Nobody budgets for freight that pays the border twice. Every cross-border return does — once coming in with duties and brokerage attached, once going back with its own paperwork, fees, and a customs identity most systems never gave it. April is when the post-peak returns wave finishes washing through and the quarter's audit work begins, which makes it the right month to say the quiet part out loud. Returns is a profitability question, not just an experience question. And for anyone fulfilling from a Texas DC with Mexico-origin inventory, the profitability math has a border running through the middle of it that the returns dashboard doesn't show.
Start with the scale of the flow. Apparel return rates run high — multiples of general e-commerce [S-cite: S21 — apparel return rates 20–40% vs e-comm 5–20%] — and the unit economics of the reverse move run roughly double the forward cost [S-cite: S22 — reverse logistics ~2x forward unit cost]. Those are the consensus industry figures, and they're bad enough before the border gets involved. Now trace one cross-border unit. It entered with duty paid, or it entered through the de minimis channel. It rode a brokered crossing at Laredo, carrying fees you cataloged in your landed-cost model — hopefully. The customer sends it back. The reverse leg is the expensive direction, and now there's a second set of questions: does it return to the Texas DC or re-cross southbound to the origin facility? If it re-crosses, that's a customs event with its own entry type and its own fees. And the duty paid on the way in? Recovering it has a name — duty drawback — and a reputation: paperwork-heavy enough that a large share of eligible refunds simply go unclaimed [S-cite: share of eligible duty drawback unclaimed by e-commerce importers]. Money you already paid, sitting in a process nobody owns.
Here's the distinction that should reorganize your dashboard: return rate is not return cost. Two SKUs with identical return rates can have wildly different return economics — one is dense, domestic-sourced, restockable; the other crossed a border, costs more to move backward than forward, and re-enters inventory in a condition that triggers inspection labor. Managing returns by rate treats those SKUs identically. Managing by cost — fully loaded, border fees and drawback included — tells you one of them deserves a different policy entirely.
That's the undermarketed move: per-SKU, per-cohort returns policy. Return windows, fees, and methods set by what the return actually costs, not one blanket policy inherited from the loyalty team's last offsite. The cross-border column makes this sharper. A SKU whose reverse path includes a southbound crossing might rationally carry a shorter window, a restocking fee, or a keep-it-and-refund threshold — while your domestic bestsellers stay generous. None of this requires punishing customers. It requires knowing which generosity you can afford, SKU by SKU.
About the platforms, because this category has good ones. Loop, Narvar, and Happy Returns have each built real products on the experience side of returns — portals, exchanges, drop-off networks, label flows. That layer matters; a clumsy return flow costs repurchases. But notice what the experience layer structurally can't see: the duty paid at entry, the brokerage on the reverse crossing, the drawback claim nobody filed. That math lives in your TMS, your customs entries, and your broker statements — three places no returns portal reads. The result is a returns program that can be excellent at experience and blind at economics, reporting a healthy NPS on a flow that loses money per unit on specific SKUs. Both layers are real. Only one shows up in the demo.
This is also audit-quarter material, which is why April. The same invoice-audit muscle you point at carrier billing — the 1-to-3 percent of parcel spend that's typically recoverable [S-cite: S10 — audit-recoverable share of parcel spend] — has a returns-shaped blind spot: reverse-leg billing errors, duplicate accessorials on return labels, brokerage charged on entries that qualified for simpler treatment, and the drawback file that was never opened. Down at the Laredo end of I-35, the southbound paperwork is its own discipline, and the buildings up the corridor around Alliance that staged the returns wave all spring have the receipts — literally — to reconstruct what each return actually cost. Pull them while the wave is fresh.
The tradeoff, honestly: per-SKU returns economics is real analytical work, and the first pass will tell you things the merchandising team doesn't want to hear about a hero product. There's also a genuine tension — tightening a return policy on a high-cost SKU can dent its conversion, and that trade should be made with eyes open, not discovered in a quarterly review. The alternative, though, is pricing every SKU's returns risk at the portfolio average, which means your best products quietly subsidize your worst. That's not a policy. That's an accident with a spreadsheet.
Concrete next step: the returns P&L worksheet. One row per SKU cohort, columns for forward cost, reverse cost, border fees both directions, drawback status, restock condition, and net cost per return. Request the template and populate it yourself — or send us sixty days of returns data, customs entries included, and we'll populate it for you and flag the ten SKUs where a policy change pays back fastest. Either way, by May you'll know which returns you're funding on purpose.