US tariff landscape October 2026

Published October 2026 · Data: USITC Harmonized Tariff Schedule export (build 2026-10-09T17:43:36Z, 23,193 lines) and competitor measurements taken 2026-10-11. All figures recomputed from the raw datasets by scripts/gen_report_oct2026.py.

This month’s report reads the United States tariff schedule the way a working importer should: not as a wall of numbers, but as a structured landscape with a predictable geography. We walk through the structural families of rate expressions across all 97 chapters, locate where the cost mass actually sits, quantify the column-2 premium facing non-NTR origins, dissect specific and compound duties that break naive calculators, and then turn to the commercial landscape — what competing trade-data tools charge, and how visible they are to AI engines that increasingly mediate how exporters find answers. Every claim carries its source inline.

Executive summary

The United States tariff schedule in October 2026 is a study in asymmetry. Of the 23,193 HTS8 lines in the current schedule, 8,504 (36.7%) enter duty-free under the general column, while only 16 of 97 chapters average 10% or more: a narrow tail of sectors carries almost all of the burden, and the single highest general rate reaches 350.0% on other (2401.10.65). [src: data/hts.json · build 2026-10-09T17:43:36Z USITC HTS REST export]

Duty-free access is not evenly spread either. Column 2 rates — the statutory ceiling applied to countries without normal trade relations — average 48.5% across the 11,253 lines where both columns carry a pure ad-valorem rate, versus 9.3% in column 1, and they exceed the general rate on 99.1% of those paired lines. [src: data/hts.json · build 2026-10-09T17:43:36Z USITC HTS REST export]

On the tooling side, the market we measured splits into pay-per-call transparency ($0 subscription floor) and monitoring subscriptions clustered between $29 and $130 per month (median $39), while AI-readiness signals such as llms.txt remain rare: only 1 of 6 measured vendors publish one. [src: data/competitors.json · measured 2026-10-11 exportsonar.com/pricing]

1. The shape of the schedule

Classifying every line’s general-rate expression yields five structural families. A majority of lines are pure ad valorem, but a meaningful minority — 2,285 lines, or 9.9% of the schedule — still expresses duty at least partly in physical units (cents per kilogram, dollars per litre, cents per pair). These mixed and specific duties concentrate in agriculture, apparel and chemicals, and they are precisely the lines where naive calculators misprice shipments, because the payable duty depends on declared quantity as well as value.

Two caveats matter for reading Table 1. First, 294 lines defer their rate to footnotes or to the rate “applicable to the natural juice”, “the garment in the ensemble”, and similar conditions; we exclude them from every average in this report rather than impute values. Second, our ad-valorem parser accepts only clean percentage expressions; 33 lines embed percentages inside longer conditional phrases (for example, rates applying to “the movement and case” versus “the battery”) and are counted as note-deferred under the strict caliber. [src: data/hts.json · build 2026-10-09T17:43:36Z USITC HTS REST export]

2. Where rates concentrate

Across the 12,077 purely ad-valorem lines the median general rate is 6.0% and the mean is 9.1% — a gap that signals a right-skewed distribution. Nine out of ten dutiable lines face rates at or below 15.0%. The workhorse band is 0–5%: fully 5,010 lines sit there, 41.5% of all strictly ad-valorem lines and more than any other single band. [src: data/hts.json · build 2026-10-09T17:43:36Z USITC HTS REST export]

The tail is where trade policy does its loudest talking. 116 lines carry 30–50% and 55 lines exceed 50%, including the 350.0% peak on other. For exporters, the practical takeaway is that sampling a typical rate is meaningless: the modal experience is near-zero duty, but a handful of chapters behave like protectionist walls, and knowing which side of that divide your HS code falls on changes landed-cost math by orders of magnitude. [src: data/hts.json · build 2026-10-09T17:43:36Z USITC HTS REST export]

3. Chapter league table

Averaging the strict ad-valorem lines by two-digit chapter exposes the fortress sectors. Tobacco (chapter 24) averages 256.6% — dominated by high specific-plus-percent compound rates that our strict caliber reads conservatively — followed by oilseeds (12) at 58.2% and footwear (64) at 14.8%. Apparel chapters 61 and 62 contribute 803 and 1,235 strictly ad-valorem lines respectively (of 958 and 1,451 dutiable lines overall), anchoring the double-digit tier. [src: data/hts.json · build 2026-10-09T17:43:36Z USITC HTS REST export]

Highest-rate single lines cluster in chapters 24 (tobacco), 12 (oilseeds) and 2 (meat), where the maximum ad-valorem expressions reach 350.0%, 163.8% and 26.4% respectively. Example: 2401.10.65 “Other”. [src: data/hts.json · build 2026-10-09T17:43:36Z USITC HTS REST export]

