Turn TikTok Mini Game signals into a test matrix by extracting observable variables and changing one variable at a time. The current Golf Orbit query returned 22 candidate records, but only five contained enough visual description to analyze opening action, input cue, progress feedback, reward, and CTA. That evidence supports a creative-variable pool, not a TikTok performance benchmark.

Why do Mini Games need a separate creative variable sheet?

TikTok’s official documentation says Mini Games launch directly inside the TikTok app. A shorter path means a mismatch between the ad promise and the first play experience can become visible immediately. The testing sheet therefore needs more precision than “change the hook” or “try a brighter color.” Record the first action shown, the input method implied, the rule feedback, and whether the transition into play preserves the same promise.

Which variables appeared in the Golf Orbit sample?

Five of the 22 candidate records contained usable visual evidence. The reviewed materials showed swing actions, fan-shaped trajectories, “Perfect” feedback, neon rails, rainbow planets, confetti, and title-card closures. The highest cumulative-impression analyzable record showed 1,421,036 impressions, 1,170 active days, and two linked ads. It can identify variables worth reviewing, but it cannot be labeled a TikTok winner or the best creative in the current period.

How should the three-by-three matrix be built?

Place opening action on the horizontal axis: immediate swing, fail-and-retry setup, or visual spectacle. Place gameplay proof on the vertical axis: explicit touch cue, trajectory prediction, or before-and-after result. Hold reward feedback and CTA constant in every cell. The first round tests the promise and proof combination. A second round can vary reward intensity. This sequence tells the team whether the difference came from attention, rule comprehension, or the payoff.

Which metrics belong to creative review and which belong to delivery?

The creative sheet records the presence of input cues, rules, progress, failure, reward, and CTA continuity. TikTok’s reporting documentation defines Active, Launch, and Day X ad-revenue ROAS as platform reporting fields that become relevant after delivery. Do not substitute cumulative impressions, active days, or linked-ad counts for those outcomes. The Golf Orbit records were not filtered to TikTok media, so they form a creative input pool rather than a placement-specific distribution.

How should the team prioritize the first tests?

Start with the variable that remains least explicit in the current sample. Several Golf Orbit materials show the ball path and celebratory result without a clear finger or button action. An explicit input cue should therefore rank above another background change. Failure feedback is the second priority because success-only footage does not fully explain the challenge. Visual exaggeration comes third; it can amplify a clear rule, but it should not be asked to replace one.

Which quality fields belong beside the matrix?

Each row should also retain evidence coverage, material ID, first and last observed dates, cumulative impressions, active days, linked-ad count, and whether an explicit CTA was captured. Evidence coverage determines how strongly the team can describe a variable. A record with only a metric and no visual description can remain in the retrieval queue, but it cannot support an inference about input. This prevents analysts from filling empty matrix cells with product assumptions simply to make the grid look complete.

How should the matrix be completed after launch?

Link every cell to an ad group and version ID, then add Active, Launch, and Day X ad-revenue return after delivery. Confirm that event definitions and time windows are aligned before comparing cells. If creative version, audience, and bid strategy all changed together, record the result as a mixed experiment rather than attributing the difference to the hook. The next cycle should retest the most interpretable one-variable combination.

What is the final action framework?

Choose one opening action, one gameplay-proof method, and one fixed reward loop for each round. Before launch, confirm that the promise can be reproduced in the first playable experience. After launch, use the official TikTok reporting fields to evaluate Active, Launch, and revenue return. Creative semantics define what changed; delivery data evaluates what happened. Keeping those jobs separate makes the matrix interpretable and reusable.