Ideation

Experiment board / hypotheses

Hypothesis, test, metric, success criteria, result, learning. One row per experiment.

When. Turns ideas into cheap tests before building them. The board keeps the memory of learnings.

Duration
1.5 h
Participants
Product trio
Output
3 planned experiments with criteria
Stage
Idea · Pre-PMF · Post-PMF · Scale-up · Mature
Hypothes…TestMetricSuccess …ResultLearningExperiment 1Experiment 2Experiment 3
Open in FigJamFind the right sequence

In FigJam: File → Duplicate to your drafts, then keep this workshop's section. Get the file on Figma Community

Filled example

HypothesisTestMetricSuccess criteriaResultLearningExperiment 1Experiment 2Experiment 3photo-only mode inweek one doublesday-5 activationdefault for 20 newcompanies, full formfor 20 otherstechnicians with 10jobs logged by day 540 % vs 20 %baseline37 % vs 19 %parts are theblocker, not photosan onboarding kitgets technicianstold before day onekit sent to 10companies, none to10 otherstechnicians who logon day one60 % vs 30 %55 % vs 28 %the video mattersmore than thechecklisttechnicians will payattention to acounter to 10fake-door counter onthe home screen, taptrackedtap rate in week one30 % tap it at leastonce11 %dropped, the counteris a Fieldnote idea,not a technicianneed

Example data. Fieldnote is a fictional company used to fill all 61 boards, so the examples tell one consistent story.

Post-PMF Fieldnote: the trio turns the two winning ideas plus one assumption into experiments for the next three weeks.

What the team took away

Two of three held up. Photo-only mode shipped as the default, the kit went to Julien's onboarding, the counter died cheaply.

How to run it

Preparation

Agenda

  1. 15 min
    Rewrite the assumptions as hypotheses
    Format: 'we believe [segment] will [behavior] because [reason]'. One per row. If you cannot name the behavior, the hypothesis is not testable yet.
  2. 25 min
    Choose the cheapest test
    For each: fake door, concierge, landing page, prototype interview, data pull, or a real small release. Pick the one that gives a signal in two weeks or less. Write who runs it.
  3. 25 min
    Define metric and threshold
    One metric per test and the number that means 'validated' or 'killed', decided now. 'We will see' is not a criterion. Include the sample size you need.
  4. 15 min
    Check for confounds
    For each test, ask what else could explain a positive result. If the answer is obvious, change the test. Add the control where needed.
  5. 10 min
    Schedule and commit
    Start dates, review date, owner in each row. Book the review in the calendar before leaving.

Pitfalls

Afterwards

Each row is reviewed at the agreed date; result and learning are filled in the same meeting. Validated hypotheses move to the story map or the roadmap; killed ones update the assumption map and the OST.

Source : Eric Ries / Ash Maurya

Related workshops

Back to the compassAll workshopsVersion française