Thinking · Newsletter
Facebook ads for games: the 2026 playbook for Meta app campaigns, creative and measurement
Facebook ads for games in 2026: Meta's two app campaign setups, what 23 game brands run on Meta, a weekly testing cadence, and how MMP, SKAN and AEM fit.
TL;DR. Facebook ads for games in 2026 come down to one setup decision and a long run of creative decisions. Meta has moved app campaigns onto a streamlined Advantage+ setup that relies on machine learning rather than manual adjustments, so the ads you upload do most of the work of deciding who sees them. Across 12,311 Meta ads from 23 game brands read by Flixr Studio, 84 percent are video and 16 percent static images, and most of the ads launched in early October stopped running within a week. Measure with an MMP on top, SKAN and Meta's AEM side by side on iOS, and the Meta install referrer on Android.
What changed about Facebook ads for games
The levers a UA manager used to pull on Meta have mostly moved inside the machine. Meta's engineering team describes its ad retrieval stage as "selecting ads from tens of millions of ad candidates into a few thousand relevant ad candidates", and says that with generative AI "the number of ads creatives in Meta's recommendation systems is expected to grow significantly."
Our read: when the system chooses the audience from what the ad shows, the ad is the targeting. We made that argument at length in the networks know your users better than you do. For a game studio it means the budget conversation and the creative conversation are the same conversation.
A disclosure before the playbook: Flixr is a performance marketing agency that runs and makes Meta ads for games, so this is a practitioner's view, labelled as ours where it is opinion.
Meta ads for mobile games: the two campaign setups
Meta's documentation sets out two ways to build an app campaign: manual app ads and Advantage+ app campaigns. On 29 May 2025 Meta's developer blog said it had it "extended our new streamlined Advantage+ campaign setup for sales and app campaigns", adding: "This will also be the case for Advantage+ app campaigns (AAC) and manual app campaigns."
| Manual app campaign | Advantage+ app campaign | |
|---|---|---|
| Structure (Meta's docs) | 1 campaign, multiple ad sets, multiple ads | 1 campaign, 1 ad set, multiple ads |
| Who adjusts | "Reliance on manual adjustments." | "Reliance on machine learning adjustments." |
| Where we would use it | Controlled creative tests, new markets, a launch you need to read cleanly | Scaling proven concepts once measurement is trusted |
Advantage+ app campaigns were "formerly known as Automated App Ads." They "do not support operating system targeting, however SKAdsNetwork Advantage+ app campaigns will only target iOS14.5+ users." Meta's documentation lists their optimisation goals, including APP_INSTALLS ("Optimizes for people more likely to install your app") and VALUE ("Optimizes for maximum total purchase value within the specified attribution window").
If you buy through the Marketing API, note the breaking change Meta flags at v25.0: the old smart_promotion_type=SMART_APP_PROMOTION field can no longer be used to create these campaigns, and whether an app campaign is Advantage+ is shown in a field called advantage_state.
What game brands actually run on Meta: Flixr Studio data
Flixr Studio reads public Meta Ad Library ads from game brands with a model that watches each ad. The set used here is 12,339 ads from 23 game brands, launched between October 2024 and November 2026 (nearly all, 11,893, since July 2026), and watched since October 2026. Of those, 12,311 have a video or image reading. The brands lean towards social casino (8 of the 23), with strategy, tower defence and merge games among the rest, and no single brand is more than 15 percent of the read ads. These counts describe what brands chose to run, not what performed: competitors' spend and results are not public.
How the read ads were made:
| What carries the ad | Read ads | Share |
|---|---|---|
| Gameplay (real or invented) | 4,294 | 34.9% |
| CGI or animated story | 2,467 | 20.0% |
| Motion graphics or static | 2,204 | 17.9% |
| Staged skit, then gameplay | 1,994 | 16.2% |
| Creator over gameplay | 823 | 6.7% |
| Live action, talking head, character to camera, other | 529 | 4.3% |
Four more things the data shows:
- Video is the default. 10,340 of the 12,311 read ads (84.0 percent) are video and 1,923 (15.6 percent) static images. 2,349 of all 12,339 ads (19.0 percent) ran as dynamic creative, where Meta assembles the ad from several assets.
- Openings lean on people and plain gameplay. In the first three seconds, 29.6 percent of read ads open on a character or an emotion and 23.5 percent on plain gameplay; 10.6 percent lead with a reward or offer.
- Invented gameplay is common. 30.8 percent of read ads show some gameplay that is not the game's core loop as a new player meets it, and 21 of the 23 brands ran at least one such ad. 44.5 percent show only real core gameplay and 24.6 percent show none.
- Rotation is fast. Of the 3,732 ads Studio saw launch between 2 and 20 October 2026, 1,440 (38.6 percent) were still running at day 7 and 861 (23.1 percent) at day 14. How long an ad runs tells you a brand keeps paying for it, not that it wins.
Two more numbers worth knowing for planning: 12.6 percent of the 12,331 ads with an ad-level reading address lapsed players (win-back), and 85.1 percent are aimed at new players.
