Skip to content
AI Ping
한국어English

Product research: from reviews to a test

Separate observations from interpretations, then design an observable prototype task.

A learning flow from source material to the next decision
Illustration: AI Ping

Product research2 guidesAbout 30 min reading, plus practiceBeginner → Intermediate · practical

Start with the first guide →

Before you start

  • Practise requesting, revising and checking in AI First Steps.
  • Use the supplied fictional notes rather than real work records.

What you will be able to do

  • Separate original review statements from your interpretations.
  • Group evidence while retaining contrary cases and unknowns.
  • Write a prototype task that does not reveal its solution.
  • Explain what a small sample can and cannot establish.

Read → try it → check against the source

Take one lesson at a time. Examples help you assess your own work; you do not need to copy their wording. Drawings are explanatory illustrations, not actual product screenshots.

Course

  1. Research basics: read reviews as evidence

    Use three illustrated review cards to understand observation, interpretation, themes and samples.

    BeginnerExplainerAbout 15 min reading

  2. Research intermediate: plan a prototype test

    Turn a hypothesis into an observable task and distinguish independent success from assisted completion.

    Intermediate · practicalExplainerAbout 15 min reading

Put it into practice

Group the three fictional reviews with source IDs, then write a research question, task and observation rules.

How to check your work

  • Statements and interpretations are distinct.
  • Contrary cases and unknowns remain.
  • The task does not reveal its solution.
  • Small records are not population estimates.

04 · Advanced project

Research project · Build an evidence map that keeps contrary cases

Group fictional usability notes by source ID, derive interpretations and design the next study. Distinguish duplicate observations and assisted completion instead of selecting only convenient evidence.

All source material, figures and drafts below are editor-created practice data, not customer cases or recorded AI outputs.

First: evaluation, review and measurement lessons →

Source material

R1
Participant A found the venue unaided, then said the text was small. These are two observations about the same participant.
R2
Participant B found the venue after the moderator pointed out the link. B described the task as easy, but completion was assisted.
R3
Participant C opened an old notice and gave the wrong venue. Participant D found the current venue unaided.

Step-by-step work

  1. 1. Count participants and preserve observations

    Keep observation and participant units separate: there are four people, A–D. Do not count quotes as participants.

    Produce: A table of participant, behaviour, statement and assistance

  2. 2. Connect claims and contrary evidence

    Examine how R1 complicates 'small text prevents venue finding'. Group behaviour separately from stated preferences.

    Produce: An evidence map with interpretation, contrary cases and unknowns

  3. 3. Design the next observation

    Choose whether to investigate freshness recognition or link discovery. Do not reveal the target location in the task.

    Produce: A research question, neutral task and assistance-recording rule

  4. 4. Bound the conclusion

    Record that A and D completed unaided in these notes, without treating two out of four as a population estimate.

    Produce: A memo separating findings, limitations and next questions

Find the problems in this draft

Most of the five comments are positive, so there is no usability problem.
Read the review

This double-counts observations and drops B's assistance and C's wrong venue. Judge task behaviour against completion criteria, not the positivity of comments.

Open after your first draft · New condition

New evidence: D has organised the group before. Does this establish ease for first-time visitors?

Check the revision approach

Retain D's success but do not treat it as evidence about a first-time visitor. Add experience to the record and recruit new visitors for a fresh task rather than selectively hiding observations.

Ready when you can explain every item

Attach your deliverable or source ID to each check. Return to the relevant step if you cannot support an item. This is self-review, not automated grading or certification.

  • Participant count is distinct from observation count.
  • Unaided, assisted and failed attempts are separate.
  • Claims retain contrary evidence and unknowns.
  • New evidence changes the scope and next study.

Where to go next

  • Two results compared against common evidence and criteria

    Going furtherEveryone

    Advanced practice: evaluate, review and improve

    After drafting work in your role, learn to evaluate it, design review points and measure a pilot. Complete three lessons without code, then apply them in a role project.

    What you will be able to do

    • Compare results consistently and verify changes on fresh cases.
    • Design drafting, review, approval and failure handling.
    • Choose adoption scope from quality and total work, including review and repair.

    In this lesson

    1. Evaluate AI work beyond first impressions
    2. Design an AI workflow with clear review points
    3. Measure an AI pilot and decide when to stop

    3 guidesAbout 54 min reading, plus practiceAdvanced

  • A learning flow from source material to the next decision

    Courses by rolePlanning

    Planning: from problem to priority

    Beginner and intermediate lessons on turning ideas into evidence and small tests.

    What you will be able to do

    • Separate supported facts from untested hypotheses.
    • Compare ideas and explain which one to test first.
    • Design a small test and decide when to revisit the choice.

    In this lesson

    1. Planning basics: frame a problem with AI
    2. Planning intermediate: prioritise and test ideas

    2 guidesAbout 30 min reading, plus practiceBeginner → Intermediate · practical

  • Two copy drafts compared beside a small results sheet

    Courses by roleMarketers

    For marketers: from copy to data

    Draft ads and content faster, then move on to reading your campaign results.

    What you will be able to do

    • Separate customer hypotheses from research findings.
    • Create distinct copy drafts without inventing product claims.
    • Calculate campaign metrics and state what the data cannot establish.

    In this lesson

    1. Define and test three customer personas
    2. Draft A/B ad copy with AI
    3. AI drafts, people finish: a blog workflow
    4. Find insights in campaign results

    4 guidesAbout 54 min reading, plus practiceBeginner → Intermediate · practical