Internal tooling for marketplace ops teams — surfaces broken listings, missing variants, and policy violations before they cause suppressions. Next.js + Supabase, Vercel-hosted.
The signals that predict a listing suppression — content gaps, detached variants, policy flags, stale inventory data — exist, but they're scattered across half a dozen marketplace reports that nobody reads until something breaks. Ops teams were discovering broken listings from sales dips, which is the most expensive possible way to find out.
Pull all of those scattered signals into one place and reduce them to a single number a team can rally around: a health score per listing. The dashboard ranks the catalog by risk, explains exactly which factors are dragging each score down, and alerts on the listings trending toward suppression — so the morning standup starts from a sorted worklist instead of a hunch.
Ingestion pipelines that normalize marketplace report exports into a single Supabase schema
A health-score model per listing that weights content completeness, variant integrity, policy flags, and inventory signals
Risk-ranked catalog views with per-listing factor breakdowns, so 'why is this score low' is always one click away
Alerting on downward score trends — the dashboard's job is to be read before the suppression, not after
Listing problems get surfaced and fixed before they become suppressions, and 'health score' became the shared vocabulary for catalog standups — the team stopped arguing about what to work on and started working the ranked list.
I take on a small number of builds like this each year — agentic systems and tooling for marketplace teams.
GET IN TOUCH