120 ad variations a month, without burning your creators
A pipeline diagram, a brief template, and a scoring rubric — the engine we use across every D2C client.
The number that gets asked about most often when we describe our creative process is the 120 — as in, 120 variants shipped in a month, for a single account. The reaction is usually some version of "that sounds like it burns out your creators and floods the account with junk." It's a fair worry, and it's exactly what happens if you try to hit that number by just asking people to work faster. The volume only works because of what surrounds it, not because of raw output.
Why volume is the actual point
Ad platforms reward freshness structurally: the same creative shown to the same audience gets more expensive every week the algorithm has already extracted the easy wins from it, regardless of how good the ad is. A single hero video, however well produced, has a shelf life measured in weeks. If your production cadence is slower than your fatigue cycle — which for most brands running one polished campaign a quarter, it badly is — you're structurally guaranteed to watch CPMs climb no matter how good the bidding strategy is.
120 variants a month isn't about flooding the account. It's about matching production speed to the rate creative actually decays, so there's always a fresh angle ready before the current one runs out of runway. Most of those 120 will lose — that's expected and fine. The volume exists to guarantee there's always a next winner in the pipeline, not to make every asset a hit.
The brief template that makes volume possible
The thing that actually makes 120 a month sustainable rather than a burnout machine is a brief template that does most of the creative thinking upfront, once, instead of re-litigating it for every asset. Ours specifies the hook (the first 1.5 seconds, which is where almost all the drop-off happens), the core claim, the proof point backing that claim, and the specific placement it's cut for — a 9:16 Reels hook reads completely differently from an OEM in-device unit, and briefing for the placement up front avoids a lot of wasted editing.
Creators get a brief that's specific enough to shoot against without a back-and-forth, and editors get a brief that tells them exactly what the finished cut needs to hit. That specificity is what turns a slow, bespoke production process into something that can run every week without everyone re-deciding the strategy each time.
The scoring rubric — and why likes don't appear on it
Every variant gets ranked on two numbers: downstream revenue attributed to that specific asset, and a measured creative half-life — how long it holds efficient CPMs before fatigue sets in. Engagement metrics like likes, comments and shares don't factor into the ranking at all, because we've seen them repeatedly diverge from what actually pays; a highly engaging ad and a highly profitable ad are sometimes the same asset and often aren't.
Winners get identified within days of going live and scaled hard immediately — there's no month-end review that delays acting on a signal that's already clear. Losers get retired without ceremony, and whatever made a winning hook work gets folded straight into the next week's brief. That feedback loop, run weekly rather than quarterly, is what compounds: the pipeline gets measurably better at predicting winners the longer it runs, instead of resetting to zero every campaign.
