How to Keep HTS Classification Accurate as US Tariffs Keep Changing (2026 Guide for Brokers and Forwarders)

For most brokers and forwarders, HTS classification was treated as a setup task: classify a product once, store the code, reuse it. That model no longer holds. The tariff base now moves faster than any team can manually track, and a code that was correct one month can carry the wrong duty the next. This guide explains why keeping classifications accurate has become a continuous operating job rather than a one-time task, what that shift costs operators who are still running the old way, and how high-performing brokers and forwarders are restructuring to stay ahead of it.

The job changed, even when the work looked the same

A part gets classified. The code goes into the system. Six weeks later the same part ships again, the stored code is reused, and the entry is filed. On paper, nothing went wrong.

But the duty base underneath that code may have moved in those six weeks. A measure was added, a rate changed, an exclusion expired. The code looks right and the entry looks clean, but the duty calculation behind it is now wrong. Nobody sees it until a post-entry review or a penalty notice surfaces it.

This is the quiet shift in customs work. The task, assign a code, looks identical to what it always was. The job, keep every classification correct against a base that keeps moving, is now a different thing entirely.

The base is moving faster than manual work can track

Tariff classification has always mattered. What feels different in 2026 is the pace of change.

Over the last eighteen months, customs teams have had to navigate successive rounds of tariff adjustments, evolving trade measures, changing exclusions, and updates to the Harmonized Tariff Schedule itself. A classification decision that remains technically correct can still produce the wrong duty outcome if the tariff treatment behind it changes. 

That has increased the operational burden on brokers and forwarders. Teams are no longer managing classification against a relatively stable tariff environment. They are managing classification against a moving target.

The challenge is not understanding the rules. It is keeping pace with how quickly they change, maintaining an audit trail of decisions, and ensuring every shipment reflects the latest duty position available at the time of entry.

For customs operators, the risk is no longer isolated to unusual shipments or edge cases. It increasingly sits within routine, repeatable volume where stored classifications are reused day after day.

 Two numbers make the point.

The United States International Trade Commission (USITC) issued 32 revisions to the Harmonized Tariff Schedule in 2025, and another 10 by early June 2026. That is the published tariff schedule changing roughly every couple of weeks, sustained across eighteen months.

And the stakes behind each entry have risen sharply. U.S. Customs and Border Protection (CBP) collected more than $200 billion in tariffs, taxes and fees between 20 January and 15 December 2025, a record, driven by a wave of new trade measures. The agency has been explicit that it uses data analytics to flag under valuation and misclassification.

The operating reality is clear. The thing you classify against changes constantly, and getting it wrong is now expensive in a way it was not a few years ago.

Where the old model breaks for brokers

Brokers feel this first, because they carry the burden: classification research, entry quality, corrections and audit defence all sit with them. Three pressure points show up consistently.

Research time caps throughput. When the base keeps moving, more lines need a fresh look rather than a confident reuse, and senior broker time becomes the bottleneck on the whole filing queue.

Reclassifications and corrections eat the most expensive hours. The work that should be reserved for genuine judgement calls gets consumed by re-checking things that were already classified once.

And liability rises when entry data is stale. A correct-looking code built on an out-of-date duty position is exactly the kind of error CBP's analytics are designed to surface. The exposure is real, and it lands on the broker.

Where it breaks for forwarders

Forwarders feel a different version of the same shift. Margin in transport-only services keeps compressing, and customers increasingly expect landed-cost and compliance help as part of the relationship rather than as a separate service they go elsewhere for.

Industry research points the same way: in the 2026 CSCMP State of Logistics Report, produced with Kearney, classification and duty analysis rank among the most-requested services customers want from their forwarders. That is a commercial opening, but only for forwarders who can deliver fast, defensible classification at the volume their shipments demand, without throwing manual headcount at it. Doing it manually turns a differentiator into a cost centre.

Signs Your HTS Classification Process Is Becoming a Risk

‍Many customs teams don't discover weaknesses in their classification process until a correction, audit, or customer dispute forces the issue.

Some common warning signs include:

• Classifications are managed primarily through spreadsheets.
• Product descriptions arrive incomplete and require repeated manual clarification.
• HTS decisions are stored in email chains or individual broker notes.
• The same products are reviewed repeatedly because there is no trusted catalogue of validated classifications.
• Only a small number of people understand why specific codes were selected.
• Tariff updates are tracked manually across multiple sources.
• Duty calculations depend on individuals remembering recent policy changes.
• Teams spend more time checking previous decisions than     making new ones.

None of these issues creates a problem on its own. Together, they create an operating model that becomes harder to scale as shipment volumes increase and tariff complexity grows.  

What a continuous operating model looks like

The operators scaling through this have not just bought a faster tool. They have changed where and when classification happens, and what stays current on its own. Four moves matter.

