The version of this page I published in May 2026 carried a genre CPI ladder that sat one full tier above the source it cited, a learning-phase budget rule that was arithmetically impossible, a channel mix that summed to 120%, and a headline claim that TikTok delivers 15-25% cheaper installs than Meta, repeated four times, supported by neither of the two sources listed at the bottom of the page.
I have rewritten it against what the primary pages actually say. What survived is smaller and more useful: a genre CPI scale from a named managed portfolio, TikTok’s own published learning-phase criteria, and an audience profile for game advertisers. Everything else was either relabelled with its real scope or removed, and the removals are listed at the end.
Key Takeaways
- TikTok gaming CPI runs $0.25-$0.80 hypercasual Android and $2.50-$6.00 midcore, per Admiral Media’s managed portfolio. The mid-teens to low-twenties midcore CPIs quoted across TikTok content, including on this page until today, trace to nothing
- The learning phase resolves at about 25 campaign results or 7 days, per TikTok’s documentation. That threshold is not a budget rule, and any article converting it into a daily-budget multiple of your target CPA is inventing arithmetic
- $5,000 per month buys data on one platform. Splitting iOS and Android with separate creative pools needs the $15,000 tier, which is the same source’s figure for exiting learning inside two weeks
- Over 60% of TikTok users identify as mobile gamers and download roughly 50% more games than non-users. Under-30s are about 40% of the base, so the “Gen Z channel” framing is out of date
- The CPI comparison against Meta is unpublished. I looked for it and did not find it in any source I would cite. Run the comparison in your own accounts, on your own game
What a TikTok Gaming Install Actually Costs
One source publishes a genre-level TikTok gaming CPI scale against a stated portfolio. Admiral Media reports more than $30M of TikTok gaming spend across 150+ game campaigns, and publishes these bands.
| Genre | Android CPI | iOS CPI |
|---|---|---|
| Hypercasual | $0.25 - $0.80 | $0.50 - $1.50 |
| Casual | $0.80 - $2.50 | Not split by platform |
| Midcore | $2.50 - $6.00 | Not split by platform |
Read the limitations before you plan against this. Admiral splits only hypercasual by platform, so the casual and midcore bands are blended across iOS and Android and will understate iOS. It does not state the geography behind the numbers, so these are portfolio-level figures rather than Tier 1 figures. And they describe one agency’s inventory, which is not the market.
Two anchors sit inside the same source and are worth carrying. Its hypercasual scale case held an Android CPI at $0.35 while spend went from $10K to $400K per month over eight weeks, which is at the bottom of the hypercasual band. Its midcore RPG case moved an iOS CPI from $6.80 to $3.90 in six weeks, which lands inside the midcore band rather than above it. A midcore iOS CPI in the mid-teens or above has no support in this source, and I did not find another that gives one.
For how these bands relate to the wider market, and why published CPI figures disagree by an order of magnitude depending on whether they are volume-weighted global blends or Tier 1 estimates, see our mobile game CPI benchmarks for 2026. The short version: a global blended figure and a Tier 1 figure can differ by a factor of fourteen while both being accurate, so never place them in the same table.
The Learning Phase, and Why It Is Not Your Budget Rule
TikTok’s help centre states it plainly: “Typically, volatility starts to decline after about 25 campaign results or 7 days from when the campaign enters the learning phase.”
That sentence gets misquoted in three distinct ways across TikTok UA content, including in the version of this article it replaces.
- The number is 25, where the figure in circulation is 50
- It applies at campaign level, where most playbooks apply it per ad group
- It reads 25 results or 7 days, an either/or with a clock, rather than a conversion count you must reach
The last point matters most, because it is what breaks the budget rules people build on top of it. If the exit criterion is 25 results in 7 days, that is roughly 3.6 results a day. At a $2.50 casual CPI, the pure threshold is under $10 a day. Nobody runs a $10-a-day TikTok campaign and learns anything, which tells you the learning phase exit criterion was never the binding constraint on your budget.
The binding constraint is statistical readability. You need enough installs to tell a winning creative from a losing one, and enough days at stable delivery to trust the number. That is a different and much larger figure, and it is the one Admiral publishes.
