Skip to main content

Reading Results

Once your test is running and has been analyzed, the test detail page shows everything you need to make a decision.

Page layout

The test detail page has a header and three tabs:

  • Test name, description, and status badge
  • Progress bar showing how far through the test period you are
  • Action buttons: Run Analysis, Complete, Archive, Duplicate, Delete

Overview tab

This is where you spend most of your time. It contains:

Recommendation card — the headline: SCALE, KILL, or CONTINUE with a confidence score and supporting checklist. See Recommendations for how these are determined.

Cumulative effect chart — a time-series showing the estimated treatment effect over the test period, with confidence intervals that narrow as data accumulates.

DiD results card — the Difference-in-Differences regression output:

  • Point estimate (treatment effect)
  • Standard error
  • 90% and 95% confidence intervals
  • Baseline mean (control group average)
  • Observation count

Bayesian card — probability-based results:

  • P(positive effect) — probability the experiment helped
  • P(exceeds target) — probability you hit your MDE target
  • P(harm) — probability the experiment hurt
  • 90% credible interval

Guardrail metrics — secondary metrics you're monitoring for side effects. See Guardrail Metrics.

Data quality warnings — automated checks for issues like sparse data, zero variance, or group imbalance. See Data Quality.

Locations tab

Lists all treatment and control locations with names and cities.

Data tab

Shows the underlying dataset information and historical data details.

Analysis states

Your test moves through analysis stages:

StateMeaning
ScheduledTest is set up but hasn't started yet
Collecting DataTest is running, accumulating observations
ReadyEnough data to run analysis
AnalyzedAnalysis complete, results available

Click Run Analysis at any time after data collection begins to get fresh results. You can re-run as often as you like—each run uses the latest available data.

tip

You don't have to wait for the test period to end. ProofPod's Bayesian approach lets you check results at any point and will tell you if you have enough data for a confident call.