How does nexus work when you sell on Shopify and Amazon at the same time?
At $20M to $80M on Shopify or Shopify Plus, a brand running direct alongside Amazon, Walmart Marketplace, and TikTok Shop is operating four parallel tax flows on top of the same nexus map. The brand collects on direct. Each marketplace collects on its own channel. FBA inventory moves independently of either revenue stream. The state cares about all of it, but on different terms in different places.
Three rules govern the picture. First, every marketplace channel runs under the destination state's marketplace facilitator law. In every state with a sales tax, plus DC, the marketplace is the statutory collector on its own transactions once it exceeds the state's marketplace-side threshold. Every major marketplace (Amazon, Walmart, eBay, TikTok Shop, Etsy) clears those thresholds in every state with a statute. Second, the seller's own economic-nexus threshold is a different question. It asks whether the brand has to register and collect on direct-channel sales, and in some states the facilitated volume counts toward that calculation even though the marketplace already remitted the tax. Third, physical nexus from FBA inventory operates independently of either of the first two and creates a registration obligation on the seller's direct channel regardless of marketplace remittance.
The defining multi-channel tax reality at this scale is that the brand owns the seams between channels. The marketplaces handle their own collections cleanly. Direct-channel calculation runs through Shopify Tax or another engine. Where the model breaks is between them: in the threshold math the marketplaces do not run, in the physical-nexus footprint Amazon creates but does not register against, and in the return that has to reconcile both channels to the state's satisfaction. The rest of this guide walks the three seams.
How marketplace and direct sales aggregate for the seller's own threshold by state
The most common multi-channel nexus mistake at $20M to $80M is a brand excluding Amazon-facilitated volume from threshold tracking because Amazon remits the tax, then missing a registration trigger in a state where facilitated sales count toward the seller's own threshold. The marketplace covered the transaction. The seller still needs to register on direct-channel grounds because the state is measuring against combined volume. By the time the brand realizes it, the unregistered period is six to nine months long and direct-channel tax was uncollected the entire time.
The split as of 2026, drawn from each state's economic-nexus statute and the marketplace-facilitator codification that runs alongside it:
| Treatment | States | Effect on combined threshold |
|---|---|---|
| Marketplace volume included | AK (ARSSTC), CA, CT, DC, HI, ID, IA, KS, KY, MD, MI, MN, MO, NE, NV, NJ, NY, NC, OH, RI, SC, SD, TN, TX, VT, WA, WV, WI | Combined direct + marketplace volume counts. A brand at $60K direct + $50K facilitated in WA is at $110K and over the WA threshold. |
| Marketplace volume excluded | AL, AZ, AR, CO, FL, GA, IL, IN, LA, ME, MA, MS, NM, ND, OK, PA, UT, VA, WY | Only direct-channel volume counts. The same $60K + $50K split in FL is at $60K direct and under the FL threshold. |
Same channel mix, same dollars, opposite registration answer across a state line. The implication shows up clearly in a single worked example. A brand with $80,000 in direct-channel Shopify sales into Florida and $50,000 in Amazon-facilitated sales into Florida sits at $80,000 against Florida's threshold (excluded) and under the $100,000 line. The same brand into Washington sits at $130,000 against Washington's threshold (included) and over the $100,000 line. Treating Florida and Washington with one rule produces a wrong answer in at least one of them.
Three further wrinkles compound the math. California, New York, and Texas measure the seller threshold against $500,000 rather than $100,000, so the inclusion question matters at a higher revenue band but produces the same kind of mistake. New York is an AND test: $500,000 in cumulative gross receipts and more than 100 transactions, measured over the immediately preceding four sales tax quarters. Connecticut also uses an AND test at $100,000 and 200 transactions over the 12-month period ending September 30. Most dual-threshold states use OR. Reading the conjunction wrong has produced real audit exposure.
The operational pattern that holds at this scale is to tag every order at intake with its channel (direct, Amazon, Walmart, TikTok Shop, eBay) and its facilitated-or-not flag, then feed both flows into one state-by-state tracker that applies the inclusion rule per state. A spreadsheet that lumps Amazon into the threshold count everywhere over-registers in the 19 exclusion jurisdictions. One that drops Amazon entirely misses triggers across the 28. TaxCloud's threshold-monitoring view ingests Shopify and Shopify Plus orders alongside Amazon, Walmart, and TikTok Shop settlement through the reporting API, with the per-state marketplace-inclusion rule embedded so the same data set produces the right answer in CA, WA, FL, and PA at the same time.
