Duty Drawback for E-Commerce: Returns and Re-Exports Are Refundable — Almost Nobody Claims Them

The Double Duty Problem in E-Commerce

Run the lifecycle of a typical DTC apparel import:

• You import $3,000,000 of garments at a 14% blended duty (MFN apparel rates run high): $420,000 duty paid.
• Returns run 20-30% in apparel e-commerce. Say a quarter of returned units are unsellable — wrong season by the time they're processed, hygiene rules, damaged packaging.
• Unsellable returns plus end-of-season overstock get one of three exits: shipped back to the overseas manufacturer, sold in bulk to a foreign liquidator, or destroyed.

Every unit taking one of those exits carried duty into the country and carried value back out — the protective purpose of the tariff was never served. That is precisely the scenario drawback exists for, and the refund is 99% of the duty attributable to those units.

The sector-wide blind spot is structural: e-commerce reverse logistics teams optimize for processing cost and recovery value, not customs recovery. The duty embedded in an exported return is invisible on a 3PL invoice. Nobody in the workflow is paid to notice it — so nobody does.

Which E-Commerce Flows Qualify

Map your reverse-logistics exits against the statute:

• Returns-to-vendor (RTV) shipped overseas. Unsellable or excess units shipped back to the foreign manufacturer or supplier. Qualifies as unused merchandise drawback — customer try-on and return doesn't constitute 'use' any more than retail handling does, provided the goods weren't genuinely put into service.
• Bulk liquidation to foreign buyers. Pallets of overstock or returns sold to liquidators in Latin America, Eastern Europe, or Africa. The export event qualifies; the price you sold at doesn't matter — drawback refunds the duty you paid at import, not a percentage of resale value.
• Destruction of imported inventory. Goods destroyed under CBP supervision (or per CBP-approved procedures) qualify the same as exports. Relevant for hygiene-restricted categories — cosmetics, intimates, supplements — where re-export isn't an option.
• Cross-border marketplace fulfillment. Inventory imported to US warehouses, then shipped to fill Canadian or Mexican orders. Those outbound units are exports of unused merchandise.

What doesn't qualify: returns resold domestically, donated goods (with narrow exceptions), and goods destroyed without proper documentation. The qualifying universe is the export-and-destruction slice — which at e-commerce return rates is bigger than most operators assume.

The Math at Realistic Return Rates

Worked example — mid-size DTC apparel brand:

• Annual imports: $3,000,000 at 14% duty = $420,000 duty paid
• Customer return rate: 25%; of returns, 40% are unsellable and exit via overseas RTV or foreign liquidation; plus 5% of total inventory exits as exported overstock
• Export-qualifying share of imports: (25% × 40%) + 5% = 15%
• Annual recovery: $420,000 × 15% × 99% = $62,370
• 5-year lookback: up to $311,850 on the first claim

Second example — electronics marketplace seller with Canadian fulfillment:

• Annual imports: $8,000,000 at 6% blended duty (with Section 301 on Chinese SKUs) = $480,000
• 18% of US-warehoused units ship to Canadian buyers; 4% exit as RTV/liquidation = 22% qualifying
• Annual recovery: $480,000 × 22% × 99% = $104,544 — every year, plus the retroactive sweep.

Note what's *not* in the math: profitability of the exits. Drawback pays on duty, so even returns you liquidated at 10 cents on the dollar refund their full import duty at 99%.

Your Platforms Already Keep the Records

The documentation burden is the standard objection, and in e-commerce it's weaker than anywhere else — because the sector runs on systems that log everything:

• Import side: entry summaries in ACE via your broker, exactly as for any importer. SKU-level landed-cost tools (which most operators already run) map duty to product.
• Return processing: your 3PL or returns-management platform logs every returned unit, its disposition (restock / RTV / liquidate / destroy), and timestamps.
• Export evidence: RTV shipments generate commercial invoices and carrier documentation; bulk liquidation sales produce export bills of lading; marketplace cross-border fulfillment produces shipment-level records from the platform itself.
• Matching: TFTEA substitution at the 8-digit HTS level means you don't trace an individual returned SKU to its original entry — apparel imported in March and exported returns of like-classified apparel in November match.

The build is a data join: disposition records × HTS classification × entry data. For operators on Shopify/Amazon plus a major 3PL, the inputs are exports away from existing dashboards. The destruction path needs more care — CBP notice requirements apply — which is a process change worth making given what hygiene-category destruction volumes are worth.

How to Start: The 90-Day Returns-Drawback Playbook

A realistic sequence for an e-commerce operator standing up drawback:

1. Weeks 1-2: Size it. Pull 12 months of disposition data from your returns platform. Compute the export-qualifying share and multiply: imports × duty rate × qualifying share × 99%. (Our 60-second estimator at /lp/drawback-estimate runs the same math with the 5-year multiplier.)
2. Weeks 2-4: Screen the lookback. Request your 5-year entry file from your broker or ACE. The 2021-2022 entries — imported at peak Section 301 exposure if you sourced from China — are the richest and the closest to expiring.
3. Weeks 4-8: Fix the destruction gap. If you destroy unsellable imports, route future destructions through CBP-compliant procedures so they start qualifying.
4. Weeks 4-12: File. Engage a drawback filer (contingency arrangements are standard — the network behind our estimator works that way), submit privilege applications for accelerated payment, and file the lookback claim oldest-entries-first on CBP Form 7551 through ACE.

From standing start to first filed claim is typically a quarter. From filing to first check, weeks — if accelerated payment privilege was requested up front.