Skip to content
scrutinizr.
Customer decisions

Find out why customers do what they do.

Run surveys and interviews where a neutral AI asks one or two follow-up questions, so you get the reason behind each answer.

Who it’s for

Built for teams about to commit.

Most surveys tell you what people chose. They rarely tell you why, so the team ends up guessing.

Founders

Talk to 30 target customers about the problem before writing code.

Product teams

Ask every cancelling customer what made them leave, and group the reasons.

Agencies

Run the same discovery interview across client projects and compare the results.

Methods

Pick the right test for the question.

Before you build

Problem discovery interview

Ask about the last time the problem happened, what people did and what it cost them.

Problems ranked by how often they happen and how painful they are, with quotes.
New sign-ups stall

Onboarding survey

A short survey sent after sign-up asking what nearly stopped them and what they expected.

Activation blockers grouped by customer segment.
Customers cancel

Churn interview

A cancellation path with branching questions and a neutral follow-up on the main reason.

Ranked reasons for leaving and the ones you can fix.
Ongoing health check

NPS with follow-up

Score from 0 to 10, then one follow-up asking what drove the score.

Score over time and the themes behind promoters and detractors.
How it works

From question to decision.

  1. Step 1

    Pick a template

    Start from one of five interview or survey templates and edit the questions.

  2. Step 2

    Choose follow-ups

    Turn on up to two AI follow-ups per text question. They ask for detail without leading the answer.

  3. Step 3

    Share the link

    Send it to your own customers, embed it in your product, or recruit participants or experts through us.

  4. Step 4

    Read the themes

    Answers are grouped into themes with counts. Click any theme to read the original answers.

  5. Step 5

    Act on it

    Turn a theme into a task, a follow-up test or a shareable report.

What you get

A result you can act on.

Sample output

Why customers cancel: 42 interviews

Top three reasons account for 71% of cancellations.

Could not share with family38%
Too much manual entry21%
Price compared to free apps12%
“My partner could not see the same list without a second account.”
Illustrative example.

Themes are based on the answers you collect. Small samples show patterns, not proof. Every percentage shows the number of people behind it.

Success stories

Decisions made on evidence.

Consumer app

Validated the problem before writing code

Would pay—31%
Problem
A founder wanted to build a family meal planner but wasn’t sure families had the problem.
Evidence
Twenty interviews with AI follow-ups showed the pain was shopping lists, not recipes.
Action
Changed the first version to shared shopping lists.
Result
31% of the next 40 people tested said they would pay, against 9% for the original idea.
B2B software

Found the onboarding step that stalled trials

Task success60%85%
Problem
Trials signed up but rarely reached the first report.
Evidence
A prototype test showed 7 of 20 people stuck on importing data, and follow-up interviews said why.
Action
Added a sample dataset and a manual entry option.
Result
Task success in the retest rose from 60% to 85%.
E-commerce

Found the price people would actually click

Buy clicks9%18%
Problem
A subscription box was about to launch at £12 a month with no evidence for the price.
Evidence
A pricing test showed people randomly assigned £8 or £12; the main objection at £12 was free alternatives.
Action
Launched at £8 with an annual plan for the most engaged buyers.
Result
Buy clicks doubled from 9% to 18% at the lower price.

Illustrative examples showing the method. They are not customer results.

Pricing

Tests and surveys are in every plan, from free.

Recruited participants are pay as you go, and you see the total before you pay.

Questions

Good to know.

Will the AI lead people to an answer?

No. Follow-ups are limited to two per question and are instructed to ask for detail, not to suggest answers. You can review every follow-up that was asked.

Can I use my own customers?

Yes. Share a link or embed the form. If you need participants, add recruited people or experts when you order.

How many responses do I need?

For themes, 15 to 30 interviews usually show the main patterns. For percentages, plan for at least 100 survey responses.

Test it before you launch.

Run your first interview, survey or test free.