Move the decision before the shipment, not at entry. Classification and duty exposure get resolved before freight moves, so landed cost and routing are decided with certainty rather than discovered at the border.

Catch missing data upfront. A line item described only as “50mm tape” cannot be classified correctly until you know whether it is aluminium or fabric, that difference changes both the code and the duty. A continuous model flags the gap and requests clarification early, instead of letting it surface as rework later.

Keep the catalog current automatically. Validated codes are stored per importer and reused but the model also watches for changes in the underlying schedule, duties and policy, updates the stored classification when something moves, and flags what changed and why. The reuse stays safe because the base it rests on is maintained.

Separate exceptions from volume. The majority of line items are routine and should beresolved automatically. Licensed brokers are reserved for the genuine exceptions, the ambiguous descriptions, the controlled goods, the judgement calls, where their expertise actually creates value.

How Ripple runs it

Ripple operates as an AI system embedded in existing broker and forwarder workflows. It reads commercial invoices as they arrive, no templates, no fixed layouts, then assigns compliant 10-digit HTS codes across every line, evaluates tariff exposure as part of the classification, and pushes validated codes and line data straight into CargoWise and other systems with no re-keying.

Classification doesn't stop at the code. For controlled goods, Ripple assigns export control codes alongside the HTS code and screens them against destination before shipment. And because it stores every validated classification against the specific importer it belongs to, repeat items match instantly, while the catalog is kept current as codes, duties and policy change, with the team alerted to what moved.

The result is the model above, running in production: routine volume cleared automatically, the catalog held current without manual upkeep, and human expertise pointed only at the exceptions that need it.

Proof in production: Pentagon Freight Services

Pentagon FreightServices handles complex oil and gas cargo, where classification delays previously held customer shipments pending approval. Running Ripple in live US production:

Multi-line invoices that took two to three hours now process in under two minutes. First-passa ccuracy runs at 95%+ across line items. Week-long classification cycles dropped to seconds, and the operation scales with the same team rather than adding headcount as volume grows.

The take away

The challenge in US customs operations is no longer the volume of work. It is the variability of the inputs and the pace of change in the base you classify against and the inability of manual or template-based systems to keep up with both, reliably, at speed.

For brokers, that shows up in throughput and liability. For forwarders, it shows up in margin and in the service customers now expect. In both cases, the operators pulling ahead are the ones who stopped treating classification as a task they complete and started treating it as an operation they run.

If you are classifying into the US right now, the question worth asking is whether your current setup keeps every classification correct as the base moves or onlyuntil the next revision.

If you're a broker or forwarder feeling this, book a 15-minute call with Ripple — no prep needed.

Frequently asked questions

Why is HTS classification described as "continuous" now?

Because the tariff base changes constantly. The USITC issued 32 revisions to the Harmonized Tariff Schedule in 2025 and 10 more by early June 2026. A code stored once and reused can become attached to an out-of-date duty position, so classification has to be maintained against the changing schedule rather than set once. 

How often should HTS classifications be reviewed?

There is no fixed review cycle. Classifications should be reviewed whenever tariff measures, duties, exclusions, product specifications, or regulatory requirements change. For high-volume importers, brokers, and forwarders, this increasingly requires continuous monitoring rather than periodic reviews. 

What happens if an HTS code becomes outdated?

The code itself may still be valid while the duty treatment behind it has changed. That can result in an entry that appears correct but carries an incorrect duty calculation, creating potential exposure to post-entry corrections, additional duties, and compliance issues. 

What's the risk of reusing a stored HTS code?

The risk is that changes to tariffs, exclusions, or trade measures are missed. A code that was accurate when it was first assigned may no longer reflect the correct duty position at the time of entry. 

How can brokers keep up with tariff changes at scale?

Most brokers combine expert review with technology that continuously monitors tariff schedules, duty changes, exclusions, and policy updates. The goal is to focus licensed broker time on genuine exceptions rather than routine monitoring. 

How do freight forwarders manage HTS classification at volume?

Leading forwarders increasingly move classification earlier in the shipment lifecycle, maintain validated classification catalogs, automate routine line-item processing, and reserve specialist expertise for ambiguous or high-risk products. 

Can AI automate HTS classification?

AI can automate a significant proportion of routine classification work when combined with business rules, importer-specific context, audit trails, and human oversight. The highest-performing models do not replace customs expertise; they allow customs experts to focus on the exceptions where judgement adds the most value.

Does Ripple integrate with CargoWise?

Yes. Validated 10-digit HTS codes and structured line-item data are pushed into CargoWise via API, removing manual data entry during entry preparation. 

How does Ripple keep classifications current?

Validated codes are stored per importer and reused for instant matching, while the system monitors changes to schedules, duties, and policy, updates affected classifications, and alerts teams to what changed and why.