Experiment Workbench
1) Console specs & variants → 2) Experiment & telemetry → 3) Validation & approval after you run.
API is offline — use Refresh in the top bar, then run analysis.
Step 1
Console specs & variants
Each card is a different test variant (arm) in the benchmark environment — control plus three product profiles. Pick the profile that frames how we interpret difficulty, rewards, and onboarding in step 2.
Variants in the benchmark
A full multi-arm run (e.g. exp_sanity_001_calibrated) compares all four arms side by side. The cards below are the same variant types you assign when building a custom experiment.
- Baseline (control)(control)Standard difficulty, light onboarding, no aggressive promos.
- Fast progression(fast_progression)Higher rewards and faster progression for engagement tests.
- High challenge(high_challenge)Harder curve for skilled players; minimal hand-holding.
- Guided onboarding(guided_onboarding)Lower friction and guided flows for new-user retention.
Select your framing profile for step 2
Baseline (control)
- difficulty
- 0.0
- reward rate
- 1.0×
- matchmaking
- 12s
- onboarding
- light
- promotions
- none
Experiment source
Step 2
Experiment outline
Select which test to analyze and run telemetry. Console profile from step 1: Baseline (control). Preview below lists the variants in that experiment.
After you run analysis, validation and recommendations appear here (step 3).
New here? Read how the system works