Writing the pre-mortem

After the causal chain, every hypothesis goes through a pre-mortem. The point is to deliberately imagine the bet failing before you commit a confidence number. This calibrates the next step and surfaces risks while you can still act on them.

You'll see this page as Step 2 of 3 in the creation flow.

1. Weakest link

Which step in your causal chain is most likely to break? Be specific — name the claim, not the whole bet.

Example: "Behavior Change — users may complete the shorter onboarding but skip the integration step that drives activation."

2. Base rate

What does history say about similar bets — yours, your team's, or the industry's? If you've never tried something like this, say so.

Example: "Past onboarding shortenings here gave +3–5pp on completion but no lift on D7. Industry reports suggest +5–10pp."

3. Concurrent confounds

What else is changing in the same window that could move the same metric? Marketing campaigns, pricing tests, seasonality, other PMs' shipments.

Example: "Growth team is running a paid acquisition test in the same week. Different traffic source, but it will skew our cohort."

4. Earliest failure signal

If this bet is going to fail, how will you know early — not at the resolution date? Define the leading indicator.

Example: "Onboarding-completion rate at day 3. If it doesn't move within the first week, the rest of the chain won't either."

5. Kill criteria

A concrete trip-wire. If a specific metric crosses a specific threshold, you stop and reassess. This converts a vague worry into a real decision rule.

• Metric — the number you'll watch.
• Threshold — the value below which you act.

6. Continue to confidence

Once you've worked through all five prompts, click Continue to confidence →. Doing the pre-mortem first is intentional: it lets the act of imagining failure adjust your confidence, instead of the other way around.

Previous: Creating a hypothesis ←    Next: Setting your confidence →