Why FBA inventory creates direct-channel obligations Amazon's remittance does not cover
The brand discovers at audit that FBA inventory created physical nexus in a state where Amazon was already handling the marketplace tax. Amazon covered every Amazon-channel transaction in that state cleanly. The state's position is that the brand owed its own direct-channel return there from the day inventory first arrived. The brand never registered, because it read Amazon's collection as covering the state. The two obligations operate in parallel.
The framework
Physical nexus is the obligation created by tangible presence: inventory, employees, contractors, property. South Dakota v. Wayfair, Inc., 138 S. Ct. 2080 (2018) authorized economic nexus on top of physical nexus, not in place of it. Every sales tax state still asserts physical-presence rules through "engaged in business" provisions (Cal. Rev. & Tax. Code §6203; Tex. Tax Code §151.107; RCW 82.08.052). Marketplace facilitator laws move collection on facilitated transactions to the platform. They do not eliminate the seller's separate presence-based registration obligation. California's marketplace facilitator statute (AB 147, Stats. 2019, ch. 5) requires Amazon to collect on facilitated sales but does not remove the presence-based registration requirement for sellers with FBA inventory in the state.
The FBA mechanic
Amazon places inventory across fulfillment centers it chooses, optimizing for delivery speed. The seller does not direct which states receive product. A brand at meaningful FBA volume typically has inventory across 15 to 20 states simultaneously, often within weeks of enrolling. California, Washington, Texas, and New York each assert that FBA inventory in their fulfillment centers creates physical nexus regardless of whether the seller controlled placement. The Pennsylvania Commonwealth Court ruled the opposite in Online Merchants Guild v. Hassell, No. 179 M.D. 2021 (Pa. Cmwlth. Ct. Sept. 9, 2022), holding that FBA placement alone does not create physical nexus in Pennsylvania where sellers lack control over where Amazon stores their goods. The PA DOR appealed the decision to the Supreme Court of Pennsylvania. The holding binds Pennsylvania courts only. No federal circuit and no other state has adopted the voluntary-control theory; the rest of the FBA footprint remains contested or actively asserted. The current state-by-state postures are mapped in Does Amazon FBA inventory still create nexus in 2026?.
The lookback math. Economic nexus is bounded by its measurement window. Physical nexus is not. When FBA inventory arrives in a state before economic nexus crosses, the state can assert physical nexus back to the date inventory first landed. A brand that crossed California's $500,000 threshold in Q3 2024 has economic-nexus exposure from that period. If FBA inventory first landed in California in January 2021, California can assert physical nexus back to January 2021, expanding the audit window by roughly three and a half years on the direct channel at combined state-and-local rates that can exceed 10% in Los Angeles County and San Francisco. The brand's audit-defense position depends on the Amazon Inventory Event Detail Report by fulfillment center, filtered by state, which establishes the date inventory first arrived. Seller Central retention on that report is approximately 18 months, so the data the brand needs at audit is the data it archived quarterly and stored outside Seller Central.
For brands operating that footprint, TaxCloud handles the registrations triggered by FBA placement across the state grid and produces the per-state documentation trail (registration effective date, filing history, nexus position log) auditors ask for on day one.
How to reconcile direct and marketplace sales on the return in non-marketplace states
The reconciliation that breaks: direct and marketplace sales aggregate differently for the threshold than they do for the return, and brands that use one number for both produce a filing that does not tie out. The threshold count answers "do we register here." The return answers "what do we report on the form we already file because we are registered here." Different question, different inputs, different math.
The return mechanics. Almost every state with a sales tax requires a registered seller to file periodic returns reporting gross sales in the state, including periods when the marketplace collected the tax on a portion of the volume. The return reports the seller's total in-state sales, breaks out the marketplace-facilitated portion on a separate line, and treats marketplace-collected tax as an offset that zeros out the seller's remitted amount on the facilitated volume. The seller remits tax only on the direct-channel portion. Florida and Texas have explicit lines for marketplace-facilitated sales on their remote-seller returns. Most states follow a similar structure. Filing nothing in a state because the marketplace collected on a portion is a non-filer position, not a $0-due position, and triggers separate penalty exposure even when no tax is owed on the net.
