What changed
The 2025 tariff regime pushed import-duty payments to record highs — a single state (Arkansas) shows $925M paid in duties Mar 2025–Mar 2026 (FACT, source). Duty drawback under 19 USC 1313 lets importers reclaim duties on goods later exported or destroyed, but it is chronically under-claimed by small/mid importers (inference). Higher duty deposits mechanically enlarge the recoverable pool.
Why now
Tariffs are at a multi-decade high, so the dollar value of each unclaimed drawback and each excess-deposit CBP protest is larger than it has been in years, and the 5-year statutory look-back means 2025's inflated deposits are claimable through ~2030. The pain is fresh and quantified in the press.
Converging signals
Three signals meet: (1) a documented surge in duties paid ($925M single-state, FACT); (2) a standing statutory refund mechanism (19 USC 1313 drawback + 19 USC 1514 protests) that is under-utilized by small filers (inference); (3) a paperwork/portal barrier (CBP Form 7551 in ACE, bills of materials, proof of export/destruction) that is exactly the founder's build shape.
Customer pain
Small/mid importers who export or destroy goods leave refunds unclaimed because drawback is procedurally hard (matching imports to exports, HTS-level line matching, evidence assembly, ACE filing). Customs brokers who could file it either don't offer it or outsource to specialist firms that skim a percentage.
Who pays
Primary BUYER: licensed customs brokers and drawback specialist firms wanting a faster claim-assembly tool (white-label, per-claim fee). Secondary BUYER: mid-size importers with in-house trade/compliance staff who file their own drawback (SaaS seat). BENEFICIARY (the importer getting the refund) and BUYER may be the same only in the self-file case.
Solved today
Specialist drawback firms (Alliance Drawback, Charter Brokerage, Comstock) run it as a managed service on a success-fee (percentage of recovery); enterprise trade software (Descartes, Thomson Reuters ONESOURCE) has drawback modules aimed at large filers. Small importers mostly do nothing.
Why current solutions are bad
Success-fee specialists take 15–30%+ of the recovery and prioritize large accounts; enterprise modules are priced and scoped for big filers. The small/mid importer and the generalist broker are underserved — no cheap, self-serve claim-assembly layer that undercuts the percentage model with a flat software fee.
Proposed product
A drawback eligibility + claim-package builder: (1) ingest ACE entry-summary and export data (ACE report exports / ABI data, or manual/CSV upload for MVP); (2) match import lines to export/destruction events and compute recoverable duty per line; (3) surface likely CBP protest opportunities where deposit rate exceeded final assessed rate; (4) generate the 7551 claim package (data + evidence checklist + supporting schedules) ready for a licensed broker or self-filer to submit in ACE. NOT filing-as-a-service.
MVP version
A screener + calculator: importer/broker uploads ACE entry-summary export and an export/BOM list; tool computes an estimated recoverable amount and produces a claim-readiness report and 7551 line schedule. Sell the estimate as the hook, the package builder as the paid product.
30-day build
Map the 7551 data model and drawback types (manufacturing, unused-merchandise, rejected-merchandise); build the import-to-export matching engine and duty-recovery calculator against sample ACE report formats; interview 8–10 customs brokers to validate white-label demand and pricing.
60-day build
Ship the claim-package builder (7551 schedules + evidence checklist + protest flagging); pilot with 2–3 brokers on real historical entries under a white-label agreement; add ACE report-format ingest robustness.
90-day revenue plan
Convert pilots to per-claim white-label pricing; add a self-file SaaS tier for importers with trade staff. First revenue via per-claim fees on real recoveries assembled during pilots.
Distribution path
Direct outreach to customs brokers and trade-compliance managers (LinkedIn, trade associations like NCBFAA, drawback/CBP communities); content marketing on 'how much drawback are you leaving on the table' with the free estimator as lead magnet.
Pricing hypothesis
White-label: flat $200–$750 per assembled claim package to brokers (undercuts the 15–30% success fee). Self-file SaaS: $300–$1,500/mo per importer seat by volume. Optional one-time recovery-audit report as a paid entry point.
Technical difficulty
Moderate-to-high: drawback rules and line-level matching are genuinely complex, ACE data formats are fiddly, and correctness matters (a wrong claim is a compliance problem). Tractable for a solo AI-assisted founder over 3–6 months but not trivial.
Legal / regulatory risk
Real but manageable: preparing/filing a drawback claim ON BEHALF of an importer requires a licensed customs broker. The founder must sell SOFTWARE to brokers/self-filers, not filing-as-a-service, OR partner with a licensed broker. As long as the product outputs a package the licensed party submits, the founder needs no broker license. Disclaim that estimates are not filing advice.
Platform dependency
Low platform-policy risk — CBP/ACE is a government system with no deplatforming owner. Dependency is on ACE data-export formats and any future ABI integration.
Founder fit
High. This is his proven shape — public money owed, a government portal (ACE), per-transaction monetization, and a filing/paperwork barrier — closely mirroring the FMCSA ELDT per-upload app. Caveats vs. a pure forced-filer play: drawback is discretionary (importers can opt not to claim), the subject matter is more complex, and incumbents exist. Software-not-service sidesteps the license constraint.
Breakout potential
Good: 50-state / national importer base, recurring annual claims plus a 5-year look-back backlog, natural expansion into CBP protests (Form 19), reconciliation, and broader trade-compliance data products. A defensible data/matching engine could become the small-importer drawback standard.
Final recommendation
PURSUE with validation-first discipline. Strong founder-fit and a genuinely large, tariff-inflated recoverable pool, but this is claimable (discretionary) money with real incumbents and complexity — not a deadline-forced filer class. Build the free estimator + 7551 package builder, sell white-label to brokers first, and stay strictly software-only to avoid the broker-license line. Validate broker willingness-to-pay in the first 30 days before committing to full ACE integration.
Next action
Interview 8–10 customs brokers and 3–5 mid-size exporting importers this week to confirm (a) they'll pay a flat per-claim/white-label fee vs. the incumbent success-fee model and (b) they can provide ACE entry-summary exports; in parallel, prototype the import-to-export matching + duty-recovery calculator on sample ACE data.