4. The column-2 premium

Column 2 of the schedule is the fallback wall: rates that apply to countries without normal-trade-relations treatment. Restricting both columns to purely ad-valorem expressions leaves 11,253 comparable line pairs. Their averages differ by a factor of more than five: 9.3% in column 1 against 48.5% in column 2. On 11,157 of the 11,253 pairs (99.1%) column 2 is strictly higher; equality holds on just 96. [src: data/hts.json · build 2026-10-09T17:43:36Z USITC HTS REST export]

The premium is not uniform — it is engineered. chapter 60, chapter 93, chapter 71 show the widest mean gaps once chapters with fewer than 20 paired lines are set aside, mirroring the historical structure of protection, while most industrial inputs (chapters 84–90) show modest premiums. Businesses sourcing from non-NTR origins therefore face not a flat surcharge but a product-mix-specific penalty of anywhere from a few points to more than a hundred percentage points. [src: data/hts.json · build 2026-10-09T17:43:36Z USITC HTS REST export]

5. Anatomy of specific and compound duties

The schedule’s physical-unit duties are dominated by weight: “per kg” appears in 1,621 of the 2,285 specific-or-compound lines. Per-piece and per-litre formulations follow far behind. This matters operationally: a calculator that ignores the unit column will silently understate duty on the 2,285 lines that mix money and percentage terms. [src: data/hts.json · build 2026-10-09T17:43:36Z USITC HTS REST export]

Among all dutiable (non-free) lines, the declared-unit column is led by kilograms (5,280 lines), numbers of items (2,612) and dozens (2,013); 2,186 dutiable lines state no unit at all. Compound lines combine both worlds — for instance $1.104/kg + 14.9% — and require quantity-aware computation to price correctly. [src: data/hts.json · build 2026-10-09T17:43:36Z USITC HTS REST export]

Three readings from the raw file show how differently the families behave. A specific line, 0102.29.40 “Other”, carries 1¢/kg; a compound line, 0402.29.50.00 “Other”, reads $1.104/kg + 14.9%; and the single widest column-2 spread runs from 5.5% general to 125.0% on 2903.11.00 “Chloromethane (Methyl chloride) and chloroethane (Ethyl chloride)”. Any tool that collapses these three expressions into one number is guessing — which is precisely where lookup tools differ in quality. [src: data/hts.json · build 2026-10-09T17:43:36Z USITC HTS REST export]

6. What competitors charge

We re-measured 6 competing trade-data tools on 2026-10-11, recording official list prices and free-tier terms from primary pages. Two pricing philosophies emerge. Pay-per-use vendors (SimplyDuty at £0.10 per call; Zonos at $2 + 10% of duties collected) keep a $0 subscription floor and monetize volume. Subscription vendors cluster between $29 and $130/month — median $39 — with the step-up priced on monitoring, alerts and batch capability rather than raw lookups, which are increasingly commoditized as free tiers. [src: data/competitors.json · measured 2026-10-11 exportsonar.com/pricing]

Quantitatively: 4 of the 6 measured vendors sell a subscription with a positive floor, 2 monetize purely per call, and the interquartile entry price spans $29–$130. The crawlable-surface spread is equally stark — 9,999 versus 42 indexable URLs, a 238-fold multiple that translates directly into how often each vendor can appear as an answer rather than an advertisement. Pricing power in this market correlates less with data breadth than with workflow depth: monitoring, alerting and export, the three capabilities no free calculator replicates. [src: data/competitors.json · measured 2026-10-11 exportsonar.com/pricing]

Free-tier generosity does not track price. The cheapest subscription in the set offers no free tier at all, while the reference baseline row — ourselves — keeps unlimited no-login US tariff lookup and monetizes monitoring instead. A cap of five queries per day (SimplyDuty) effectively gates sustained research workflows behind registration, whereas uncapped free lookup shifts the paywall to higher-value features: watchlists, change alerts, batch export. [src: data/competitors.json · measured 2026-10-11 exportsonar.com]

7. AI visibility of tariff tools

As buyers shift from search boxes to AI assistants, being quotable by machines becomes a distribution channel. Our measurements show the competitive field wide open on machine readability: 6 of 6 vendors allow all five major AI crawlers through robots.txt, yet only 1 publishes an llms.txt guidance file, and programmatic-SEO surfaces range from 42 to 9,999 indexed URLs — a ninefold spread in the raw material AI engines can cite. [src: data/competitors.json · measured 2026-10-11 exportsonar.com/robots.txt]

The gap between crawler openness (universal) and llms.txt adoption (1/6) is the clearest near-term opportunity in the set: publishing structured, quotable answers about tariff rates remains cheap, while the measurement tooling for tracking whether AI engines cite you is sold as a premium feature by adjacent vendors. [src: data/competitors.json · measured 2026-10-11 otterly.ai/pricing]