The four-channel reconciliation. At month-end the operating-side question is how the brand assembles the return inputs from systems that each export their tax data in their own schema. Amazon publishes a tax-collected-by-state and gross-marketplace-sales-by-state breakdown in the Sales Tax Report inside Seller Central. Walmart Marketplace exposes the same data on a different cadence and with different field names. TikTok Shop's seller-side reporting is newer and still maturing. Each marketplace's settlement export is what the seller's return treats as support for the offset line. Direct-channel data comes from Shopify Tax (or whichever calculation engine the brand uses) and reports tax collected by jurisdiction for the period. Normalize all four into a common state-by-state schema before any return preparation begins.
The reconciliation that fails. Brands at this revenue band typically fail one of three ways. First, they file gross sales from a Shopify-only dashboard and omit Amazon entirely, producing a return that under-reports gross by the size of the Amazon channel and matches the marketplace-offset disclosure to nothing. Second, they file gross sales as the sum of all four channels and forget the offset line, producing a return that double-counts marketplace-collected tax (the marketplace already remitted; the seller is now remitting on top). Third, they file the return correctly but discard the marketplace settlement export after filing, so the audit-defense documentation for the offset line is gone two years later when the state opens the period.
The audit-defense pattern. Retain the marketplace settlement export alongside the return for every period filed. Document the methodology in a per-state schedule: which channels contributed to gross, which were treated as facilitated, what offset was claimed, and the source export that backs each line. For brands registered in the 24 SST states (23 full members plus Tennessee as an associate member), TaxCloud handles consolidated SST filing across all 24 in a single process, including the marketplace-offset disclosure line on each state's return. The 21 non-SST states with a sales tax (plus DC) file individually. The reconciliation pattern for the rest of the multi-channel footprint, including the Shopify-direct-to-Amazon seam at month-end, is in Multi-channel sales tax orchestration and Cross-channel sales tax reconciliation view.
The monitoring view that keeps both channels visible at month-end
The brand that runs into the multi-channel wall typically has a clean Shopify tracker, a clean Amazon settlement export, and no view that ties them together by state. The threshold math runs on one of them. The return math runs on the other. The physical-nexus footprint from FBA lives in a third system. Auditors do not accept three disconnected views as a defensible position.
The defensible monitoring view is one dashboard, refreshed at least monthly, that ingests five inputs and applies state-specific logic on top.
- Shopify or Shopify Plus direct-channel orders by state and tax period. For Shopify Plus brands running B2B order flows, the B2B transactions pull separately and need to be flagged as wholesale at intake so per-state wholesale-inclusion logic applies correctly.
- Amazon Seller Central settlement and tax reports by state. Pull both the gross marketplace sales by state and the tax-collected by state, monthly. Tag every order with the facilitated flag preserved.
- Walmart Marketplace, TikTok Shop, eBay, and Etsy settlement exports. Same shape as Amazon, different schemas. Normalize before reconciliation.
- FBA Inventory Event Detail Report by fulfillment center, archived quarterly. This is the document that establishes when physical nexus began in each FBA state. Standard Seller Central retention is approximately 18 months; the brand's own archive runs longer.
- HRIS data on remote hires by state. Every W-2 employee in a state generally creates physical nexus there, independently of FBA, marketplaces, or direct-channel volume. Hire dates establish the nexus trigger date.
On top of those inputs, the dashboard applies four per-state rules: the marketplace-inclusion flag (included vs. excluded for the seller's threshold), the measurement-basis flag (gross vs. taxable vs. retail per state), the AND vs. OR conjunction (NY and CT are AND), and the measurement window (calendar year, current or previous, rolling 12 months, or preceding four sales tax quarters). The dashboard's primary output is two columns per state: the seller's threshold position and the physical-nexus status. Proximity alerts fire at 75% of the dollar threshold for advance warning, and registration triggers fire at 90% with a 30-day operational lead time so registration is in place before the brand books the order that crosses. The cadence by revenue band (quarterly at roughly $10M, monthly at roughly $30M, monthly plus event-driven above roughly $70M) is in Nexus monitoring at scale.
The reader here is past wondering whether multi-channel volume changes the nexus picture. The question is what the operating model looks like when a $50M brand on Shopify Plus also runs Amazon FBA across 18 states, Walmart Marketplace at a smaller share, and a TikTok Shop pilot that just crossed $100,000 into Washington. TaxCloud is built for that operating model. The reporting API pulls Shopify, Shopify Plus, Amazon, Walmart, and TikTok Shop into one threshold view that applies the 28-jurisdiction inclusion rule and the 19-state exclusion rule per state. Consolidated SST filing across the 24 member states handles the marketplace-offset disclosure line on each return. The documentation trail ties direct-channel collections, marketplace settlement, and FBA inventory together for audit defense.