The weekly returns monitoring dashboard: six numbers that flag fraud early
Returns abuse shows up in your numbers weeks before it shows up in your margin. The brands that catch it early all watch the same handful of metrics on a weekly rhythm. Here are the six that matter, what each one tells you, and what a bad reading looks like.
One: return rate by customer cohort, not blended. A blended return rate hides everything. Split it: first-time buyers, repeat customers, and flagged accounts. When the flagged cohort's rate climbs while the rest stay flat, you have an abuse problem, not a product problem. When first-time buyer returns spike, you have a sizing or description problem. The cohort split tells you which fire to fight.
Two: claim rate per hundred orders. Damage, loss, and wrong-item claims are the noisiest channel for organized abuse because a claim bypasses the physical return entirely. Track claims per hundred orders weekly. Honest brands sit in a narrow band; a sudden doubling is almost never a coincidence. Break it out by carrier and warehouse if you can, because internal leakage shows up here too.
Three: resalable percentage of returns. This requires your warehouse to grade condition, and it is worth the discipline. When the share of returns arriving in resalable condition drops, items are being worn, used, or damaged before they come back. A ten-point drop in resalable rate over a quarter is a wardrobing signal hiding in plain sight.
Four: concentration, or what share of returns cost comes from the top one percent of returners. In a healthy business this number is boring. When it climbs, a small group of accounts is doing a disproportionate amount of damage. This metric decides whether your problem is policy or people: diffuse return costs need policy work, concentrated return costs need account-level action.
Five: manual review queue outcomes. Of the returns you routed to manual review, what share got approved, warned, and denied. If approvals dominate, your threshold is too aggressive and you are annoying honest customers. If denials dominate, your threshold is too lenient and abuse is sailing into the instant-approve lane. The queue is a calibration instrument; read it weekly and adjust the threshold monthly.
Six: time from delivery to return by category. Fast returns in multiple sizes mean bracketing. Returns clustered at the edge of the window mean wardrobing. Track median days-to-return per category and watch for shifts. A category whose median slides from twelve days to twenty-eight did not change fit; it changed use.
Review all six every week, in the same meeting, with the same owner. The dashboard only works if someone is accountable for the readings. Most brands can build this from exports they already have: orders, returns, claims, and warehouse grading. The technology is a spreadsheet. The discipline is the product.