8. Free-trade programs soften the headline

The general column is the ceiling most importers actually see, but not the floor they pay. The special column grants at least one free treatment on 14,689 of 23,193 lines, and among strictly ad-valorem dutiable lines 100.3% carry at least one free-program pathway. Coverage is broad but not uniform: Oman (OM) appears on 14,689 lines and Bahrain (BH) on 14,658, while deeper-harmonization programs cover materially fewer lines. For an exporter holding eligible origin documents, the effective landed cost can be zero on lines whose general rate reads double digits — which is why origin management, not rate lookup, is where sophisticated shippers concentrate their effort. [src: data/hts.json · build 2026-10-09T17:43:36Z USITC HTS REST export]

The high-rate tail deserves its own look. Above the 90th-percentile cutoff of 15.0%, 1,201 lines remain, and 62 (356), 61 (282), 64 (101), 42 (48), 24 (43) account for the majority of them — apparel, footwear, leather goods and tobacco again. In other words, the same chapters that anchor the league table also dominate the extreme tail, so a sourcing team screening only average rates will understate its worst-case exposure precisely where it hurts most. [src: data/hts.json · build 2026-10-09T17:43:36Z USITC HTS REST export]

9. Implications for exporters

• Know your family before your rate. With 9.9% of lines carrying unit-based components, the first question is not “what percent?” but “percent of what, plus what per unit?”. Tools that answer only the first question misprice up to one line in eight. [src: data/hts.json · build 2026-10-09T17:43:36Z USITC HTS REST export]
• The tail dominates cost planning. The median dutiful line pays 6.0%, but chapters 24, 12, 64, 61 and 62 average 256.6–14.8%+. Product-level screening beats portfolio averages when choosing what to ship to the US. [src: data/hts.json · build 2026-10-09T17:43:36Z USITC HTS REST export]
• Column 2 is a product-mix penalty. A 39-point average premium, concentrated in consumer categories, can flip sourcing decisions; monitor it line-by-line rather than headline-to-headline. [src: data/hts.json · build 2026-10-09T17:43:36Z USITC HTS REST export]
• Free-tool economics favor the buyer — for lookups. Uncapped free lookup exists; what is scarce and worth paying for is change detection on 23,193 lines that refresh quarterly. [src: data/competitors.json · measured 2026-10-11 exportsonar.com/pricing]
• Being citable is underexploited. Only 1/6 vendors publish llms.txt while all allow AI crawlers; early movers capture the citation surface. [src: data/competitors.json · measured 2026-10-11 docs.zonos.com/llms.txt]

10. Methodology and data notes

Data sources. Tariff data comes from data/hts.json, our local archive of the official USITC HTS REST export (hts.usitc.gov), built 2026-10-09T17:43:36Z: 23,193 HTS8 lines across 97 chapters, each with description, declared unit, general rate, special-rate programs and column-2 rate. Competitor data comes from data/competitors.json, measured 2026-10-11 by direct inspection of vendor pages; every cell stores its own source URL, scrape date and verification status. [src: data/hts.json · build 2026-10-09T17:43:36Z USITC HTS REST export]

Calibers. “Ad valorem” statistics use the strict caliber: the general-rate field must be exactly N% or N.N%. Lines mixing percentages with money terms are classified compound; money-only lines are specific; prose-deferred rates are excluded from averages (counted as note-deferred). A sensitivity reading that extracts the first percentage token from those 33 conditional lines yields a mean of 3.5%, confirming that excluding them does not materially shift chapter averages downward.

Pairing rule. Column-2 comparisons use only lines where both columns parse as pure ad-valorem (n=11,253 of 23,193); this understates column-2 exposure because free lines and specific lines drop out, and we flag it as the conservative bound. Chapter-gap rankings require at least 20 paired lines per chapter.

Unverified cells. The competitor dataset marks cells that could not be measured with a sentinel (9999, legacy) or null; we render these as “n.m.” and never treat them as zero. One entry price (Easyship) currently carries third-party status from a review aggregator and is labeled as such in Table 5.

Reproducibility. Regenerate this page and verify byte-identity with:
python3 scripts/gen_report_oct2026.py && python3 -c "import hashlib;print(hashlib.md5(open('docs/reports/2026-10/index.html','rb').read()).hexdigest())"
Run twice; identical hashes confirm determinism. Every number and table above is emitted by the generator from the two JSON files — nothing is transcribed by hand.

Limitations. General-column analysis ignores special-program rates (free-trade partners, preferences), which appear in the special column of 14,689 lines and often reduce duty to zero for eligible origins. Competitor prices are list prices observed on one date and may omit enterprise discounts. Figures describe the schedule as archived on the build date; quarterly refreshes will shift counts slightly.

© 2026 ExportSonar research · Tariff data: USITC (public domain) · Competitor measurements: ExportSonar research · Generated by scripts/gen_report_oct2026.py