Replay runs applied AI against its own stock, with Recommend
Fashion demand moves faster than a production cycle. A theme worth building around this week is worth nothing in six. So brands either move fast and guess, or move carefully and arrive late.
Replay set out to move fast and be right. Applied AI reads live demand across Germany, Austria and Switzerland, matches it against the catalogue — and against the stock behind it — and turns the match into pages and communication. Twelve themed pages, live inside a month.
What changed
Twelve themed pages built and live within a month
Every page matched to product that was actually in stock
Matched communication for each one, run end to end by Recommend
A double-digit revenue uplift on the products featured
Stock is the constraint that makes it commercial
Plenty of tools will tell a fashion brand what is trending. Acting on it is a different problem: a theme matched to a product that is out of stock generates interest nobody can fulfil.
So the object being matched is not the theme and not the catalogue. It is the stocked item. Applied AI holds live demand against what Replay can actually ship this week, which is the difference between a communication idea and a commercial decision.
Speed was the brief, not a bonus
A theme that has to wait for a production cycle is no longer a theme. Replay's requirement was explicit: detect demand and convert it into assets before the window closes.
Twelve pages in four weeks, each built around a specific direction — bootcut, denim on denim, spring jeans — with matched communication for every one, targeted to the segment that direction belongs to.
Growth without discounting
The brief ruled out the usual lever. No discount mechanics, no waiting for a seasonal drop — visibility and conversion earned by being relevant rather than cheap.
Replay measured a double-digit revenue uplift on the products featured.
Every cycle informs the next
Each week's match is a record: which direction was worth building around, which products it led to, and what happened when it went live.
The following week does not start from a blank read of the category. It starts from what the previous ones produced — which is what turns a pilot into a way of working.