Outcomes

Creating an outcome

Outcomes are always linked to a Bet. Open the bet and click Log outcome. You'll be asked for:

• Ship date — when did this actually go to production? You can enter it manually, or if Jira/Linear is connected and detected the ship event, it's pre-filled.
• What shipped — a brief description of what was actually built. This may differ from the original hypothesis — that's worth noting.

Once created, the outcome moves to Measuring state. This is the window where your analytics are pulling data against the predicted impact.

Connecting analytics

Connect your analytics integration from Settings → Integrations. Findry supports PostHog, Amplitude, Mixpanel, and GA4. Once connected, configure the metric mapping for your project — which event or property in your analytics tool corresponds to which metric in Findry's metric library.

After connecting, open the outcome and click Pull metrics. Findry queries your analytics tool for the metric value over the timeframe you defined in the predicted impact. The raw data appears in the outcome's Metrics section alongside the predicted value.

Reading the variance report

The variance report compares your predicted impact against the observed outcome:

• Predicted — what you committed to in the Bet's structured predicted impact
• Observed — what your analytics reported over the measurement window
• Variance — the delta — expressed as both absolute and percentage difference
• Statistical significance — for A/B experiments connected via PostHog or Amplitude, Findry shows the p-value and confidence interval

The variance is color-coded: green for outcomes within or above the predicted range, amber for outcomes that partially met the prediction, red for outcomes that missed.

“A variance report isn't a scorecard. A miss that's well-documented is more valuable than a hit that nobody wrote down. The record is the record.”

Writing the verdict

After reviewing the variance report, write the verdict. Three options:

• Hit — the outcome met or exceeded the predicted impact
• Miss — the outcome did not meet the predicted impact
• Unclear — the data was insufficient, the metric moved for unrelated reasons, or the ship was too narrow to measure

The verdict is accompanied by a required Lessons field — one to three paragraphs explaining what you learned. Resist the urge to skip this. The lessons are what make the record useful six months from now when someone asks "why did we ship this?"

Loop closure

After saving the verdict, Findry prompts you to generate signals from the outcome. These are new observations — what you learned from measuring what shipped. They go into the Signals backlog, tagged with the outcome they came from, ready to seed new hypotheses. This is what closes the loop: the evidence from one decision becomes the raw material for the next.