Creative when everyone has the same tools

Tool parity is the wrong thing to worry about. Two studios with the same generation stack, the same editing suite and the same testing platform will still produce creative that performs differently, and the gap will be wider than it was when tools were scarce, not narrower. The tools remove the production bottleneck. They do nothing about the judgement bottleneck, and judgement was always where the performance came from.

Plenty of creative leads will push back on this. Their experience of the past two years is that outputs across the market have converged, that the feeds look alike, and that the auction has started treating ads as interchangeable. All of that is true. The mistake is to blame the tools for it.

What actually converged

The convergence is in the inputs, not the outputs. Most teams feed their generation tools the same things: competitor ads pulled from intelligence libraries, the top-performing hooks of the past quarter, and a brief that says roughly "like that one, but ours". The tool does what you ask of it and produces a competent variant of a thing that already exists.

AppsFlyer's State of Gaming for Marketers 2026, published in January with Unity and Newzoo, put top advertisers at 2,400 to 2,600 creative variations per quarter. That number describes production capacity. It says nothing about how many distinct ideas were in the pool, and the evidence from the feeds suggests the answer is very few. UA Ledger argued in February that AI creative volume lowers the value of volume. The corollary is that it raises the value of whatever sits underneath the volume.

The mechanism is simple. When production is expensive, a bad idea costs a week of an artist's time and gets killed early by the budget. When production is nearly free, a bad idea costs nothing to produce, so it gets produced and tested, then allowed to compete for a share of the test budget alongside good ideas. Cheap production does not make teams more creative. It makes them less selective, unless something replaces the budget as the filter.

The auction punishes sameness, and then something worse

The first-order effect of convergent creative is well understood. Delivery models cannot distinguish between ads that look alike, so they fall back to price and the advertiser with the higher bid wins the impression. Creative stops being a lever and margin becomes the only one.

The second-order effect gets less attention. When a network's models see thousands of near-identical ads from dozens of advertisers, they learn that the creative is not predictive of user response for that genre. The models stop weighting creative features in their selection and lean harder on user-level signals and bid, which is a market-level change that persists after any individual team fixes its own creative. The team that finally produces something distinct is competing in an auction that has partly stopped looking.

This is why differentiation needs to be visible at the level the model reads: the first two seconds, the dominant motion, the colour field, the sound. A clever narrative twist at second eleven does not differentiate anything if the opening frame is the same as everyone else's. The brief has to specify difference where the model can see it.

Where the advantage moved

Three places, in order of how hard they are to copy.

The first is the brief. A brief that starts from a player observation the competitor does not have produces a different prompt and a different ad. That observation might come from your own game's telemetry or from support tickets. It might come from watching people play. Nobody else can generate from your data; everyone shares the tool, and nobody shares the input.

The second is the kill decision. With free production, the scarce resource becomes test impressions. A team that kills a concept after a small and well-defined read, then redirects the impressions, is running more distinct ideas through the same budget than a team that lets every variant run to statistical comfort. The skill here is knowing what an early signal looks like for your genre, and having the standing to act on it before the dashboard agrees.

The third is taste, and it is the hardest to talk about. Someone on the team has to be able to look at forty competent variants and say which three deserve budget, and be right more often than the average. Taste was always part of the job. It used to hide inside production, because the person making the ad also chose what to make. Now the two come apart, and taste has to be an explicit role.

An illustrative week

Take a mid-size puzzle studio with a creative team of four and a shared generation stack. Under the old process they might brief twelve concepts a month, produce each at two variants, and test all twenty-four. Under a brief-led process the same team might brief four concepts, each grounded in a specific player observation, produce each at eight variants that differ in the first two seconds, and test thirty-two. Same production cost, fewer ideas, far more information per idea, and a much sharper read on which observation actually moved behaviour. The numbers are illustrative, but the shape is the point: fewer sources of difference, more tested expressions of each.

The trade-off worth naming

A brief-led process is slower at the front end. Someone has to look at the data and talk to players before writing a thing that says more than "like the winner". Teams under pressure to hit variant counts will resist this, because the variant count is the metric leadership sees and the quality of the brief is not.

The way through is to make the brief the reported unit. Report concepts tested, concepts that beat control, and the observation each winning concept came from. Once the room can see that three winners this quarter came from one support-ticket theme, the argument for spending time on inputs makes itself. The tools are the same for everyone. What you point them at is still entirely yours.

Related archive reading

These articles provide related context and remain subject to their stated review status.

• The creative brief is a hypothesis document
• The misleading-ad arbitrage and why it still pays