How does the auditor treat marketplace-facilitated sales during a direct seller's audit?
At $20M to $80M on Shopify or Shopify Plus, a brand running direct alongside Amazon, Walmart Marketplace, TikTok Shop, and Faire enters the audit assuming the marketplace-facilitated portion is the marketplace's problem. The state agrees on the statute. The audit still touches the marketplace volume because the seller is registered, files its own return, and is the party the state is examining. The question is whether the seller's return correctly separated the two flows.
The framework is set by each state's marketplace facilitator statute. California's AB 147 (Stats. 2019, ch. 5) designates the marketplace as the collector and remitter on facilitated transactions (Cal. Rev. & Tax. Code §6041; CDTFA Special Notice L-805). [1] New York's Part G of Ch. 59 of Laws of 2019, codified at NY Tax Law §1101(e)(1), does the same. [2] Texas's SB 70 (2019), codified at Tex. Tax Code §151.0242 and implemented through 34 TAC §3.286(g), follows the pattern. [3] The statutory effect is the same across the 47 sales-tax jurisdictions with marketplace facilitator laws: tax on a facilitated transaction is the marketplace's collection and remittance obligation, not the seller's.
What the audit examines is different from what the statute moves. The audit examines the seller's records: the seller's return, the seller's ledger, the seller's reconciliation. The marketplace's statutory role is the predicate for what the seller should have reported, not a wall keeping the marketplace volume out of the examination. The auditor traces the marketplace-facilitated volume through the seller's books to verify three things: that the return correctly reported gross sales including the marketplace channel, that the offset disclosure correctly excluded the marketplace-collected tax from the seller's remittance obligation, and that the offset is supported by settlement-level evidence from the marketplace.
When any of those three breaks, the audit pulls the marketplace volume back into the assessment by default, on the theory that the seller has not proven the marketplace covered it. The default direction at audit is toward including, not excluding.
How auditors separate marketplace from direct-channel sales in the production
The first IDR cycle in a multi-channel audit asks for the seller's records broken out by channel. The CA CDTFA Audit Manual, [4] NY DTF Publication 130-D, [5] and TX Comptroller Audit Procedures Manual [6] each describe the same opening: gross sales by state and period, transaction-level calculation logs from the calculation engine, filed returns with offset disclosures, and marketplace settlement reports by state and period. The auditor does not run the separation; the seller does. The auditor tests it.
What separates marketplace from direct cleanly is a channel flag at intake. Every order entering the seller's ledger carries a tag identifying its source: Shopify direct, Shopify Plus direct, Amazon, Walmart Marketplace, TikTok Shop, Faire. The tag flows from order capture through the calculation log and into the filed return as the basis for the offset disclosure. A brand with the channel flag in place produces the audit population in two columns by state: direct-channel gross with the seller's tax collected, and marketplace-facilitated gross with the marketplace's tax credited as an offset. A brand without the flag is reconstructing the split at audit, which moves the cost of the separation from the operating layer to the audit response.
The audit failure mode is the brand that ran one ledger across all four channels with no source tag. Sales totals tie to the filed return, but the auditor cannot trace which dollars are direct and which are facilitated. The standard auditor response is to treat the entire revenue base as direct until the seller proves otherwise. Proving otherwise on a multi-year audit window, transaction by transaction, is a documentation project the seller did not anticipate budgeting.
The records the auditor compares against are independent of the seller's ledger. Amazon's Sales Tax Report inside Seller Central exports gross marketplace sales by state and tax-collected by state. Walmart Marketplace's seller-side reports expose the same data on a different cadence with different field names. TikTok Shop's seller-side reporting matures monthly. Faire publishes wholesale settlement and exempt-buyer status separately. The auditor pulls the marketplace's data from the seller, ties the marketplace's gross by state to the seller's reported gross by state, and ties the marketplace's tax-collected to the seller's offset disclosure. Gaps surface as assessment items.
TaxCloud's reporting API exports transaction-level calculation logs by channel for any period, with the source tag preserved from order capture, the jurisdiction sourcing decision, and the tax computed. The same export the seller uses to prepare the return is the export the auditor reconciles against, with the channel column already populated.
Documentation that proves the marketplace collected and remitted
Three layers of documentation establish marketplace coverage at audit, ordered by how directly each ties to the seller's offset disclosure. The auditor works through them in sequence and reaches the assessment by default on whatever the seller does not produce.