What $5,000 and $15,000 Actually Fund
Admiral gives two budget figures. “$5,000/month is the minimum to gather meaningful data” and “$15,000+ is needed to exit TikTok’s learning phase within two weeks and begin meaningful ROAS optimisation.”
Run the arithmetic before you plan against them, because the gap between the two is where most TikTok pilots fail.
$5,000 a month is about $165 a day. At Admiral’s casual band of $0.80-$2.50 that buys roughly 66 to 205 installs a day. At the midcore band of $2.50-$6.00 it buys 27 to 66. Now split that across separate iOS and Android campaigns, which every TikTok playbook including this one tells you to do, and halve it again. A midcore studio at $5,000 a month is buying roughly 14 to 33 installs a day per platform, which will not separate one creative from another inside a month.
So the honest reading of the two figures is this. $5,000 funds one platform and one creative track, and it answers a single question: does this game find an audience on TikTok at all. $15,000 funds the architecture, meaning separate iOS and Android campaigns with their own creative pools and their own budgets, read at two-week intervals. If your plan calls for the architecture, quoting the $5,000 floor at your CFO is setting the pilot up to fail.
Budget in that second tier for at least six to eight weeks. A creative wave from brief to shipped ad takes most small studios two to three weeks, so a four-week pilot gives you one wave and no chance to act on what it taught you.
Ready to put a real UA engine behind your game? Book a strategy call and we will pressure-test your TikTok plan against your LTV math before you spend.
Ad Formats That Matter for Mobile Game UA
| Format | Best for | What the sources support |
|---|---|---|
| In-Feed App Install | All genres, top of funnel | The default. Native scroll placement, optimised for installs |
| Spark Ads | Casual, hybrid-casual, hypercasual | Boosts a live creator post as a paid ad, keeping comments and follower context |
| Playable Ads | Casual, puzzle, hybrid-casual | AppFollow states playables are “about five times more effective at driving game downloads than standard mobile ads”. Vendor claim, no dataset published |
| Branded Effect / AR | Brand pushes, IP launches | Cultural reach rather than direct response. No install-efficiency figure published |
| TikTok Pulse | Premium adjacency, midcore IP | Places your ad against the top 4% of trending gaming content, at a CPM premium |
The workhorse combination for most studios is in-feed app install plus Spark Ads, with playables layered on once you have a stable install flow to protect. I have removed the 40/40/20 budget split this page used to prescribe, because it was an assertion rather than a finding and nothing in either source supports a specific ratio.
On playables specifically, note how large the AppFollow figure is. Five times more effective is a far bigger claim than the 25-40% uplift this article previously published, and both are unverifiable in the same way. What holds regardless of the multiplier is the mechanism: a playable makes the user demonstrate interest before you pay for the install, so it shifts cost from acquisition to qualification. Whether that trade is worth it depends on your production cost per playable, which varies too widely by studio for a published range to be meaningful.
Spark Ads and Creator Programmes
Creator activation through Spark Ads is where TikTok stops resembling a media channel and starts resembling a supply chain. The mechanics are worth getting right.
- Recruit creators who play your genre, and brief them on game pillars and emotional beats rather than scripts
- Negotiate Spark Ads authorisation up front, since TikTok requires the creator to authorise their own post for paid amplification and retrofitting that permission is painful
- Run several variants per creator and scale only what clears your hook-rate bar
- Rotate the pool before performance decays, rather than after
I have removed the claim that micro-creators deliver 2-3x better ROAS than mega-influencers. It is repeated widely, it matches what I have observed, and I could not source it. What I will say without a number: a creator whose audience already plays your genre converts better than a larger creator whose audience does not, and the price difference compounds that advantage. Test it in your own account before you build a budget on it.
The studios that succeed here treat creators as UA suppliers, with volume deals, briefs written to the same standard as an agency brief, and CTR and CVR read weekly. The studios that fail treat them as influencer marketing, expect virality, and cannot explain their CPI afterwards.
