Part two · Everything that touches them · Chapter 8

WHAT THE DATA CAN’T SEE

Data shows where customers stop. It rarely shows why. For that you need their words, a small number of them, and an honest label on what you’re guessing.

The cohort table can tell you that 1,900 customers a month never show a first-use signal. It can’t tell you whether the product was confusing, was a gift, or was used without triggering your signal. Those are different fixes, and the only way to tell them apart is to ask.

Four sources, and what each misses

SourceWhat it showsBlind spot
Support ticketsProblems in the customer’s own words, tagged by stageOnly customers who bothered to write; most who leave don’t
Return reasonsWhat didn’t work, by productVague codes; nothing from customers who kept an item they didn’t like
Session recordingsWhere people hesitate, scroll back or quit on the siteOnly on-site stages; no view of the cupboard, the bathroom or the dog
InterviewsThe circumstance, the goal and what got in the waySmall numbers; people explain their past behavior tidily

Use the first three to form a view, and interviews to test it. For reading tickets and return reasons as a defect log, see The Return Trip. For session recordings and on-site research, see The Honest Test.

How many conversations

Fewer than most teams think. Abbie Griffin and John Hauser, studying how firms identify customer needs, found that “interviewing 20 customers identifies over 90% of the needs provided by 30 customers”, and hypothesized that 20 to 30 interviews find 90% to 95% of needs. Two one-on-one interviews were about as productive as one focus group Published. For website problems, Jakob Nielsen’s rule of thumb is that five users surface about 85% of usability issues, with more needed for distinct groups Reported.

For a journey stage, that suggests eight to ten interviews with customers who stalled and five or so with customers who passed, per round. The comparison is what makes it useful: the difference between the two groups is the fix.

Who to call, and what to ask

The script is in Appendix B. Offer a small thank-you, and be clear it doesn’t depend on what they say.

Label every part of the map

Mark each stage, and each claim about why customers stall there, as measured (from data), heard (from customers in their words) or inferred (believed, not yet checked). The Nielsen Norman Group’s advice on hypothesis maps is to validate them with research before using them for decisions Reported. The label keeps that visible. An inferred claim can still guide a test; it shouldn’t drive a budget.

Do this

If this is your store

See where your first-time buyers stall.

The free Klaviyo audit reads your own account and scores it. Start with your store’s address. A read-only key gets the full report.

Get the free audit →

Written by Andrew Lauchner, a growth and retention operator for consumer brands. The paid work is one ninety-day Sprint.

This is one chapter of The Journey Map, which is free and readable in full on a single page with no form in front of it.