The marketplace's settlement report, by state and period
This is the primary document. Amazon's Sales Tax Report (Seller Central, Reports, Tax Document Library, Sales Tax Reports) exports gross marketplace sales by state and tax-collected by state for any period. Walmart Marketplace exports the equivalent through Seller Center's Tax & Compliance section. TikTok Shop's seller-side reporting publishes monthly settlement with a tax-collected-by-state breakdown. Faire exposes settlement separately for wholesale and DTC marketplace transactions. The audit-defense rule is to retain the settlement export alongside the filed return for every period filed and to store it outside the marketplace's seller portal. Retention windows on settlement data through standard seller portals run from twelve to thirty months depending on platform.
The marketplace's statutory certification
When settlement reports are incomplete or contested, the secondary documentation is the marketplace's certification under the state's marketplace facilitator statute. California publishes the certification framework in CDTFA Special Notice L-805 (October 2019), which clarifies that registered marketplace facilitators are responsible for collection and remittance on facilitated sales under Cal. Rev. & Tax. Code §6041. [1] New York's marketplace provider definition (NY Tax Law §1101(e)(1), enacted by Part G of Ch. 59 of Laws of 2019, A.7508-A) carries the same effect. [2] Texas's marketplace provider certification under 34 TAC §3.286(g) and SB 70 (2019), codified at Tex. Tax Code §151.0242, mirrors the framework. [3] Amazon, Walmart Marketplace, and TikTok Shop publish their state-by-state marketplace facilitator registration and tax-collection status on their respective seller-help pages. Print and retain those pages as supporting documentation for each audit-period quarter.
The seller's return offset disclosure
Most states' remote-seller return forms include a separate line for marketplace-facilitated sales, treating marketplace-collected tax as an offset that zeros out the seller's remittance on the facilitated portion. Florida (Fla. Stat. §212.0596) [7] and Texas (34 TAC §3.286) [3] have explicit marketplace lines on remote-seller returns. New York's ST-100 series and ST-810 series accept marketplace-facilitated sales on Schedule N. [2] California's CDTFA-401 series treats facilitated sales on the deduction schedule. [1] The audit examines whether the offset on each return reconciles to the marketplace settlement export for the period, on the gross and on the tax-collected.
A complete production produces the three layers together. A partial production produces an assessment, on the dollars the missing layer would have offset.
Why FBA inventory creates direct-channel liability the audit still reaches
FBA inventory is the trap door in a marketplace audit. The seller enters confident the marketplace handled Amazon-channel transactions cleanly. The state agrees on the Amazon-channel tax. The audit still finds direct-channel exposure because FBA inventory in the state created physical nexus on the brand's direct channel, separately from the marketplace coverage. The marketplace's role applied to facilitated sales. The brand's direct-channel sales into the state were never covered by the marketplace's remittance, and the brand may never have registered or collected on them.
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 [8]. Every sales-tax state still asserts physical-presence rules through "engaged in business" provisions (Cal. Rev. & Tax. Code §6203 [1]; Tex. Tax Code §151.107(a)(3) [3]; RCW 82.08.052 [9]; NY Tax Law §1101(b)(8) [2]). California CDTFA, Washington DOR, Texas Comptroller, and New York DTF 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 where sellers lack control over which fulfillment centers receive product. [10] The PA DOR did not appeal. The ruling binds Pennsylvania courts only. No federal circuit and no other state has adopted the reasoning.
The audit math at this scale: a $40M brand running Shopify Plus direct plus Amazon FBA has FBA inventory across 15 to 20 states from the months after enrolling, placed by Amazon's distribution network without seller direction. In states where FBA inventory creates physical nexus (California, Washington, Texas, New York, and most others outside Pennsylvania), the brand has a direct-channel registration obligation from the date inventory first arrived. That obligation runs even when Amazon collected and remitted on every Amazon-channel transaction in the same state. The audit examines two flows: the Amazon-channel transactions Amazon covered, and the direct-channel Shopify Plus sales the brand should have collected on but did not, because the brand never registered.
The Amazon Inventory Event Detail Report, filtered by fulfillment center and by state, establishes the date inventory first arrived. Standard Seller Central retention on the report is approximately 18 months. Brands that have not archived it quarterly arrive at audit unable to prove the nexus start date in their favor. The state's default position is the earliest plausible date the brand cannot rebut.
