Run and read A/B tests on your warehouse data, with no extra tracking or data copies. ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­  
Mitzu_logotype_box_melon_gradient (1)

Today's drop: Experimentation 🧪

Hi there,

 

Mitzu Launch Week is here. Five releases, one a day, starting now.

 

Today: Experimentation. Run and read A/B tests directly on your warehouse data, with no extra tracking or data copies. Your experimentation or feature-flag tool keeps assigning users to variants. Mitzu reads the exposure event it leaves in your warehouse and measures every variant with the saved insights your team already trusts.

 

Here's what that looks like.

1. A verdict in plain words

Press Evaluate and the Results card leads with the answer: which variant won, by how much, and how sure you can be. Below it sits every metric for every variant: the primary metric that decides the test, secondary metrics, and guardrails that must not get worse.

Mitzu Experiments results card: Annual first beats Monthly first on Signup to Trial Started, +11.4%, significant win, with secondary and guardrail metrics below

2. Health checks before you trust a win

Six checks run on every evaluation. If the traffic split doesn't match the plan (a sample ratio mismatch) or users saw more than one variant, Mitzu withholds the verdict instead of handing you a confident wrong answer.

Six health checks in Mitzu Experiments: traffic split, users in several variants, sample against plan, window coverage, metric tracking and users left out, all passing

3. A decision on record

When the test ends, Finalize records the outcome, the decision and a summary drafted from the results, and freezes the numbers. Save each variant's users as a cohort and mark the roll-out on every chart, so months from now anyone can see what was tested, what won and why.

The Finalize the experiment dialog in Mitzu: outcome Win, winning variant Annual first, decision Roll out, a roll-out date and a summary drafted from the results

All of it runs on your own warehouse: no SDK, no sync, no copy of your data. Mitzu never assigns users, changes a running test or touches your feature flags.

See it in action

Experiments in Mitzu: plan, measure and decide A/B tests

📖 The full story, with a step-by-step setup guide, is on our blog: Experiments in Mitzu: Warehouse-Native A/B Test Analysis

 

Every setting is covered in our docs.

 

Four more drops this week. Tomorrow: the numbers your data team already calculates in the warehouse, on any Mitzu chart.

Try Experiments

Mitzu, Inc., 3500 South Dupont Highway, Dover, Delaware 19901, United States

Unsubscribe Manage preferences