Creative Performance

The algorithm buys the media.
Creative decides what it's worth.

On Meta, Google and Amazon the bidding is automated. What you say, to whom, in what format is the lever that's left — and it carries most of the result. We run it as a weekly loop with a memory.

Our approach

Creative is the lever.
We pull it every week.

Most accounts ship a handful of assets, let them run until frequency climbs, and read the platform's 'top performer' label as a strategy. The platforms now give you far better tools than that — creative cells, experiments, generative production, placement-level data — and almost nobody uses them as a system. We do. Every brief names the claim, the audience, the format and what it has to beat. Variants run in cells with enough spend to read. Winners scale, losers come out, and the registry remembers both, so the next brief starts from evidence instead of taste. You get creative that compounds instead of fatigues.

The registry

UGC · 'ships in two days'Broad · Reels
Won

+38% new-customer orders vs control · scaled

Studio · lifestyle heroBroad · Feed
Lost

CTR fine, orders flat · retired week 2

Founder to camera · price anchorRetarget · Stories
Won

Contribution per order up · promoted to prospecting

Product on white · feature listBroad · Feed
Lost

Frequency 4, CTR down a third · retired

Review carousel · 4.8★Broad · Reels
Testing

Cell live · read lands Friday

Illustrative, not a client registry. Every brief starts from what the registry already knows.

Why it's different

A loop with a memory.

Most creative testing forgets. A variant wins, scales, fatigues, and six months later someone briefs the same idea from scratch. The registry keeps every claim, audience, format and result, so what worked in March is the starting point in September — and what lost never gets paid for twice.

How the loop runs

Brief, produce, test, promote. Every week.

Brief

From the demand map and the product's margin, not a mood board. Which claim, which audience, which format, and what it has to beat.

Produce

Partner studios, UGC creators or the platform's generative tools, each working to a content standard per placement.

Test

Variants run in cells on a schedule, with enough spend to read. Winners are decided on orders and contribution, not on CTR.

Promote and retire

Winners scale; losers come out before the CPM tells you; the registry records both.

What we do and don't

Honest about the split.

We do

Direction, standard, testing, registry.

The brief, the content standard, the test calendar, the read, and the record of what won. The judgment that compounds.

We bring in

Production.

Studios, editors and creators we have worked with, under our brief, priced as their own line. Or the platform's generative tools where they clear the standard.

We don't

Charge you to be a studio.

No in-house production retainer. You pay for the work that's made and the judgment that directs it — nothing else.

We make profit predictable.

Commerce strategy, technology, operations and
growth economics, run as one system.