Who the customer is and why they tolerate the problem

AI for startups · Lesson 2 / 22

A tolerated problem is worth less than it looks

Most early products die not because the problem is absent. The problem exists, but it is built into the customer's life like weather: unpleasant, familiar, and cheaper to endure than to change behaviour around. Your question is not "does it hurt" but "what have they already done to make it hurt less". People who suffer quietly do not pay. People who assembled a workaround from three spreadsheets and a nine dollar subscription almost always do.

Mapping the current solution

  • What they do today. A sequence of steps, not a general description.
  • What it costs them. Money, hours, nerves, standing with their boss.
  • Who would notice a change. If nobody would, there is no budget.
  • What keeps them there. A contract, a habit, an integration, fear of breaking a process.
  • When patience runs out. The trigger event: an audit, a hire, growth, a lost client.
Segment profile (fill in for one living person):
name and role - company size - who approves the spend
current solution and its price - frequency of pain (daily / monthly / yearly)
what they did in the last 90 days to make it easier
the trigger after which people start looking for a replacement

Frequency beats intensity. A once-a-year pain worth a thousand dollars loses to a weekly pain worth twenty: the second one is remembered, searched for, discussed and paid for on a card without approval. Annual pain is always postponed to next quarter.

And separately: the person who suffers and the person who pays are often different people. The bookkeeper struggles, the director pays, and the IT lead blocks the rollout. Until you can name all three, you have an image, not a segment.

Insight. An existing homemade workaround is the strongest early signal available. It proves the person already spent resources on the problem without your involvement, so you are competing with real behaviour rather than imagined behaviour.
Common mistake. Describing a segment through company demographics: industry, revenue, headcount. That is a mailing list filter, not a segment. Segments are defined by shared behaviour and a shared trigger, otherwise your interviews never add up to one conclusion.
Pro tip. Ask about the last 90 days rather than what usually happens. In 'usually' people describe an idealised version of themselves; in the last 90 days they recall specific incidents, dates and amounts.

Cheat sheet

  • Look for the workaround, not the complaint.
  • Frequency of pain beats its size.
  • Sufferer, payer and blocker are three different people.
  • A segment without a trigger does not buy.
1. Which early signal is the strongest?
2. Why does frequency of pain matter more than its size?
3. What makes a segment description useless?
Task — checked by AI

Describe your segment through behaviour rather than demographics. Take one real person you know by name and fill in the profile: role, who approves the spend, the current solution and its price, the frequency of the pain, what they did in the last 90 days to make it easier, and the trigger that would push them to look for a replacement. Separately name the sufferer, the payer and the person who could block a rollout. Finish by stating whether they have a homemade workaround, and if not, what that means for your hypothesis.

← Back

🔒 Answer the question correctly to move on to the next lesson.

Who the customer is and why they tolerate the problem — AI for startups — Skilvy