- Country: US and Canada
- Guide
- Topic: Post-entry audit
- Topic: Reasonable care
- Topic: Classification
- Topic: Duty stack
How do you evaluate tariff and trade compliance software?
Published · Updated
Short answer
Test it on your own entries, dated: the US tariff schedule has been revised 23 times since its 2026 Basic Edition took effect on December 31, 2025, most recently on November 10, 2026, so a tool is only as good as its handling of the entry date on each line. First decide which job you are buying for, because classification tools, entry-filing software and post-entry audit tools answer different questions. Then run a blind test on 20 to 50 real lines you have already checked, including at least one line that changed rate mid-year, one stacked line and one known overpayment, and score the tool on whether it got the duty right for each line's date, cited the instrument behind every rate, and flagged errors in both directions. The legal duty does not move to the vendor: under 19 U.S.C. 1484 the importer of record declares classification, value and rate using reasonable care, and keeps the records for five years in the US (19 CFR 163.4) and six in Canada.
TariffSense reads entry summaries, invoices and shipping documents in any format.
Which job are you buying software for?
"Tariff software" covers several products that answer different questions. Most evaluations go wrong by testing one job and buying for another. Start by naming the job.
| Job | Question it answers | When it runs | What it reads | What a wrong answer costs |
|---|---|---|---|---|
| Classification | What is the HTS or Canadian tariff code for this product? | Before the first shipment, or when a product changes | Product descriptions, specifications, rulings | Every future entry of that product carries the error |
| Duty calculation or landed cost | What will this line pay on this date? | Before buying or quoting | A code, an origin, a value, a date | A pricing or sourcing decision made on the wrong number |
| Entry filing | Prepare and transmit the entry or CAD | At import | Commercial invoice, packing list, bill of lading, the importer's product file | Rejected or wrong entries; the filer works under broker regulations or as a self-filing importer |
| Post-entry audit | Was each filed line right on its own entry date, and what is the money in each direction? | After filing, before and after liquidation or accounting | Filed entry summaries, CADs or broker exports, plus invoices and shipping documents | Missed refunds, unnoticed underpayments that keep accruing interest and penalty exposure |
| Trade-remedy and screening checks | Is the product in scope of an AD/CVD order, a SIMA measure, a surtax or a forced-labor rule? | At import and in audit | Codes, origins, producers, case numbers | Duties and deposits that were owed and not paid, or paid and not owed |
A broker usually owns filing. A compliance team usually owns classification policy and the audit. Finance owns the accrual. If the software is meant to serve more than one of these, run a separate test for each job.
Why does the entry date matter so much in a test?
Because the law changes between the order and the audit. The US International Trade Commission's release list shows the 2026 tariff schedule starting with the Basic Edition on December 31, 2025 and reaching Revision 23 on November 10, 2026. Each revision can move a rate, add a Chapter 99 number or change a note. A line entered in March and an identical line entered in September can owe different duty. A tool that answers "the rate for this code" without asking "on what date" is answering a different question from the one an audit asks.
Canada has the same problem in a different form: surtaxes, remission orders and SIMA measures start and end by order-in-council and Gazette notice, so the rate on a CAD depends on its accounting date.
So the first thing to test is not whether the tool knows today's rate. It is whether the tool can price a past line under the rules of that line's date, and show you which instrument it used.
How do you build a fair test set?
Use your own data, chosen so that you already know the answers. A vendor demo on the vendor's sample data tells you how the demo works.
- Pick 20 to 50 lines you have already checked. Take them from real entries, with the invoices behind them. Write down the correct code, origin, value, Chapter 99 or surtax lines, and duty for each before the trial starts.
- Include lines from both sides of a rate change. For US entries, a Section 232 metals line entered before April 6, 2026 and the same part entered after it is a good pair, because for most derivative articles the base moved from the value of the metal content to the full customs value on that date (Proclamation 11021). For Canadian accounting, a line on each side of a surtax start or remission date.
- Include stacked lines. At least one line with two or more Chapter 99 numbers, and one where a program switches another off, so you can see whether the tool knows the exclusions.
- Include known errors in both directions. One overpaid line and one underpaid line. A tool that finds only the refund is not an audit tool.
- Include your messy formats. A broker entry export, a scanned invoice, a spreadsheet from a supplier. The format your data arrives in is part of the test.
- Include a Canadian line if you import into Canada, with a CUSMA or other preference claim, a surtax or a SIMA duty.
- Run it blind. Give the tool the inputs, not your answers, and compare afterwards.
What should you score?