For a systematic framework on what makes mobile game ad creative convert across networks, see our mobile game ad creative strategy guide. For the broader validation framework, our mobile soft launch complete guide covers market selection and KPI gates.
Hybrid Creative: Casual Hooks on Midcore Games
The most replicated creative pattern on TikTok is the hybrid hook. Open with a few seconds of a casual mini-game or a satisfying mechanic such as slicing, sorting, merging or pin-pulling, then pivot into the strategy, RPG or 4X gameplay you are actually selling.
Admiral’s own casual puzzle case supports the direction without quantifying the CPI effect: gameplay UGC hybrids outperformed studio-produced ads by 3x on hook-rate for that client. I have removed the claim that hybrid hooks lower CPI by 20-40%, which was unsourced and is a much stronger claim than a hook-rate result.
A workable 15-second structure:
- 0-3s: casual hook, a mechanic that resolves visibly inside three seconds
- 3-7s: the payoff, a reveal, a win moment or a character reaction
- 7-12s: real game footage carrying the value proposition
- 12-15s: store call to action with logo and rating
This does not require you to misrepresent your game. It requires you to find a genuinely casual moment inside it, which most midcore titles have several of, in a tutorial step, a satisfying interaction or a meta layer. Most studios simply never lead with them.
Creative Cadence Without a Fatigue Number
This page used to state that TikTok creative fatigue runs on a 7-14 day cycle, then prescribe a refresh every two weeks. Those two instructions cannot both be right. A fortnightly refresh against a 7-14 day half-life replaces the creative at the exact moment the article says it is already dead.
I could not find a published fatigue half-life for TikTok gaming creative that I would cite, so I have removed the number rather than pick a side of the contradiction. Replace on a measured trigger instead.
- Track hook-rate, meaning three-second view-through, per creative from launch
- Replace a creative when its hook-rate has decayed materially against its own first 48 hours, not when the calendar says so
- Keep the pipeline shipping continuously so a replacement is always ready. Admiral’s guidance is “systematic 3-second hook testing across 10-20 creative variants per sprint”, with sprint length set by your own production capacity
On raw volume: the industry-wide creative output figures often quoted in this context, including the 2,400-plus variants per quarter attributed to top gaming advertisers, are covered with their source in our analysis of the AI creative paradox, which also explains why that volume stopped being an advantage once everyone in the auction reached it.
Tracking, Attribution, and the Privacy-First Reality
TikTok measurement on iOS is a SKAN problem before it is a media problem. You need a clean conversion schema, a working MMP integration, and a model that keeps iOS post-install signal separate from Android event-level data. If your conversion mapping is wrong, your TikTok ROAS is meaningless and you will defund a working channel for the wrong reason.
Three guardrails worth keeping in place:
- Separate iOS and Android campaigns. Different attribution, different bidding behaviour, different creative response. This is also why the $5,000 budget floor does not stretch to cover both
- Map at least four SKAN conversion values to real LTV milestones, such as tutorial complete, day-1 return, first purchase, and a revenue threshold
- Reconcile MMP, TikTok Ads Manager and your own BI weekly. Persistent gaps above 15% mean something is broken in the SDK or postback chain rather than in the campaign
For the full attribution playbook under SKAN 4 and Privacy Sandbox, see our privacy-first UA guide.
Where TikTok Sits in the Channel Mix
TikTok is the highest-leverage creative-driven channel available to a small studio in 2026. It is not a UA stack on its own. Here is the allocation I would start a mid-tier mobile publisher on, stated as my own advisory position rather than as a benchmark.
| Channel | Starting share | Role |
|---|---|---|
| TikTok | 25% | Creative-led volume, strongest on casual and hybrid-casual |
| Meta | 35% | Breadth, midcore scale, the most mature optimisation |
| Google App Campaigns | 20% | Intent and Play Store adjacency |
| Apple Search Ads | 12% | High-intent bottom of funnel on iOS |
| Mediation networks | 8% | Incremental volume on hypercasual and hybrid-casual |
That sums to 100%, which the previous version of this table did not. Move within it by genre and LTV profile, but hold the discipline of not over-concentrating: auction shocks and policy changes hit single-channel advertisers hardest.
