Customer Research

Customer research

Positioning fails on bad inputs more often than bad reasoning. This skill is about getting inputs you can trust, from the user in front of you and from what customers have already written down in public.

Two sources, used differently. The user knows the deals, the losses, and the objections, so ask them. Customers write the language, so go read it.

Interviewing the user

You are talking to someone who knows the product far better than you. Your job is to extract specifics, not to demonstrate a framework.

• Ask about the last time, not in general. "Who was the most recent customer to buy, and what made them" beats "who is your ideal customer", because the first has an answer and the second invites a persona.
• Ask about behavior over opinion. What they did tells you more than what they think buyers want.
• Follow the loss. The deal that didn't close, and who it went to, is the fastest route to the real competitive set.
• Push once on an abstraction. "Faster" gets "faster than what, by how much, measured how". Once is enough; twice is an interrogation.
• Ask what surprised them. The use case customers found that the team didn't plan is frequently the actual positioning.
• Batch three questions at a time. Twelve questions get skimmed and the answers get thinner as they go.

references/questions.md has the question sets by purpose.

Reading what customers already wrote

Reviews, forum threads, and support conversations are the cheapest research available and the only place you'll find the words customers actually use. Read them for language rather than for sentiment.

What to pull out:

• The verb they use for the job. "Chasing exports" is worth more than "data integration workflows".
• What they compare the product to, unprompted. That's the real alternative.
• The moment they decided. Reviews often name the trigger, and triggers make better campaign material than features.
• What they complain about in positive reviews. That's the honest limitation, and it belongs in objection handling.
• Which benefit they mention first. Their ordering is better evidence than yours.

Quote it verbatim, with a source. A paraphrase loses the exact phrasing, which was the point.

Telling a real finding from a polite one

Most research goes wrong here rather than in collection.

• One customer saying something is an anecdote worth quoting, not a pattern. Two independent sources saying it unprompted is a pattern.
• Compliments are noise. "This looks great" predicts nothing; "I stopped using the spreadsheet the same week" is a finding.
• Anything hypothetical is unreliable. What someone says they would pay, or would use, is not evidence.
• Beware the answer shaped by your question. If you named the benefit and they agreed, you learned nothing.
• Separate what you heard from what it means. Record the quote, then your interpretation, so the next person can disagree with the second without losing the first.

What to hand back

Findings with their sources, sorted into what's confirmed, what's a single data point, and what's still an open question. Include the language list, since that's the part messaging will use directly. When the research contradicts the positioning that exists, say so plainly rather than reconciling it quietly, and say what would settle it.