one I see in expansion work. It is not a gap of intelligence or effort. The founder had done more research in six weeks than most companies do in a year. It is a gap of category. He had assembled a very strong case that a market exists, and he was reading it as a case that the market wants him.
Every dataset answers the same question, and it is the wrong one
Customs data tells you that goods crossed a border. Search volume tells you that people typed words. A TAM tells you what the number would be if a hypothetical company captured a hypothetical share. All three describe a world in which you are absent. None of them contain any information about your arrival.
This sounds obvious written down. In practice, it is almost impossible to feel. When you look at 18,000 monthly searches for a product category in a country you have never sold in, your brain converts that number into demand for your product. It is not. It is evidence that a category has attention, most of which is already claimed by whoever has been building relationships there for the last decade.
The problem is worse in the markets that look most attractive. High import volumes usually mean an established supply chain with entrenched intermediaries, price expectations set by incumbents, and buyers who have no unsolved problem. A market with almost no import data might mean nobody wants the product, or it might mean regulation makes it impossible, or it might mean the first mover has a two-year window. The data reads identically in all three cases.
I spent years in banking innovation in Hong Kong building models on lending behaviour, and later hyper-personalisation systems that ran on far richer data than any market entry study will ever have. What that work taught me is a discipline about scope. A model is honest about what it predicts and silent about everything else. Market research is rarely presented with the same honesty. It is presented as a verdict when it is, at best, a filter. Use it to eliminate the eleven countries you should not enter. Do not use it to choose the one you should.
Two markets where the numbers were perfect
In the consumer hardware venture, we built in China and sold into Europe. One Western European market looked, on paper, like the best in our set. Category imports growing steadily, high disposable income, strong affinity for the design language of the product, retail structure that matched our channel model. We had one product then and a plan to build a family around it. Every model said go.
We went. The deals did not close. Not because buyers disliked the product, but because the buying calendar in that channel ran on a rhythm we had not accounted for, listings were decided nine to eleven months ahead, and the two people who actually decided had a relationship with an incumbent supplier that had survived worse products than ours. We were not competing on merit. We were competing on a slot that was not available for another year. No dataset in existence contains that fact. One honest conversation with a category buyer would have contained it in twenty minutes.
The second was in China, during the healthcare work, where we were expanding a private chain’s digital services across the mainland. A group of second and third tier cities showed excellent indicators: patient volumes, income growth, smartphone penetration well above the national average, weak local competition. We built for those cities. Adoption was slow and stayed slow. What we had missed was that the decision to use a private service in those cities sat with a family member rather than the patient, and that the trust pathway ran through employers and local relationships rather than through app stores. The demand was real. The route to it was completely different from the one our numbers implied.
In both cases the data was accurate. The data was also useless for the decision it was being used to make. That is the pattern I want founders to name, because once you name it you stop paying for more of it.
The only validation that has ever held up
I now treat a market as validated when three things are true at the same time. Not one, not two.
The first is ten structured conversations with people who control a budget. Structured means the same eight to ten questions in the same order, so you are comparing answers rather than impressions. Budget control means the person can say yes and cause money to move. A distributor’s sales manager is not that person. A head of innovation is often not that person. You will need roughly forty approaches to land ten of these conversations, and you will conduct most of them badly for the first three. What you are listening for is not enthusiasm, which is free and worthless. You are listening for specifics: what they currently spend on the problem, who they buy from now, what would have to be true for them to switch, when their buying cycle opens, who else signs. If eight of ten cannot answer those questions concretely, you do not have a market, you have a category.
The second is one signed pilot with money attached. The amount matters much less than the fact of it. I would rather have a €4,000 paid pilot than a €200,000 letter of intent, and I say that having held letters of intent that evaporated. Money creates an internal owner on the buyer’s side, someone whose name is now on the decision. That person will tell you things no research participant ever will, because they now have a stake in you being useful. A free pilot creates nothing. It sits in a corner of someone’s quarter and dies quietly.
The third is one local intermediary willing to attach their name to an introduction. This is the test people skip, and in Asia especially it is the one that predicts most. Anyone will take a meeting with you. Very few will spend their own credibility introducing you to someone who matters to them. When a local operator, a lawyer, a former executive, a partner in an adjacent business, says “I will introduce you to X and I will tell them you are worth their time”, they are making a statement about your fit in that market that no report can make. When they hedge, when the introduction keeps not happening, when it arrives as a cc rather than a warm handover, that is a signal. Read it.
Spend the first €20,000 on flights, not on a study
A serious market study will cost you somewhere between €15,000 and €40,000 and take eight to twelve weeks. At the end you will have a document that tells you a market exists. You already suspected that, which is why you commissioned it.
The same €20,000 buys three trips of five to seven days each, a local fixer or junior consultant on the ground for two months to book meetings, translation where you need it, and enough margin to fund a small paid pilot at a discount. Three trips is the right number: the first one is mostly wasted because you are asking the wrong questions, the second is where the pattern appears, the third is where you close something or admit you cannot.
You will come back with less quantitative confidence and far more commercial certainty. You will know the names of six people who matter, the two structural obstacles nobody wrote down anywhere, and whether a single organisation in that country will part with money to work with you. That is a decision-grade input. A TAM is not.
I am not arguing against data. I am arguing about sequence. Use the datasets to narrow twelve possibilities to three. Then close the laptop and go, because the remaining question is not whether the market is big. It is whether it wants you, and that question has only ever been answered by people, in rooms, with budgets.
If the ten conversations are dull, the pilot will not sign, and nobody local will lend you their name, the answer is no. It is a cheap no, delivered in ten weeks rather than after two years of burning a country manager’s salary against a market that was never going to open. That is worth more than any deck I have ever been handed.
I write from twenty years of building businesses between Europe and Asia. If your company is facing this, start a conversation.