EVIDENCE · SCENARIOS

Test the decision with thousands of scenarios.

Describe the decision, the assumptions and the uncertainty in each segment. The simulator runs thousands of synthetic scenarios and shows whether the decision holds, what weighs most on the result and what is still unknown.

Runs in your browser, no account, no network and no personal data. Everything shown here is a synthetic scenario, never observed data. Need a baseline? Generate a dataset in the Studio.

01 / The decision

Illustrative model scenarios

Decision Metric the decision moves

02 / Assumptions Current metric rate (%) Minimum change that supports the decision (%) Required confidence Observations per group in each segment

The change is relative to the current rate: 5% on 12% leads to 12.6%. With 0 observations the scenario shows only the uncertainty in your assumptions; with a number it adds the noise of measuring the effect with that sample.

03 / Segments and effect range

For each segment: its share of the total and the range you believe the effect falls in (a 90% interval). Shares are normalized to add up to 100%.

04 / Run Simulations

Seed

The same seed with the same assumptions reproduces exactly the same result.

SCENARIO RESULT

Synthetic

Fill in the assumptions and run the simulation. No result is presumed before that.

Most likely change
90% interval
Chance of reaching the minimum
Chance of getting worse

Each bar counts synthetic scenarios. The line marks the minimum change that supports the decision.

What weighs most on the result

How much the total change moves when one segment goes from the minimum to the maximum of its range while the others stay in the middle.

By segment

Segment	Weight	Median	90% interval

Decision envelope

SUPPORTS
DOES NOT SUPPORT: Claiming the effect exists. This is a synthetic scenario built from your assumptions.
NEXT EVIDENCE

HOW THE SIMULATOR WORKS

Explicit assumptions. Visible uncertainty. Nothing observed.

Ranges, not guesses

Each segment gets an effect range read as a 90% interval. The simulator draws the true effect from it in every scenario.

Sample size matters

With observations per group, each scenario adds the noise of measuring the effect with that sample. A small sample widens the interval.

Envelope, not verdict

The result says whether the assumptions support the decision and which real evidence would change that. Read the method