Our read of all this, for anyone planning Facebook ads for mobile games: game brands on Meta are running a volume business. They ship many ads, keep few, and spread the bets across formats rather than settling on one.
How to run Facebook ads for games: seven steps
- Wire measurement before you spend. Install your MMP's SDK, set up SKAN and AEM for iOS, and connect the Meta install referrer for Android. A campaign you cannot read cannot be optimised.
- Start manual, scale with Advantage+. Use a manual app campaign to test new concepts in a structure you can read, then move proven concepts into an Advantage+ app campaign. This is our working practice, not a Meta rule.
- Brief concepts, not cuts. Our rule of thumb is at least three distinct concepts per test, each on a different hook, so a flat result tells you something. The hook taxonomy in hooks and creative briefs is the starting list.
- Mix the formats the category already runs. Gameplay-led video, a creator or skit cut, a CGI or motion piece, and an image. Studio's data says 20 of 23 brands kept images in the mix, so do not drop them entirely.
- Ship on a weekly cadence. Most competitor ads in the October sample lasted a week or less, so a monthly batch leaves you slow. Our view on how often to refresh is in how often a mobile game should refresh its creatives.
- Read early on IPM and CPI, then on day-7 events. IPM and CPI tell you whether an ad earns the install; retention and purchase events tell you whether it earns the right player. Keep what Meta keeps spending on and retire ads as creative fatigue sets in.
- Prove the channel adds players. Platform-reported results include players who would have installed anyway. A holdout settles it; our method is in incrementality testing for ad creative.
Measurement: MMP, SKAN and AEM
On iOS, Meta's documentation describes Aggregated Event Measurement as "a protocol that allows for measurement of web and app events", and lets an app campaign set its attribution to AEM. It runs alongside Apple's SKAN. Segwise, an analytics vendor, notes that "AEM doesn't ingest any SKAdNetwork postbacks or conversion values", that with AEM "you start seeing modeled conversions just a few hours after they happen", and that "Meta began sending AEM signals to Mobile Measurement Partners on October 9, 2024."
On Android, Meta's install referrer covers more than Google Play's. Meta's docs say the Google Play Install Referrer "only provides information for same session click through installs", while the Meta install referrer also covers view-through installs and multiple-session click-through installs. You can set it up directly or through your MMP.
Our view: let the MMP be the referee across networks, use AEM for speed inside Meta, and use SKAN as the cross-network iOS check. When they disagree, the holdout in step 7 decides.
Where Flixr fits
We run Meta for game studios and make the creative it needs in house: gameplay, creator and UGC video, motion and playables, from the same brief. If you want the volume without building a team, our user acquisition for games page explains how an engagement starts. You can also see a UGC piece we made for a casual game, "Stop Scrolling", on the showcase.
Questions people ask
Are Facebook ads still worth it for mobile games? In our view, for most studios, yes: Meta's automation does the audience work, and 23 game brands in Studio's set are running thousands of ads there. The condition is creative supply: across the game brands Studio reads, most new ads were rotated out within a week, so plan for a steady flow of new concepts.
Should I use Advantage+ app campaigns or manual campaigns for my game? Use both for different jobs. Manual app campaigns give you multiple ad sets for controlled tests; Advantage+ app campaigns put everything in one ad set and rely on Meta's machine learning, which suits scaling concepts you have already proven.
What is the best ad format for Facebook game ads? There is no single best format, and the category does not behave as if there were. In Studio's read of 12,311 game ads, gameplay-led video was the largest group at 34.9 percent, but CGI stories, motion graphics, skits and creator cuts each took a meaningful share.
How do you track Facebook ad installs on iOS after ATT? With SKAN and Meta's Aggregated Event Measurement running side by side, reported through your mobile measurement partner. AEM gives faster modelled results inside Meta; SKAN gives Apple's privacy-preserving postbacks across networks.
How many creatives should a game run on Meta? Meta does not publish a number. Our rule of thumb is to launch new concepts every week and judge each one on installs and early events, because the game brands in Studio's sample replace most of their new ads within seven days.
Sources: Meta for Developers, "Advantage+ App Campaigns" (developers.facebook.com/documentation/app-ads/advantage-app-campaigns); Meta for Developers blog, Advantage+ campaign experience for sales and app, 3 June 2025 (https://developers.facebook.com/blog/post/2025/06/03/advantage-plus-campaign-experience-for-sales-and-app/); Engineering at Meta, "Meta Andromeda", 2 December 2024 (https://engineering.fb.com/2024/12/02/production-engineering/meta-andromeda-advantage-automation-next-gen-personalized-ads-retrieval-engine/); Meta for Developers, "Aggregated Event Measurement and Limitations of SKAdsNetwork" (https://developers.facebook.com/documentation/app-ads/SKAdNetwork-aem-and-limitations); Meta for Developers, "Meta Install Referrer" (https://developers.facebook.com/documentation/app-ads/meta-install-referrer); Segwise, "Meta AEM vs SKAN", 5 June 2025 (https://segwise.ai/blog/meta-aem-vs-skan-2025-ios-attribution); Flixr Studio, public Meta Ad Library game ads, queried 9 November 2026.