Returns (retour): why COD deliveries fail and what it costs

What a retour really costs

When a parcel comes back, the visible cost is the courier's return fee. The invisible costs are usually bigger: the packaging you paid for, the outbound shipping effort, one to two weeks of the product being unsellable while it travels, the risk it returns shelf-worn, the confirmation minutes already spent, and the ad money that bought the order in the first place. A returned order is the full cost chain of a delivered order — minus the revenue.

That is why return rate is the single most sensitive number in COD economics: it multiplies against everything else. Two operations with identical products, prices, and ad performance can end up with opposite profit signs on return rate alone.

Why deliveries actually fail

• Unreachable customers — the phone is off, wrong, or never answered on delivery day; the driver moves on.
• Refusal at the door — the customer changed their mind, found it cheaper, or is surprised by the total with the delivery fee.
• Weak confirmation — the order was never truly confirmed, so all of the above arrive unfiltered at the doorstep.
• Address problems — wrong commune, vague directions, or an area the courier serves poorly.
• Long transit — the more days between order and delivery attempt, the colder the intent; remote wilayas suffer most.
• Courier coverage mismatch — a courier that is excellent in the north can be slow in the south; every zone has its strong operators.
• Stop-desk expiry — the parcel arrived but the customer never picked it up within the holding window.
• Serial refusers — a small set of customers who repeatedly order and refuse; without shared history they hit you again per store.

The levers that pull the rate down

No single fix eliminates returns. The rate responds to a stack of small disciplines applied together:

• Confirm properly — the confirmation call is the cheapest return prevention that exists; restate the total with delivery fee so there is no doorstep surprise.
• Screen before shipping — block known bad phones and repeat refusers; review risky orders instead of shipping them blind.
• Collect structured addresses — wilaya and commune from a controlled list, not free text; bad addresses die at the form, not at the door.
• Offer stop desk deliberately — where home delivery is weak or the customer sounds hesitant, a stop desk converts a probable failure into a pickup.
• Match couriers to zones — route each wilaya to the courier that actually performs there, and re-check with data as volumes shift.
• Chase stalled parcels — a parcel sitting "in transit" too long is a return being born; ask the courier for a new attempt while the customer still cares.
• Segment the measurement — return rate per product, campaign, wilaya, and courier. One bad product or one bad zone can hide inside a decent average.

Measuring returns without lying to yourself

Return economics only work on delivered truth. Count an order as revenue when it is delivered and the cash is collectible — not when it is submitted, confirmed, or shipped. Attribute the return costs (fee, packaging, lost time) to the product and campaign that generated the order, so "winning" campaigns that quietly generate refusals stop looking like winners.

Averages are the enemy: a 10-wilaya operation with one terrible zone can look fine overall while burning money in that zone every day. Segment first, then decide.

How COD Scale attacks returns

• Fraud Shield — scores each incoming order using phone history, IP fingerprint, blocklists, and past outcomes across your stores; confirmed-bad orders are blocked, risky ones go to a review queue.
• Customer risk profiles — delivered-vs-returned history for the same customer across your stores, visible during confirmation calls.
• 58-wilaya address intelligence — structured wilaya/commune capture with home vs stop-desk choice at checkout.
• Suivi board — in-transit parcels that stall are surfaced with courier contact info, so you request a new attempt before the parcel expires into a return.
• Return-aware finance — per-order P&L carries return costs, and campaign verdicts (Scale / Hold / Stop) are computed from delivered orders, not submissions.

Frequently asked questions