How do you test a country or segment before committing headcount?

Summary

Four cheap tests in an order where each can stop the project before the next costs money. Define the buyer precisely enough to name thirty to fifty specific companies rather than a country or a sector. Verify demand exists by finding who supplies them now, what they are asking, and what triggers a change of supplier. Test at least two genuinely different propositions on a slice of the list, with the wording fixed for the duration. Then score on reply quality rather than reply volume — an operational question about lead times is worth more than twenty brochure requests. The decision at the end has four outcomes, and three of them are not hire.

Published

Run it as a sequence of four cheap tests, in an order where each one can kill the project before you spend money on the next. Define the buyer precisely enough to name companies. Verify that demand exists rather than assuming it from market size. Test the message on real buyers. Then judge on the quality of replies rather than the count of them. Only after all four does a headcount decision have anything underneath it.

The discipline that makes this work is being willing to stop at any stage. A test you have already decided the answer to is not a test, it is a budget approval with extra steps.

Step one: who exactly is the buyer?

Not a country, not a sector, not a company size band. A named list. If you cannot write down thirty to fifty specific organisations that should buy this, you do not yet have a market — you have a category.

The list forces precision that a market-size figure hides. A total addressable market of several hundred million tells you nothing about whether forty reachable companies exist with a live reason to change supplier this year. Building the list also surfaces the first real finding, usually within a day: whether the buying population in that market is structured the way it is at home, or whether procurement is centralised somewhere you were not looking, or the specification is written by an engineering consultancy rather than the manufacturer.

Step two: does demand actually exist?

Demand means somebody currently has this problem and is currently spending money on it, not that they logically should. Three checks, cheapest first:

  1. Who supplies them now? If the answer is nobody, be suspicious rather than excited. An unserved market is occasionally an opportunity and much more often a market that does not buy.
  2. What are they asking? Look at where the questions surface — industry forums, tender portals, trade association material, and what an AI assistant returns when you ask a buyer-shaped question in that market's language. If the assistants name three suppliers and none of them are you, that is both a finding and a baseline.
  3. What triggers a change of supplier here? Regulation, a plant expansion, an end-of-life component, a price shock. If you cannot name the trigger, your pipeline will consist of people who found you interesting.

Step three: test the message, not the product

The product usually works. What usually fails is the sentence describing why it matters here, because it was written for a different market's priorities.

The test is direct outreach with a specific claim, sent to a slice of the named list — enough for a pattern to emerge, few enough that you have not burned the market if the message is wrong. Run at least two genuinely different propositions rather than two rewordings of one: for instance a cost-reduction framing against a risk-and-compliance framing. In many industrial segments those two reach different people inside the same company, which is itself the finding.

Keep the slices separate and keep the wording fixed for the duration. A message edited mid-test produces results that cannot be compared, which is the same mistake as editing a prompt set between measurement runs.

Step four: measure reply quality, not reply volume

This is where most market tests go wrong, because volume is the easier number to report and the one that answers the wrong question.

What a reply tells you, by type
Reply What it actually means
"Send me a brochure" Politeness. Counts as a reply, not as evidence.
"We already work with X and are happy" Useful. Confirms demand exists and identifies the real competitor.
"Not us — but the plant in Ostrava buys this" Very useful. Your buyer definition was wrong in a correctable way.
"What is your lead time to Rotterdam?" The strongest signal available: an operational question implies a real evaluation.
"We need certification Y" A gating requirement you did not know about. Possibly the whole answer.
Silence Ambiguous on its own. Only meaningful if two different propositions both produce it.

Twenty replies of the first kind are worth less than two of the fourth. A test scored on volume will recommend hiring after a wave of brochure requests, which is how a good market and a polite one get confused.

What about the visibility layer?

Run it in parallel, because it has the longest lead time of anything in the project. In a market with no history of you, buyers verify before they respond — and Yext's October 2025 analysis of 6.8 million citations found 42% came from listings and profiles, which in a new market mostly do not exist yet.

The practical consequence is that a strong message sent to the right buyer can still fail because the recipient looked you up, found nothing, and quietly moved on. That failure looks identical to a bad message in your reply data. Getting the basic entity footprint in place before the outreach starts removes a confound as well as a barrier.

What does the decision look like at the end?

Four options, and three of them are not "hire":

  • Hire — demand is real, the message works, the servicing economics hold, and the constraint is now coverage.
  • Partner — demand is real but the route to market needs someone who already has the relationships or the local approvals.
  • Persist — signals are genuine but thin; keep testing at low cost rather than committing to fixed cost.
  • Stop — the trigger for change does not exist here, or the servicing cost eats the margin. Finding this out in eight weeks rather than fourteen months is the single largest return the exercise offers.

What to do with your own candidate market

Take the market you are currently arguing about internally and try to write the named list of thirty companies. If you cannot get past twelve without inventing candidates, you have your first result already, and it cost you an afternoon.

If the list comes easily, the next step is testing the message rather than approving a hire. That sequence is what entering a new market covers.

Talk through a market entry

Frequently asked questions

What is the first step in testing a new market?

Writing a named list of thirty to fifty specific organisations that should buy this. If you cannot get there without inventing candidates, you have a category rather than a market, and that is a result in itself. The list also surfaces structural findings quickly — whether procurement is centralised elsewhere, or the specification is written by a third party.

How do I know whether demand actually exists?

Three checks. Who supplies them now — if the answer is nobody, be suspicious rather than excited. What are they asking, including what an AI assistant returns for a buyer-shaped question in that market's language. And what triggers a change of supplier there: regulation, a plant expansion, an end-of-life component, a price shock. If you cannot name the trigger, your pipeline will be made of people who found you interesting.

Should I measure replies or reply quality?

Quality. Twenty requests for a brochure are politeness; two questions about lead times to a specific port are evidence of a real evaluation. A referral to a different site corrects your buyer definition, and a certification requirement may be the entire answer. A test scored on volume will recommend hiring after a wave of polite replies.

What outcomes should a market test allow for?

Four: hire, when demand is real and the constraint is coverage; partner, when the route to market needs existing relationships or approvals; persist, when signals are genuine but thin; and stop, when the trigger for change does not exist or servicing costs eat the margin. Learning the last one in eight weeks rather than fourteen months is the largest return the exercise offers.

Find out what the engines currently say about you

Send me your company name and website. I run a set of buyer questions across ChatGPT, Perplexity, Gemini, Claude and Google's AI Overviews, and send back a recorded walkthrough of what came out: where you appeared, where you did not, who was named instead, and the two or three structural reasons why. No charge, no obligation, and you keep the findings whether or not you decide to hire me.

I run these myself, so there is a queue. Expect a few working days rather than an instant report.