
Key Takeaways
Traditional duty drawback software relies on a slow, manual process that takes 9-12 months and leaves a high percentage of eligible refunds unclaimed.
Modern AI-powered platforms automate data processing and use algorithms to find the optimal import-export matches, reducing the timeline to just 10-15 days.
This algorithmic approach can recover 15–20% more in refunds compared to manual methods by evaluating every possible combination.
Zollback's platform combines this AI-driven optimization with expert compliance review and performance-based pricing, making drawback accessible to companies of all sizes.
If you've spent any time trying to navigate the duty drawback software market, you've probably felt the same frustration that one customs broker summed up bluntly: "The market blows." Between legacy desktop applications that haven't meaningfully changed in decades, enterprise platforms with opaque pricing, and a new wave of AI tools that inspire equal parts curiosity and skepticism, it's genuinely difficult to know what you're evaluating — let alone which solution is right for your operation.
This article breaks down the two dominant approaches to duty drawback software: the traditional model, anchored by industry stalwart DutyCalc, and a new generation of AI-powered platforms that are rebuilding the process from scratch. We'll compare them on speed, data handling, refund optimization, and accessibility — so you can cut through the noise.
DutyCalc Data Systems has been in the drawback space since 1988. Their flagship product, Drawback.NET, is a Windows-based application that a majority of U.S. drawback providers still rely on. The platform has a real track record: DutyCalc claims that no client has failed a CBP audit since 1981, and it has earned loyalty from specialized drawback brokers and Fortune 500 compliance teams. Tom Ferramosca of American Drawback Services calls it "extremely user friendly" and "head and shoulders above any others."
That reputation is earned — but it was built for a different era of trade.
The software itself is only one piece of the picture. The bigger issue is the workflow it sits inside, which looks roughly like this:
Manual data collection. Teams gather import and export records from scattered PDFs, commercial invoices, bills of lading, and ERP exports — often manually keying data into spreadsheets.
Line-by-line matching in Excel. The core of the work is human-driven: staff match import records to export records manually, one row at a time. It's time-consuming and can't computationally test every possible combination.
Desktop software entry. Matched data gets entered into Drawback.NET to prepare the formal claim.
Wait. The full cycle typically takes 9–12 months from data collection to claim filing.
For large enterprises with dedicated trade compliance teams and high refund volumes, this process is manageable — if not ideal. For everyone else, it's a bottleneck.
The case for DutyCalc and traditional providers is real:
Proven compliance track record. Decades of claims filed without audit failures is a meaningful credential in a highly regulated space.
Established relationships. Long-standing providers understand CBP workflows, drawback rulings, and the nuances of U.S. Customs and Border Protection (CBP) compliance.
But the limitations are just as real:
Slow by design. A 9–12 month timeline locks up cash that companies could be deploying elsewhere.
Manual matching leaves money unclaimed. Humans working in Excel can't evaluate every possible import-export combination under substitution rules authorized by the Trade Facilitation and Trade Enforcement Act (TFTEA). That means refunds are systematically underoptimized.
Inaccessible for mid-market companies. The high labor cost of manual processing makes it economically unviable for traditional providers to serve companies below a certain refund threshold. As one broker pointed out, the per-transaction cost model becomes "really brutal" for e-commerce businesses with high parcel volumes and thin per-shipment margins.
Opaque, often rising costs. Pricing is rarely transparent. Companies that budgeted for one provider discover the quote has changed — sometimes significantly — by the time contracts are signed.
Here's the insight that separates modern duty drawback software from legacy tools: duty drawback is not fundamentally an accounting task. It's a combinatorial math problem.
Every drawback claim involves matching a set of import records to a set of export records — across different Harmonized Tariff Schedule (HTS) classifications, drawback types, regulatory time windows, and substitution rules — to find the combination that maximizes the legal refund. A human doing this in Excel is making judgment calls and running partial checks. An algorithm can evaluate every permutation.
That reframing changes everything about how the software is built.