For how to structure that mix in detail beyond Meta and Google, see our UA channel diversification guide. For rewarded and offerwall inventory as a complementary layer, see our rewarded user acquisition playbook. For how UA fits the wider launch motion, our go-to-market strategy guide covers the sequencing, and our ASO guide covers the organic layer that compounds paid spend.
Who Is Actually on TikTok
The audience case for gaming is the part of this article that held up best under verification, though the figures come from a vendor blog rather than a measurement house.
AppFollow states that over 60% of TikTok users are mobile gamers, that they download 50% more games than non-users, and that users under 30 are about 40% of the base. That last figure is the one worth internalising, because it contradicts the reflex that TikTok is a Gen Z buy. The majority of the audience is over 30, which is where most paying mobile game cohorts sit.
AppFollow also notes that in January 2022, 80% of TikTok’s top Android advertisers were mobile games. That is a 2022 data point and I am labelling it as such rather than presenting it as a current state.
What We Could Not Verify
This article publishes fewer numbers than the version it replaces. Here is what I looked for and removed.
- “TikTok delivers 15-25% lower CPI than Meta.” This appeared four times in the previous version, including in a section heading and the conclusion. It is in neither Admiral Media nor AppFollow, and I found no primary source stating it. Removed entirely. Admiral does market “45% lower CPI vs benchmark” as a portfolio headline, but names no benchmark, so it is not usable either
- The US iOS CPI ladder of $0.60-$2.00 hypercasual, $2.50-$6.00 casual, $10-$22 midcore, $14-$35 RPG. The $2.50-$6.00 band is Admiral’s midcore band, republished as casual. The midcore and RPG tiers above it had no source at all. Replaced with Admiral’s actual scale, at Admiral’s actual labels
- D7 ROAS of 25-55% for hypercasual. Removed. Liftoff measured D30 ROAS for casual on Android at 15% across 2.4 billion installs, so that band would have a single week repaying between roughly one and a half and three and a half times what the whole first month repays. Admiral’s single casual puzzle client reports 2.8x D30 ROAS, which is a client result rather than a category average, and the distance between the two figures is exactly the reason single-client results should not be published as benchmarks
- “70% higher click-through and 159% higher engagement” for creator-led UGC. Attributed to “TikTok’s own data” with no link. Not present in either source. Removed
- “25-40% higher conversion” on playables. Not in AppFollow, which instead claims about five times more effective. Replaced with the source’s own wording and its own caveat
- Playable production costs of $3,000-$15,000. No published range I would cite. Removed
- “Micro-creators deliver 2-3x better ROAS than mega-influencers” and “casual openings retain at 2-3x the rate of cinematic ones”. Both plausible, both sourced nowhere. Removed, direction kept
- “Top studios route 25-35% of paid budget through TikTok.” Unsourced. The channel mix table above is now labelled as my advisory position rather than an observed market average
- A published creative fatigue half-life for TikTok gaming. Does not exist in a form I would cite. The 7-14 day cycle was removed along with the fortnightly refresh rule it contradicted
- Two sources were dropped from this article entirely. One was an SEO aggregator on our denylist, the other recycled figures without naming a dataset. Neither was load-bearing once the numbers above were removed
A note on what remains. Both Admiral Media and AppFollow are commercial parties writing about a service they sell, and neither publishes the underlying data. Admiral at least states the portfolio its figures come from, which is why its numbers carry the most weight here. TikTok’s help centre is the only genuinely primary source in the list, and it covers exactly one thing.
Conclusion
The studios winning on TikTok ads for mobile games in 2026 are not the ones with the largest budgets. They are the ones running a real creative operation: a Spark Ads creator programme with genre fit, continuous variant production read on hook-rate, hybrid casual-to-midcore hooks, clean iOS and Android separation, and a budget large enough to make a result readable.
Everything above that is a claim someone made about the channel. Before you plan against a TikTok number, ask which page states it and what portfolio it came from. On this channel, that question eliminates most of what is in circulation.
Want to pressure-test your TikTok UA plan against your LTV math? Get in touch, explore our mobile game consulting practice, or see what we cover.