creative is the new targeting: the audience is automated, so your creatives win the auction.
nielsen says creative drives 56% of sales, targeting ~30%. meta's andromeda just made that physical — the machine picks the audience now, so your mobile game's creatives are the only lever left. why creative is the new targeting.
ask a room of UA managers what drives sales and most will say targeting. nielsen's 2025 data says they're wrong by almost 2x: creative quality drives 56% of a campaign's sales lift, media and targeting about 30% — while marketers themselves attribute only around a quarter of the effect to creative. that gap was always true. what changed is that meta just made it physical. on mobile, the machine picks the audience now — so the only real lever left in your hands is the creative. this is why "who do we target" is the wrong question in 2026, and what to run instead.
what meta quietly took away from you
in december 2024 meta shipped andromeda, a new ads retrieval engine running on nvidia grace hopper chips, roughly 10,000x more efficient than what it replaced. in plain terms: it made advantage+ good enough that hand-picking audiences stopped adding value. fewer, broader ad sets now beat tightly-defined ones. the audience decision moved from your spreadsheet into meta's model — and it isn't giving it back.
here's the consequence that lands at the campaign desk: when you and a competitor both run broad advantage+ into the same cohort, the audience is identical — the machine picked it for both of you. so the auction is decided by the one variable you still control, the creative. that isn't a slogan. it's the mechanics of how the system now works.
the numbers, not the vibes
two data points settle it. first, nielsen's 2025 roi work: creative accounts for 56% of sales lift, media and targeting about 30%. second, meta's own testing under andromeda: advantage+ creative delivered around 22% higher ROAS, and one ad set with 25 creatives drove roughly 17% more conversions at 16% lower cost than the tightly-managed alternative (Meta's figures, via agency analysis — treat the exact numbers as directional). the platform is literally paying you to bring more creative and stop fiddling with audiences.
what this changes for a mobile game team
the job flipped. the old UA operator won by being clever about audiences — lookalikes, exclusions, layered interests. that skill is now worth close to nothing, because advantage+ does it and does it better. the new job is creative throughput: how many genuinely different, on-brand concepts can you get into the auction, and how fast can you read and iterate them. your mobile game's creatives are the whole ballgame now — you stopped being an audience strategist and became a creative operation.
the catch: diversity, not duplication
this is where teams misread the memo and burn budget. andromeda rewards creative diversity, not volume for its own sake. practitioner reads of the system put a creative-similarity threshold around 60% — cross it and the model treats your ads as redundant clutter and suppresses them (directional, but it matches what the engine is built to do). so fifty colour-swaps of one idea don't count as fifty creatives; the machine sees one and ignores forty-nine. the unit that matters is the genuinely different concept — a new hook, a new angle, a new mechanic shown — not the permutation. it's the same grounding problem: generic, samey creative got you nowhere before, and now it literally gets filtered out.
and it fatigues faster than it used to
the other half of the andromeda trade: because the model matches more precisely, it burns through your best audience faster. practitioner data suggests effective ad lifespan compressed from roughly 6–8 weeks pre-andromeda to about 2–4 weeks now (directional). the treadmill sped up. you don't just need more creative — you need more of it, more often, and you need to catch fatigue early, before the CPI climb, using the leading signals instead of waiting for frequency to spike.
the honest takeaway
creative is the last real lever meta left in your hands, and the data says it was always the biggest one anyway. the team that wins the next two years isn't the one with the cleverest targeting — that's automated and identical for everyone. it's the team that puts the most genuinely different, on-brand, spec-clean creatives into the auction, reads them fast, and refreshes before they fatigue. that throughput — generate on-brand statics and playables, read what's converting, iterate the winners, kill the rest — is the exact loop we built numu to run for game studios. tell us what you're launching.
