Why positioning built from customer interviews beats an AI-generated guess

Positioning from customer interviews works because it produces evidence rather than plausibility: the words your buyers use, the event that started their search, and the options they actually weighed. A generated summary can draft a hypothesis in minutes, and it can only reflect what has already been published about your category. This piece covers how to run the interviews.

Most positioning work at a 40 to 200 person company gets done in a room with a whiteboard, or in a document drafted from what the market already says. Both produce something readable by Friday. Neither tells you why your last 10 customers chose you, which is the only input that makes the positioning defensible when someone challenges it.

Key takeaways

• A generated draft is useful for structure and hypotheses, and it cannot know your buyers.
• Sample design matters more than sample size. 15 to 20 conversations across 4 groups is enough.
• Ask about the last real purchase rather than preferences in general.
• Four outputs come only from real conversations, and all 4 change the copy you write.
• Use the draft as the hypothesis and the interviews as the test.

What a generated draft actually gives you

Worth being fair about this, because the answer is not nothing.

What it does well:

• A clean structure for a positioning document, in minutes rather than days.
• A list of plausible buyer types and objections to take into the conversations.
• A summary of how your category already talks about itself, which is genuinely useful when you want to sound different from it.

Where it stops:

• It works from published material, so it reproduces the category consensus. If every competitor claims speed and reliability, that is what comes back.
• It cannot know the trigger that made your buyer start looking, because that lives inside their operations rather than in published material.
• It has no access to your lost deals, which is where the most useful information usually sits.
• It reads as familiar. Buyers skim past copy that sounds like every other option they have already dismissed.

So the draft is a starting hypothesis. The interviews are what turn it into a claim you can defend in a board meeting.

Who to interview: sample design over sample size

15 to 20 conversations is enough at this company size. What matters is the spread. Take roughly 4 to 5 people from each of these groups:

• Recent wins, closed in the last 6 months. Their memory of the decision is still accurate. Older customers remember a company that no longer exists.
• Expansion accounts. Customers who bought more after the first purchase. They tell you what value looks like once the product is in use.
• Lost deals. The most uncomfortable and the most useful. They tell you which competitor you actually lose to and on what criterion.
• Churned customers. They show you which buyer you should stop chasing, which shapes the positioning as much as knowing who to pursue.

Skip your happiest reference customers if they are the only people you speak to. A sample of advocates confirms whatever you already believe.

What to ask, and what to avoid

The single most important discipline: ask about one specific past purchase rather than preferences in general. People are unreliable about hypotheticals and precise about what they actually did.

Ask:

• What was happening in the business the week you started looking?
• What had you tried before that, and why did it stop working?
• Who else did you look at, and how did you narrow it down?
• What nearly stopped you buying?
• How did you describe this internally to get budget approved?

Avoid:

• Would you recommend us. The answer is yes and it teaches you nothing.
• What features would you like. That belongs in a product conversation.
• Any question that names your own claim, which invites polite agreement.

Two practical points. Have someone senior run the conversations, because follow-up questions are where the value appears. And record and transcribe them, since the exact phrasing is the output.

The 4 things only real conversations produce

These are the outputs that change the copy on your website, and none of them can be inferred from published material.

• 1. Verbatim language. The words buyers use for the problem, which are almost never the words your industry uses. This becomes your headline and your search terms.
• 2. The trigger event. The specific thing that started the search: a failed audit, a resignation, a new contract, a system that finally broke. Aim your content at the trigger and you reach people at the moment they are looking.
• 3. The real alternative set. Usually not the competitors you track. It is often a spreadsheet, an internal hire, or doing nothing for another year.
• 4. The order of the decision criteria. Everyone says price matters. Interviews tell you whether it ranked first, third, or last, and what actually decided it.

How to use both together

The two methods work in sequence rather than in competition.

• Before the interviews: draft the hypothesis. Who you think the priority buyer is, what you think the trigger is, which alternatives you expect. Write it down so you can be proved wrong.
• During: test that hypothesis against what people say. Note every place the language differs from your draft. Those gaps are the finding.
• After: lock the answer on the evidence, and keep the transcripts. When someone challenges the positioning 6 months later, quoting 3 customers ends the argument faster than any deck.

Positioning from customer interviews takes about 2 weeks at this company size, which keeps it well inside the first month. What to do with the answer once you have it sits in the portfolio company positioning play.

What the interviews revealed in one Nordic company

Take a newly acquired Nordic software company we work with: around 60 people, 600+ customers, two marketers, no CMO. The internal assumption was that customers bought on ease of use, which is what the website said and what any category summary would have suggested.

The conversations said something narrower:

• The buyers the company was now winning were larger, and they described the problem in operational terms the website never used.
• The trigger was rarely a search for better software. It was usually an internal event that made the old way untenable.
• The real alternative was building something internally rather than a named competitor.
• Ease of use mattered, and it ranked below confidence that the switch would not disrupt anything.

None of those 4 findings would have come out of a desk exercise. All 4 changed the copy.

FAQ

Draft the guess, then go and check it

Write the hypothesis quickly, then spend 2 weeks testing it against 15 to 20 real conversations across wins, expansions, losses, and churn. Keep the transcripts. What you get back is a positioning claim with evidence attached, in language your buyers recognise. No random acts of marketing.

If you want a lighter first look at where your current message has drifted from the buyers you now win, the Positioning Teardown sits alongside the value-creation programme for portfolio companies.

When you are ready to run the interviews for the company you just bought, book a strategy session. You will leave with a bird's-eye plan, whether or not we work together.

Book a strategy session →