Modern platforms approach the problem differently at each stage:
Automated document ingestion. Rather than requiring manual data entry, AI models parse unstructured trade documents — PDFs, CSVs, commercial invoices, bills of lading, ERP exports — and structure the data automatically. No reformatting, no manual keying.
Algorithmic refund optimization. Instead of line-by-line Excel matching, proprietary algorithms evaluate all possible import-export combinations under regulatory constraints (including TFTEA substitution provisions) to find the permutation that produces the highest refund.
Continuous monitoring. The platform connects to a company's data streams and automatically identifies new eligible claims as transactions occur — converting drawback from a once-a-year project into an ongoing cash flow source.
Direct electronic CBP filing. Claims are submitted electronically to CBP via certified Automated Broker Interface (ABI) software, as now required. No third-party handoffs, no manual submission.
The result is a process that takes 10–15 working days instead of 9–12 months — and recovers meaningfully more on each claim.
Here's how the two approaches compare across the dimensions that matter most:
Processing time: 9–12 months, end to end.
Data handling: Requires manual data entry and document formatting. Labor-intensive and prone to human error.
Refund optimization: Manual, line-by-line matching in Excel. Systematically underoptimized — no tool can test every combination by hand.
Technology: Legacy Windows-based desktop application. No cloud integration, no API connectivity, no automation layer.
Accessibility: High per-claim labor costs make it uneconomical for providers to serve companies below certain refund thresholds. SMBs are typically turned away.
Pricing: Often opaque. Annual retainers, per-transaction fees, or minimum claim requirements are common.
Compliance: Relies on specialist expertise of the human operator. Audit trails reconstructed from spreadsheets.
Processing time: 10–15 working days.
Data handling: AI automatically parses and structures documents in any format — PDFs, CSVs, ERP exports — with no manual intervention.
Refund optimization: Proprietary algorithms compute all possible import-export matching combinations under regulatory constraints, recovering 15–20% more than traditional methods in head-to-head comparisons.
Technology: Cloud-based platform, built for continuous and automated operation. Designed to integrate with modern supply chain data systems.
Accessibility: Automation lowers the cost-to-serve, making it economically viable to serve companies of all sizes — including SMBs that traditional providers won't touch.
Pricing: Performance-based. You pay a percentage of what's actually recovered — no upfront fees, no retainers.
Compliance: In-house licensed customs brokers review every claim before electronic submission to CBP via certified ABI software.
It's worth acknowledging what the user research reflects honestly: there's real skepticism in the customs broker community about AI-powered drawback. As one broker noted, "I was approached for an AI Duty drawback program but not into it. I use a service center with old school programs." When new players are mentioned, the reaction is often cautious, with some brokers expressing uncertainty about their quality.
That skepticism is healthy. The duty drawback space has a long history of overpromising providers, Byzantine pricing structures, and software that looks modern but is just a UI wrapper on the same manual process underneath.
The meaningful test isn't whether a platform calls itself "AI-powered" — it's whether the underlying architecture actually replaces manual Excel matching with genuine algorithmic optimization, and whether the compliance layer is built in from the start (not bolted on as an afterthought).
We built Zollback specifically to address the limitations of both legacy software and first-generation AI drawback tools. A few things distinguish our approach:
Our co-founder Elena Zhao holds a Stanford PhD in combinatorial optimization — the specific branch of mathematics that powers our import-export matching engine. Before founding Zollback, she spent three years as a quantitative analyst at CBP modeling drawback claim workflows, and led the customs cost optimization team at Maersk Digital. This isn't AI applied generically to trade documents — it's optimization methodology built specifically for the drawback matching problem.
Our CTO Daniel Park previously built large-scale data integration pipelines for U.S. government agencies at Palantir, and architected the customs data platform at Descartes Systems Group used by thousands of brokers and freight forwarders. The document parsing and data infrastructure underpinning Zollback is built by someone who has done it before at institutional scale.
