Experimentation
Experiment results your teams can trust.
Analyze the A/B tests you run on the data in your warehouse, next to your product analytics and KPIs.
An A/B test result: one-page checkout beats three steps on order conversion by 4.2%, with the 95% interval, p-values for each metric and health checks.
Know when a winner is real.
Lift and significance
See the lift, its interval and whether it holds.
Every metric gets its lift against control, a 95% interval and a p-value, corrected across comparisons.
+4.2%
Significant winOne page
Three steps
Guardrails and health
Check that a win is safe to ship.
Traffic split, users in several variants, window coverage and guardrail metrics, on every test.
- Traffic split50/50, p = 0.43
- Users in several variantsnone
- Window coverage14 of 14 days
- Sample against plan102% of plan
Refund rateGuardrailWithin guardrail
Your tool, your data
Keep your experimentation tool.
It assigns users and sets up variants; Mitzu reads the exposures from your warehouse and analyzes the results.
Optimizely
LaunchDarkly- or your own assignment
Governed by your data team.
- Nothing copied out of your warehouse
- Self-hosting and data residency available
- You control whether the AI sees query results
- Your data is never used to train AI models
Experimentation: frequently asked questions
Read every experiment on the numbers your business reports.
Start free on your warehouse, or book a demo and we'll show it on your data.