Research

Verified against ChatGPT · 2026-08-12

Synthesize customer interviews into jobs-to-be-done findings without averaging away the outliers

Turns a batch of customer interview notes into a jobs-to-be-done synthesis that keeps a genuinely different customer segment visible instead of blending it into a single averaged persona.

ChatGPT (GPT-5.1)3 fillable variables

The prompt

Ready to copy — highlighted parts are example details you can swap.

I'm giving you raw notes or transcripts from customer interviews. Synthesize them into a jobs-to-be-done analysis — but do not collapse everyone into a single average customer if the notes actually show meaningfully different segments underneath.

INTERVIEW NOTES OR TRANSCRIPTS
Notes from 8 customer interviews about why they signed up for a budgeting app, mostly focused on debt payoff but two mentioning saving for a specific goal instead.

PRODUCT OR DECISION THIS INFORMS
Deciding whether our next feature investment should be a debt-payoff planner or a savings-goal tracker.

NUMBER OF INTERVIEWS AND ANY KNOWN SEGMENT SPLIT
8 interviews total, all existing paying customers, no prior segmentation done.

Before synthesizing, check whether the interviews actually describe one coherent job-to-be-done or multiple distinct ones — if two or more interviewees describe wanting the product for clearly different underlying reasons, do not merge those into one blended "customers want X" statement, because the blended version will describe no real customer accurately. Where a genuine minority pattern shows up (even from just one or two interviews) that contradicts or sits outside the majority pattern, name it as a distinct finding rather than treating it as noise to average out — a minority pattern in qualitative research this small a sample is often a real signal, not statistical noise, precisely because the sample is too small for anything to average out cleanly in the first place. For the majority pattern(s), state the job to be done in the classic structure: when [situation], I want to [motivation], so I can [expected outcome] — grounded in actual quotes or close paraphrases from the notes, not a generic restatement.

WHAT NOT TO DO
Do not present findings as percentages ("60% of customers said...") from a small qualitative sample — that language implies statistical rigor this kind of research doesn't have. Do not silently smooth a contradictory interview into agreement with the majority to make the write-up cleaner.

OUTPUT FORMAT
1. The primary job-to-be-done statement(s), with supporting quotes.
2. Any distinct minority pattern found, named explicitly as such, not folded into the majority.
3. One line on what Deciding whether our next feature investment should be a debt-payoff planner or a savings-goal tracker. this most directly informs, and what it doesn't yet tell you.

Customize

Optional — swap in your own details for the highlighted parts above.

Why this works

Synthesizing qualitative interviews into a single narrative is a compression task, and the model's default compression strategy is to find the central tendency and state that as the finding — which works fine when the underlying interviews genuinely cluster around one job-to-be-done, but actively destroys information when they don't, because a blended "customers want debt payoff and also savings tracking" statement ends up describing neither group's actual motivation precisely. Explicitly instructing the model to check for multiple distinct jobs before synthesizing, rather than synthesizing first, changes the order of operations in a way that matters: it has to hold the segments apart mentally before it's allowed to average anything, rather than averaging by default and only noticing the split if asked to check afterward. The instruction to preserve a minority pattern from even one or two interviews addresses a specific statistical misunderstanding that creeps into small-sample qualitative synthesis — with 8 interviews, "only 2 people said this" is not a weak signal to be smoothed toward the majority the way it would be in a 500-person survey, because there's no law of large numbers operating at n=8 to justify treating a minority view as noise. Forbidding percentage language on a small qualitative sample matters because stating "25% of customers" about 2 out of 8 people borrows the rhetorical authority of quantitative research for a sample size where that framing is actively misleading — someone skimming the report would treat that percentage as more robust than the underlying two conversations actually support.

What you get back

Primary job-to-be-done (6 of 8 interviews): When I'm carrying credit card debt and feel like I'm not making progress, I want a clear payoff plan I can actually follow, so I can see a real end date instead of an open-ended balance. Distinct minority pattern (2 interviews): these two users weren't in debt at all — they wanted the app purely to save toward a specific goal (a house down payment), a meaningfully different job the debt-payoff framing doesn't serve.

Verified against

ChatGPT GPT-5.1 · 2026-08-12

Changelog

  • 2026-08-12 Initial publish, verified against ChatGPT GPT-5.1.

Need this built into your business?

If a prompt isn't enough — what Scult builds, built and maintained for you — that's Scult's day job.

EXPLORE WHAT SCULT BUILDS
All Research prompts

Check your AI visibility

One URL in, a 0–100 score and the exact fixes out.

RUN THE CHECK

Browse all the tools

15 tools across six categories
13 of them never send your data anywhere

Free · No signup · No trial clock

SEE THE DIRECTORY