The settlement-report-to-books reconciliation the auditor runs
The reconciliation the auditor runs is settlement-report-versus-books. The brand produces Amazon, Walmart Marketplace, and TikTok Shop settlement reports showing tax-collected-and-remitted by state and period, tied to the brand's own ledger by channel, against the marketplace-offset disclosure on the filed return. Brands that produce that reconciliation cleanly close the marketplace question at the field-auditor level. Brands that cannot get the marketplace sales pulled into the assessment by default, with the assessment standing until the brand produces what it should have retained.
The reconciliation runs three ties.
Tie 1: Settlement gross to filed-return gross by state, by period
The auditor takes Amazon's Sales Tax Report for California Q3 2024, reads gross marketplace sales (the number Amazon collected on), and ties it to the brand's California Q3 2024 CDTFA-401 return on the line reporting gross sales including the marketplace channel. [1] If the brand's return reports total gross of $X and Amazon's settlement shows $Y as marketplace-facilitated, the brand's direct-channel gross is $X minus $Y. The auditor verifies the subtraction.
Tie 2: Settlement tax-collected to filed-return offset disclosure by state, by period
The auditor takes Amazon's tax-collected-by-state figure for California Q3 2024 and ties it to the brand's California return on the marketplace-offset deduction line. If Amazon's settlement shows $Z of California tax remitted and the brand's return claims $Z as an offset, the tie is clean. If the brand's offset is larger than Amazon's remittance, the difference is exposure on the seller. If smaller, the seller over-remitted and is owed a refund the audit may surface separately.
Tie 3: Direct-channel tax collected to direct-channel gross
The auditor takes the brand's calculation log for California Q3 2024 direct-channel transactions, applies the California rate matrix, and reconciles to the tax-collected line on the return. Direct-channel gross from Tie 1, direct-channel tax from this tie, tied to the remittance confirmation. Gaps are assessment items.
The reconciliation that wins is one workpaper per state per period showing all three ties on one page, with the underlying settlement export, calculation log, and filed return attached as source documents. TaxCloud's reporting API pulls direct-channel calculation logs by jurisdiction alongside marketplace settlement data ingested through the platform's marketplace connectors, with both flows tagged to the channel of origin. The same workpaper that supported the filing supports the audit, on the same field layout the auditor reconciles against. The reconciliation is built once, used twice.
What the multi-channel audit-ready operating model looks like
The brand at audit is not asking whether the multi-channel reconciliation will close. The IDR has arrived, the opening conference is closed, and the question is whether the documentation the brand built before today is going to hold. Three patterns hold across multi-channel audits that close at the field-auditor level rather than escalating.
The channel tag was in place at order capture, not reconstructed at audit
Every order from Shopify, Shopify Plus, Amazon, Walmart Marketplace, TikTok Shop, and Faire entered the brand's ledger with its source flag and its facilitated-or-not flag. The audit population produces two clean columns by state and period without per-order reconstruction. Brands without the flag in place spend the first IDR cycle building the segregation the auditor expected to receive prepopulated.
The settlement reports were archived quarterly, outside the marketplace portal
Amazon's Sales Tax Report, Walmart Marketplace's tax compliance export, and TikTok Shop's settlement breakdown are retained per period alongside the filed return. Retention windows through seller portals run from twelve to thirty months. The brand's own archive runs the full audit-retention period for the state: eight years for unfiled-return periods in California (Cal. Rev. & Tax. Code §6487), [11] four years in Texas (Tex. Tax Code §111.0041), [12] and three years in New York for timely-filed periods (NY Tax Law §1135). [13]
The Amazon Inventory Event Detail Report was archived by state, by quarter
The document that establishes when FBA inventory first arrived in each state is the document the audit needs to set the physical-nexus start date in the brand's favor. Standard Seller Central retention is approximately 18 months. The brand that archives quarterly arrives at audit with a defensible nexus-start date for each FBA state. The brand that has not archived gets the state's default position, which is the earliest plausible date the brand cannot rebut.
The reader here is past wondering whether the marketplace volume will be touched at audit. It will. The question is what the production layer looks like when the auditor opens it. TaxCloud preserves the channel flag from order capture through the reporting API, so the same export that supports the filing supports the audit, direct and facilitated channels reconciled on one workpaper per state per period. [14]