Written in by More Playables production team
Mobile Game Creative Testing: From Hypothesis to Decision
A shared test register for concepts, executions, formats, and controlled production changes across video and playable creative.
Operational framework; campaign analysis requires the buyer's sourced data
Updated
Separate concept, execution, and delivery
A concept is the audience promise or fantasy. An execution is the way a specific asset expresses it. Delivery is the campaign context in which that asset was shown. When those levels are mixed, teams turn a result from one asset into a universal creative rule.
Test level | Example question | Production consequence |
|---|---|---|
Concept | Does rescue fantasy deserve further investment? | Create distinct expressions across one or more formats |
Execution | Does the consequence-first opening communicate rescue faster? | Change the first scene while holding the promise |
Format | Does acting on the rescue add value over watching it? | Build a playable as new creative; do not call it an automatic conversion |
Tuning | Does lower first-board difficulty improve completion? | Produce a qualified variant |
Delivery | Did one audience or placement receive different exposure? | Investigate campaign setup before changing creative |
Use a creative taxonomy
Assign stable values for concept, hook, format, mechanic, art theme, difficulty, reward, CTA, language, master, variant, and export. The taxonomy does not need to be exhaustive; it needs to be consistent enough to retrieve what actually ran.
Plan a review cadence
- Before launch, record the decision, metric definitions, allocation method, and known confounders.
- During the test, check data quality, delivery imbalance, and broken creative before reading winners.
- At the decision point, compare the primary observation with guardrails and uncertainty.
- Record keep, iterate, replace, investigate, or inconclusive.
- Send one specific learning back into the production backlog.
Avoid a volume-only system
Creative velocity is the rate of useful learning, not only files shipped. A smaller set of well-identified treatments can be more valuable than many untracked combinations. Reuse supported masters when the next question is a parameter or presentation change; commission a new master when the hypothesis needs a different interaction.
Preserve downstream quality
Starts, clicks, and installs can reveal different parts of the funnel. Also examine retention, payer or revenue signals, or another agreed quality measure when available. A creative can increase low-cost installs while attracting people who do not stay, and privacy or attribution limits can make downstream comparison incomplete.
The register is the durable asset
For every treatment, preserve the exact creative ID, files, taxonomy, campaign context, metric definitions, result source, decision, and follow-up. This allows later reviewers to distinguish a real learning from a renamed anecdote.
Use the editable test matrix as the first version of the register.
Sources
Put this into practice
- Creative test matrix
Reuse the same decision record across formats.
- Creative test metrics
Read funnel and downstream metrics together.
Make your next test playable
Turn a creative question into a scoped playable, variant set, and delivery plan.
Start a playable brief