RTO — return to origin — is the single most expensive metric in LATAM COD. Every percentage point of RTO carries: the outbound carrier fee, the inbound carrier fee, the warehouse handling cost, the lost AOV, and the inventory cycle penalty. A 35% RTO operation is, on cash terms, often unprofitable even with strong gross volume.
This is the sequence Fufills runs to drag RTO down to the 10–15% band on accounts using the full execution stack. The order matters — earlier levers compound the later ones.
This is the largest single lever. An order that should not have shipped cannot be saved by any downstream fix. See the hard-gated confirmation deep-dive for the SOP.
Order of magnitude: hard-gating typically moves RTO from 30–40% range into the 20–25% range in the first 30 days, before any other lever fires.
LATAM carrier coverage maps do not look like US/EU coverage maps. A street in zone X may be on-route for Carrier A and 6 km off-route for Carrier B. Validating the customer address against the carrier's actual coverage polygon — not just the postal code — closes the next biggest gap.
The SOP:
- Geocode the buyer address against the lane's primary carrier.
- If the lane has multi-carrier coverage, evaluate against each carrier's polygon and route to the best-fit.
- If no carrier reaches the address inside SLA, the order is paused and the buyer is contacted with three options: pickup point, alternate address, refund.
Expect another 3–5 percentage points off RTO from clean addressing.
Single-carrier COD is brittle. The carrier has no competition on the lane, so its operational rigor decays. Multi-carrier execution means each lane has at least two active carriers competing on weekly performance: on-time delivery, RTO, cash-collection accuracy. Underperformers get demoted; outperformers get more volume.
We typically run a 4-week rolling window. A carrier that drops below threshold for 2 consecutive weeks loses volume the following week. A carrier that beats threshold for 2 consecutive weeks gains volume.
Order of magnitude: another 2–4 percentage points off RTO, with secondary effect on SLA adherence.
Not every confirmed order should ship in the next dispatch wave. An order confirmed for "Friday afternoon" should not be in the truck on Tuesday — that creates an extra holding step and an unnecessary delivery attempt before the customer is even home.
The SOP routes confirmed orders into dispatch buckets keyed off the confirmed window. Tighter windows ship later; looser windows ship sooner. This is the cheapest lever in the stack and shaves another 1–2 points off RTO.
A buyer who is not home on Attempt 1 is not necessarily a failed sale — they are a paused sale. The dispatch SOP runs:
- Attempt 1 — original confirmed window. If failed, code as
NOT_HOME. - Re-confirm via call center — same day if before cutoff, next business day otherwise. Buyer offered three windows.
- Attempt 2 — at the new confirmed window. If failed and no answer in re-confirmation, code as
UNREACHABLE. - Reverse logistics — package routed back to fulfillment warehouse, inspected, returned to picking inventory if SKU re-shelvable.
The trap most operators fall into: ungated Attempt 3, Attempt 4, Attempt 5. Past Attempt 2 + re-confirmation cycle, the marginal cost of each attempt exceeds the marginal revenue. The cap is the SOP's most important rule.
A surprising amount of "buyer refused" disposition is product appearance, not product quality. Packaging that looks cheap, COD-style brown box without branded labels, no visible payment receipt — these correlate with refusal at the door even when the product itself is what the buyer ordered.
The fix is mechanical, not creative: branded carton (not plain), per-country language label on the outer ply, payment receipt visible through a window panel, tamper-evident tape. None of this changes the product. All of it changes the refusal rate.
Order of magnitude: 0.5–1.5 points off RTO once the other levers have stabilized. Small, but cheap.
If a merchant can only fire one lever in the next 30 days, it is Lever 1. If they can fire two, add Lever 3. The rest compound on top.
Here is the cumulative effect Fufills sees on a typical mid-AOV LATAM book ($35–$70 SKU, 1,000+ orders/month, single country to start):
| Lever stack | Typical RTO band |
|---|---|
| None (ungated, single carrier, generic packaging) | 30–40% |
| Lever 1 only | 22–28% |
| Levers 1 + 2 | 18–24% |
| Levers 1 + 2 + 3 | 14–20% |
| Full stack (1–6) | 10–15% |
The 10–15% target is not a marketing claim. It is the operational floor Fufills has measured on accounts running the full stack with us across the 16-country LATAM footprint. Below 10% becomes possible on premium-AOV books with tight geographies — but is not the published target.
Lever 1 produces the biggest single move. Levers 2 and 3 require carrier relationships and routing infrastructure that take time to build. Most teams plateau between Lever 1 and Lever 2, which still places them at 18–24% RTO — better than industry average, but well above the 10–15% achievable with the full stack.
This is the gap Fufills exists to close: not a single fix, but the full sequence, operationalized as SOPs, run across 16 LATAM countries by one operator. The same standard everywhere.