Score each line on a short, fixed list, and weight the items by what a wrong answer costs you, not by how impressive the screen looks.
| Criterion | What to look for | Red flag |
|---|---|---|
| Right duty on the line's own date | Correct rate, base and Chapter 99 or surtax lines for the entry or accounting date | Today's rate applied to an old entry |
| Citation behind every rate | The instrument, heading or order and its effective date, so you can check it | A rate with no source, or "per our database" |
| Both-direction findings | Overpayments and underpayments reported side by side, with the amount | Only refund opportunities shown |
| Line-level evidence | Which field on which document drove each finding | A summary score with no drill-down |
| Reads your formats | Broker exports, scanned and digital invoices, spreadsheets, CADs and Form 7501s | Needs your data re-keyed into a template |
| Update discipline | How quickly a new notice reaches the tool, and who reviews it before it goes live | No answer, or "automatically" with no review |
| Export and records | Findings and data exportable in a form you can keep for the retention period | Findings locked inside the tool |
| Handover to the filer | Clean, checked data and findings the broker can act on | A promise to file or "one-click" submission that bypasses your broker |
A tool that scores well on the first three and badly on the rest is a good engine with a weak workflow; the reverse is a good workflow wrapped around answers you cannot trust. Weight the first three most.
What questions should you ask the vendor?
These are the questions that separate a demo from a working tool. Ask for the answer in writing.
- "Price this line as of this entry date, and show the instrument and effective date behind each rate." Then change the date and ask again.
- "When the Federal Register, CBP or the Canada Gazette publishes a change, how long before it is in the tool, and who checks it?"
- "Show me a finding where the importer underpaid. What does the tool tell me to do with it?"
- "Which of my files can you read as they are: broker entry exports, scanned invoices, CADs, Form 7501s, spreadsheets?"
- "Where does a classification suggestion come from, and what do you show when the tool is unsure?"
- "Does the tool cover both US and Canadian entries, and does it handle the Canadian surtax and SIMA side, not only the tariff rate?"
- "Can I export all my data and every finding, with its source, at any time?"
- "Who in your team reviews regulatory updates, and what is their customs background?"
- "Where is my data stored, who can see it, and is it used to train anything?"
- "What does the tool not do?" A vendor that cannot answer this has not thought about where its answers stop.
What stays with you, whatever you buy?
The legal duty. In the US, 19 U.S.C. 1484(a)(1) says the importer of record, in person or by an agent authorized in writing, makes and completes entry using reasonable care, declaring the value, classification and rate of duty. Under 19 U.S.C. 1592(c), a negligent violation can cost up to the lesser of the domestic value of the goods and two times the lost duties, taxes and fees (or, where the violation did not affect duty, 20 percent of the dutiable value), and a grossly negligent one up to the lesser of the domestic value and four times the lost duties (or 40 percent of the dutiable value). Records relating to an entry are kept for five years from the date of entry under 19 CFR 163.4. In Canada, section 40 of the Customs Act requires importers of commercial goods to keep records, and section 2 of the Imported Goods Records Regulations sets the period at six years following importation.
What software changes is the evidence. A dated, repeatable check of every line, with the source for each rate and a record of what was corrected in each direction, is what reasonable care looks like in practice. A tool that produces that record helps; one that produces confident answers without sources makes the record worse.
Where does TariffSense fit?
TariffSense is customs compliance software for companies importing into the United States and Canada. It reads entry summaries, invoices and shipping documents in any format, checks classifications, duty rates and declared values against current tariff measures, and gives compliance teams, customs brokers and forwarders the tools to classify products and calculate duties with confidence. In the terms of this guide it is an audit, classification and duty-calculation tool: it reads Form 7501 and CAD lines with the documents behind them, prices each line on its own entry date with the instrument behind every rate, and reports overpayments and underpayments side by side. Filing stays with your broker. Run it through the same blind test as anything else; the post-entry audit pillar and its Canadian counterpart, auditing CAD/B3 entries after accounting, show what a line-level audit covers, and the duty calculator lets you price a single line before any trial.
TariffSense stays up to date with the latest trade regulations, and our internal team reviews and validates regulatory updates before they're published. However, the tool is intended for research and planning purposes. We recommend working with a licensed customs broker to prepare and file your customs entries.
Sources
- 19 U.S.C. 1484 Entry of merchandise (govinfo.gov, checked )
- 19 U.S.C. 1592 Penalties for fraud, gross negligence, and negligence (govinfo.gov, checked )
- 19 CFR 163.4 Record retention period (ecfr.gov, checked )
- Customs Act (R.S.C., 1985, c. 1 (2nd Supp.)), section 40 Importers' records (laws-lois.justice.gc.ca, checked )
- Imported Goods Records Regulations (SOR/86-1011), section 2 (laws-lois.justice.gc.ca, checked )
- USITC, Harmonized Tariff Schedule release list (2026 Basic Edition to Revision 23) (hts.usitc.gov, checked )
- Proclamation 11021, Strengthening Actions Taken To Adjust Imports of Aluminum, Steel, and Copper Into the United States, 91 FR 18201 (April 9, 2026) (federalregister.gov, checked )
- Updated : Re-checked every source; rates, caps and deadlines updated.
