Reallocated six merch roles to growth, ran 14x more A/B tests.

Manual merchandising and tier-1 CS were eating the entire growth team's capacity. We automated the routine decisions and freed the team for the work that actually compounds — assortment strategy, customer-cohort experiments, and the loyalty program.

What was actually broken.

The merchandising team of nine spent three weeks a month doing the same things: re-ranking PLPs based on last week's sell-through, writing 200-character copy for incoming SKUs, triaging tier-1 returns and CS tickets, and rebuilding the homepage hero for whatever Marketing decided was the next launch. The CMO wanted them running cohort experiments and assortment strategy. They had not had time to run an A/B test in 11 months. Tickets sat in the CS queue for 14 hours on average.

What we did, and what we deliberately did not do.

We did not replace the team. We replaced the routine work. A merchandising agent re-ranked PLPs nightly using last-7-day sell-through, margin, and inventory pressure, with copywriter human-in-the-loop on the first 30 days. A CS triage agent answered tier-1 returns, sizing, and tracking questions with the brand's voice (calibrated on 50K prior CS conversations) and escalated anything sentiment-flagged. The team that had been doing the work supervised the agents for the first 90 days and ran experiments the rest of the time. The hire freeze on the growth team was lifted four months in.

“The most expensive thing about manual merchandising was not the labor. It was the experiments we never got to run. We did fourteen times more tests in the year after FirmAdapt shipped, and that compounded into the AOV lift.”

What the numbers did, twelve months in.