"Our RTO is 25%" is the least useful sentence in D2C ops. Averages hide the structure, and RTO is nothing but structure: a handful of pincodes, one courier lane, a product category and a repeat-offender segment usually carry a wildly outsized share of the returns. You can't fix an average. You can absolutely fix a cluster.
Cut the number four ways
- By pincode and state: certain areas return parcels at multiples of your baseline — remote lanes, hostel clusters, specific localities. This is your map.
- By courier: the same pincode can perform differently across carriers. Sometimes the 'RTO problem' is a routing problem.
- By product: COD magnets — trend items, gifts, low-price impulse buys — run hotter than replenishment products people actually need.
- By customer history: the network's serial returners are few, identifiable, and responsible for far more than their share.
Graduate the response
The blunt instrument — turning COD off everywhere — punishes the innocent majority. Geography deserves a dial, not a switch: high-RTO pincodes go partial-COD (a small advance keeps honest buyers and filters the rest), the truly hopeless prefixes go prepaid-only, and everyone else feels nothing. Product gates work the same way — a collection that keeps coming back can be prepaid-only without touching the rest of the catalogue.
Test before you trust
Every gate is a conversion trade-off, which is why rule changes deserve a dry run: simulate the decision for a real phone, pincode and cart, and see exactly which verdict each rule produces — before shoppers meet it. Then watch the breakdown weekly. Geography shifts, couriers change, festivals distort everything; the map is alive.
RTO is not a weather condition to be endured. It's an address list to be managed.
Want these numbers for your store?
Try the RTO calculator or book a demo.