Direct answer — What is pre-production concept testing for mobile games and when should a studio do it? Pre-production concept testing for mobile games means buying paid traffic against a game that does not exist yet — a theme, an art direction, a hook, a store listing — and reading install rate and install cost as a proxy for market appetite. You run it in pre-production, before engineering budget is committed, which is exactly what separates it from a soft launch (a working build, live in real markets) and from creative testing (a live game, scaling paid UA). The output is not a green light. It is a price of entry: if your concept cannot buy attention affordably on a mocked-up listing, a finished build will not rescue it.
Most studios I meet run this sequence backwards. They build for nine months, then discover at soft launch that the concept was never marketable — when the only remaining lever is a rebuild nobody has budget for. Mobile game concept validation moves that discovery forward, to a stage where changing the art direction costs a creative wave instead of a milestone.
Why this matters more in 2026 than in 2022 is volume. AppsFlyer’s State of Gaming for Marketers 2026, published 14 January 2026, reports “top gaming advertisers producing between 2,400 and 2,600 creative variations per quarter, up 25–30% YoY.” You are not competing against other concepts. You are competing against an industrial creative machine.
Key Takeaways
- Concept testing answers one question: what will attention cost? It says nothing about retention, monetization, or whether the game is good.
- Test CPI is a ranking instrument, not a planning input. Lancaric: “Your estimated CPI is usually 1.5–2× wrong.”
- Click tests measure the weakest signal in the chain. Install tests need a store listing and tell you something real.
- Google Play’s closed testing rule adds three to five weeks to any Android install test on a personal developer account.
- Write the kill criteria before the test runs. A test with no pre-agreed threshold always concludes “promising”.
What a concept test is, and what it is not
Three practices get called “testing”. They sit at different points in the timeline, cost different money, and answer different questions.
| Practice | What exists already | What it measures | Typical decision |
|---|---|---|---|
| Concept test (pre-production) | Assets and a listing, no game | Install rate, cost of attention | Build it, reshape it, or drop it |
| Soft launch | A playable, monetizing build | Retention, ARPDAU, early payback | Scale, iterate, or sunset |
| Creative testing (live) | A shipped, scaling game | Creative win rate, cost trend | Which creatives get budget |
Confusing the first two is the most expensive mistake in the sequence. A complete soft launch guide is about a game that already works; concept testing before soft launch is about whether the market wants what you are describing at all. And ad creative strategy for a live game is volume mathematics applied to a title that already has a P&L.
Separating those three is usually the first thing I do in a mobile game consulting engagement: studios arrive asking for a soft launch plan when what they need is permission to not build the game at all.
The two tests, and why only one of them decides anything
There are two honest ways to run mobile game marketability testing, and they are not equivalent. A click test reads click-through against a landing page or an unbuilt listing. A CPI test reads cost per install and IPM against a real store listing with attribution in place.
Matej Lancaric declared the first category finished in December 2025: “Marketability tests are dead.” The argument is narrow and correct. “CTR tests only measure clickers,” while “CPI tests simulate the real funnel.” A click is a reaction to a thumbnail. An install is a decision, and only the second predicts your acquisition cost.
My position after 20+ years across Gameloft, SFR and Blacknut, and €12M+ of managed P&L: run a click test only to rank art directions quickly, and never let its output leave the art department. Anything touching a greenlight or a financial model needs an install signal.
| Click test | CPI test | |
|---|---|---|
| Build required | No | Yes, store-listed |
| Time to a read | 3-5 days | 2-3 weeks including listing setup |
| Signal quality | Weak, measures clickers | Strong, measures installers |
| Fit | Ranking art directions | Greenlight and budget decisions |
What a readable concept test costs
The media line is smaller than most founders expect; the setup is bigger. In the engagements I run, a readable wave on Tier-1 Android lands at roughly $600 to $1,200 in media for two or three concepts, with six to ten creative variations costing $1,200 to $3,000 if you buy production rather than make it in house. Those are planning figures from my own practice, not a published benchmark.
Supersonic, Unity’s mobile publishing arm, sets the creative floor plainly: “We suggest creating 4 versions of your creative to test in the marketability stage,” and notes that “The first 3-5 seconds of a marketability creative are the most important.” Four variations per concept is the minimum that lets you tell a bad concept from a badly executed one.
What breaks a creative concept testing budget is not price, it is spread. Below a few hundred dollars per day per ad set the platform cannot separate signal from noise. Three concepts funded from a wave sized for one buys three unreadable results — worse than testing one concept properly, because it feels like data.
One scheduling trap deserves its own line. Google Play requires developer accounts created after 13 November 2023 to run a closed test with “a minimum of 12 testers who have been opted in continuously for at least 14 days” before production access and pre-registration unlock. If your test needs a Play listing, that lands three to five weeks ahead of the media spend, and I have watched it derail a launch calendar twice.
Not sure whether your concept is ready to test? Book a strategy call and we will go through the concept, the creative brief and the kill criteria.
Reading the result without fooling yourself
Concept testing produces one number that lies and one comparison that does not.
The number that lies is the absolute CPI. Small budgets, narrow targeting and fresh creative all flatter it, which is why Lancaric’s “1.5–2× wrong” is the most useful line in this discipline. Never carry a raw test CPI into an LTV model. The reason acquisition costs keep drifting upward is structural rather than tactical, and your test sits outside that structure by design.
The comparison that holds is concept A against concept B: same day, same geo, same budget, same creative count. AppAgent’s rebrand test for Pixel Federation is a clean published example of it working on a rebrand rather than a new title: “The test delivered a 38% increase in iOS IPM for the Seoul Secrets concept and a 20% decrease in Android CPI,” with no change to the core game.
Three rules I write into every test brief:
- Kill criteria first. Define, before launch, the install-cost level at which the concept does not proceed.
- One variable per concept. If concept B changes the art direction and the hook, a win tells you nothing actionable.
- A win is a licence to build, not to plan. Carry the ranking forward, then re-derive your acquisition costs during soft launch market selection, where the numbers behave.
What we could not verify
A figure circulates widely on this topic: that a $5,000 to $10,000 pre-production test prevents around $500,000 in misdirected launch spend. Every trail I followed led back to aggregators recycling each other, or to vendor blogs with no dataset behind them. It is not in the source list above and not repeated here as fact.
The underlying logic is sound: killing a bad concept early is cheaper than killing it late. But treat the arithmetic as a hypothesis to test against your own cost base, not as a benchmark. I also found no published absolute go/no-go install-cost threshold for any genre, and do not expect one, because the threshold is relative to lifetime value and therefore specific to your game.
Conclusion
Pre-production concept testing is not a way to be certain. It is a way to be wrong for four figures instead of seven, while being wrong is still recoverable. Run the install test when money is attached to the decision. Treat the ranking as the deliverable and the absolute CPI as noise.
Deciding how much of your pre-production budget to spend on validation? Book a Strategy Call with Game Growth Advisor and we will size the test against your build cost.