1
Day 1
Record the baseline for play-through rate, average session, D1 retention, and current page assets.
Baseline note and screenshots of the current title, icon, thumbnail, and description.
Soft launch testing
Roblox Soft Launch Experiment Planner
Turn a weak metric or creator hunch into a 7-day experiment that changes one variable and produces a readable decision.
7-day experiment schedule
2
Day 2
Prepare exactly one change: test a reward-focused thumbnail instead of the current danger thumbnail.
One finished change that fits 3-5 hours, with no extra gameplay or monetization edits.
3
Day 3
Publish the change and write down the exact time it went live.
Launch note with timestamp and what changed.
4
Day 4
Check early movement in play-through rate, but do not make another change yet.
Early read: up, flat, or down.
5
Day 5
Read comments, private feedback, and playtest observations for the same bottleneck.
Three user phrases that explain the metric movement.
6
Day 6
Decide whether the experiment needs one more day of data or is clearly failing.
Keep observing, revert, or prepare the next variable.
7
Day 7
Keep, revert, or iterate based on the decision rule.
Next experiment brief with one metric and one variable.
Prepare the experiment assets
Diagnose analytics firstMake sure this experiment targets the current bottleneck.Check game page clicksUse this when the experiment targets title, icon, thumbnail, or description.Build Launch PackTurn the experiment into a full launch plan.Plan update roadmapUse this if the test points to retention or update cadence.
How to run a useful soft launch test
A soft launch should reduce confusion. The goal is not to fix everything in one week, but to learn which lever actually moves.
How to use it
- + Pick one metric: play-through, average session, D1 retention, payer conversion, or update return.
- + Write one hypothesis.
- + Change one variable.
- + Record baseline and decision rule before publishing.
Example outputs
- + Thumbnail test: change thumbnail only and watch play-through rate.
- + First-session test: change spawn and first reward only, then watch average session.
- + Retention test: add one daily reward and watch D1 retention.
Common mistakes
- + Changing the thumbnail and tutorial and game passes on the same day.
- + Judging a test without a baseline.
- + Keeping a change because it looks better even when the target metric falls.
It is a small controlled test before or during early launch where you change one variable and watch one main metric.