We are both a technology platform and a licensed customs brokerage. Every claim is reviewed by in-house licensed customs brokers — including former Charter Brokerage experts — before electronic submission to CBP via our certified ABI software. That compliance layer is not optional or third-party.
Our tiered model means you pay a percentage of the refund actually recovered, and that rate decreases as your total refund value increases. There are no upfront fees, no setup costs, no retainers. We earn only when you do.
Since launching, we've processed over $10 million in tariff refunds and crossed $1 million in booking revenue within the first year of our seed round. We're backed by Y Combinator (S24 batch) and Cherubic Ventures, and are SOC 2 Type 1 certified.
If you're evaluating duty drawback software — whether legacy or AI-powered — these are the questions that separate the real from the theoretical:
What is your optimization method? Do you use algorithms to compute all possible import-export matches, or does a person do the matching in Excel?
What is your average end-to-end processing time? From document upload to claim filing.
How do you handle raw documents? Can your system ingest unstructured PDFs and invoices automatically, or do we need to reformat data for you?
Who reviews claims before submission? Is there a licensed customs broker on your team, or is the review outsourced or skipped entirely?
Are you a certified ABI filer? Or do you hand off to a third party for CBP submission?
What is your pricing structure? Fixed fee, per-transaction, percentage of recovery, or some combination? Are there minimum claim thresholds?
The answers will tell you quickly whether you're looking at a genuine rebuild of the drawback process or a traditional workflow with a new coat of paint.
Billions of dollars in eligible tariff refunds go unclaimed in the U.S. every year. A high percentage of eligible import refunds are never claimed — not because companies don't qualify, but because the traditional process was too slow, too opaque, and too expensive for most businesses to pursue seriously.
Legacy software like DutyCalc played an important role in building the foundation of the U.S. drawback industry. But it was designed for a world of dedicated drawback specialists at large enterprises, not for the mid-market manufacturer paying significant import duties on components and trying to understand what they're owed.
If your company imports goods, exports finished products, or re-exports unsold merchandise — and you haven't taken a serious look at what duty drawback could recover — that gap is worth closing. A free eligibility assessment takes 30 minutes and will tell you whether the math works in your favor. No upfront commitment, no fees unless you recover something. Just a clear answer on what you could get back.
Duty drawback is a U.S. Customs and Border Protection (CBP) program that provides a refund of up to 99% of customs duties paid on imported goods that are later exported or destroyed. This program allows companies to recover tariffs and fees, significantly lowering their net duty spend on exported products.
AI-powered duty drawback software automates the refund process by ingesting trade documents in any format and using algorithms to identify the optimal matches between imports and exports. Instead of manual Excel matching, it computationally evaluates every possible combination to maximize your legal refund under CBP substitution rules.
AI-powered drawback is faster because it eliminates manual data entry and matching, which are the main bottlenecks in the traditional process. Automating document parsing and using algorithms to find matches reduces the end-to-end timeline from a typical 9–12 months down to just 10–15 working days for claim preparation.
The amount you can recover depends on your import duties paid and export activities. Our AI-driven approach can potentially recover 15–20% more in refunds compared to manual methods by ensuring no eligible matches are missed. A free eligibility assessment can help estimate your potential refund based on your unique trade data.
Any company that imports goods into the U.S. and later exports them, or uses imported parts to manufacture products for export, is likely eligible for duty drawback. This includes manufacturers, distributors, and e-commerce businesses, even if the imported and exported goods are not identical but are commercially interchangeable.
Yes, when operated correctly. Zollback is a licensed customs brokerage, and every claim our AI platform generates is reviewed by in-house licensed customs brokers before electronic submission to U.S. Customs and Border Protection (CBP). This ensures full compliance with all drawback regulations.
Zollback uses performance-based pricing, meaning you only pay a percentage of the refund we actually recover for you. Unlike traditional providers who may charge retainers or setup costs, our model has no upfront fees. Our success is directly tied to yours, making drawback accessible to more businesses.