# Principal vs. Agent, Explained: Gross or Net Revenue Under ASC 606

You run a marketplace that processed USD 500,000 last year. Your seller payouts were USD 425,000, and your platform kept USD 75,000. Now your lender asks a question that sounds simple: what was your revenue — USD 500,000 or USD 75,000? The answer is not a preference and it is not whichever number looks better on a loan application. Under ASC 606 it is a conclusion about control, and getting it wrong overstates your top line by nearly seven times while collapsing your gross margin from a healthy service margin to a razor-thin retail one.

This guide explains the principal-versus-agent test that decides whether you report gross or net, the three indicators that point to control, how marketplaces, agencies, and resellers each tend to land, and the bookkeeping entries that keep the answer consistent in your ledger.

## Why This One Judgment Moves Your Top Line but Not Your Profit

When another party helps deliver what your customer buys, ASC 606 asks about the nature of your promise. If your promise is to provide the specified good or service yourself, you are the principal and you report the full amount the customer pays as revenue, with what you pay the other party as cost. If your promise is only to arrange for the other party to provide it, you are the agent and you report only your fee or commission as revenue.

Take that USD 500,000 marketplace year:

```
Principal (gross)              Agent (net)
Revenue        USD 500,000     Revenue         USD 75,000
Cost of sales  USD 425,000     Cost of sales         USD 0
Gross profit    USD 75,000     Gross profit    USD 75,000
Gross margin          15%      Gross margin         100%
```

Gross profit dollars are identical either way, and net income does not change. Everything else does. Revenue, growth rates, gross margin percentage, revenue-based covenants, and valuation multiples all swing on this one call. A lender testing a minimum-revenue covenant, an acquirer applying a revenue multiple, or a peer comparison against a competitor that reports the other way will read two very different businesses from the same cash flows.

Auditors treat this as a material presentation judgment, not a policy election. You do not choose gross because it looks larger. You conclude principal or agent from the facts, apply it consistently to each revenue stream, and document the reasoning so it survives diligence.

## The Control Test That Decides Everything

The guidance lives in ASC 606-10-55-36 through 55-40, as clarified by ASU 2016-08. The test is a single sentence: do you control the specified good or service before it is transferred to the customer?

Control means the ability to direct the use of the good or service and to obtain substantially all of its remaining benefits. For a physical product, that usually looks like taking title, even briefly, or having the right to redirect the goods to a different buyer. For a service, it looks like performing the work yourself or engaging and directing the subcontractor that does, rather than merely introducing the customer to someone else who performs it.

Three scoping rules matter before you apply the test:

First, you assess each specified good or service separately. A platform that sells its own private-label goods and also hosts third-party sellers can be the principal for the first stream and the agent for the second, on the same checkout page. One company-wide conclusion is almost always wrong.

Second, the specified good or service is what the customer contracted to receive, not the administrative tasks around it. Processing the payment, hosting the listing, or providing the checkout page does not make the listing itself something you control.

Third, the test is applied before transfer to the customer. Rights you obtain only after the customer already controls the goods — a back-to-back indemnity from a supplier, for example — do not make you the principal.

If you controlled the specified item before transfer, you are the principal and you report gross. If you did not, you are the agent and you report the net fee or commission you keep.

## The Three Indicators That Point to Control

ASC 606-10-55-37A through 55-37B names three indicators that support the control assessment. None is decisive on its own. They are evidence, and you weigh them together against the definition of control.

### 1. Primary responsibility for fulfilling the promise

Who does the customer blame when things go wrong, and who has to make it right? If you are on the hook for acceptable fulfillment — you handle defects, re-performance, or refunds out of your own obligation rather than passing a claim through to someone else — that points toward being the principal.

A reseller that warehouses inventory and ships under its own return policy clearly carries primary responsibility. A referral platform where the contractor warrants the work points the other way. Customer support alone does not settle it — the question is who bears the fulfillment obligation in the contract and in practice.

### 2. Inventory risk before or after transfer

Do you hold the goods, or bear the loss if they go unsold, get damaged, or get returned? Inventory risk before the customer takes control is strong evidence of being the principal. So is bearing the loss on returns after transfer without recourse to the supplier.

A distributor that stocks widgets and absorbs obsolete units has inventory risk. A drop-ship intermediary that never takes title and owes nothing for unordered units generally does not — though momentary title or a non-cancellable purchase commitment can create risk even there. For services, the analog is capacity risk: paying a subcontractor for a block of hours whether or not your client uses them looks like a principal; paying the provider only when the end customer pays you looks like an introducer.

### 3. Discretion in establishing the price

Who sets what the customer pays? A reseller that buys at USD 10 and independently decides to sell at USD 25 has clear pricing discretion, which supports a principal conclusion. A platform whose commission is a fixed contractual percentage of a price the seller sets has far less.

Pricing discretion is the most over-read indicator. Agents can have pricing latitude too — the standard says so directly — so this never decides the question alone. What matters is whether the pricing power reflects control of the underlying item or merely flexibility in your fee. Discounting your own commission is not control of the goods.

### What does not decide the question

Two common proxies are explicitly unreliable. Exposure to credit risk — the risk the customer does not pay — is not an indicator of control. An agent that collects the buyer payment and remits the seller share still bears collection risk, and that changes nothing. Likewise, payment processing, invoicing, or tax collection is administrative, not control. Clearing the card transaction does not make you the seller of the goods.

## How Marketplaces, Agencies, and Resellers Each Land

The same test produces different answers in different business models. Here is where each common pattern usually comes out, and what flips it.

### Online marketplaces: usually the agent

The typical marketplace lists seller-owned products, lets sellers set prices, never takes possession, and earns a commission — say 15 percent. The seller ships, warrants the goods, and accepts returns. That marketplace arranges the sale rather than making it, so it reports the commission as revenue and records seller obligations as a payable, not cost of sales.

The same marketplace becomes the principal the moment it controls goods: first-party inventory, private-label products, or third-party units purchased outright before resale are all gross-revenue streams. The tell is operational — purchase orders, warehouse receipts, and inventory write-downs belong to a principal, not an agent. Many platforms run both models, so keep seller payables in a liability subledger for the agency stream and inventory plus cost of sales for the principal stream, or the income statement blends two economics into one misleading margin.

### Ad and creative agencies: the pass-through trap

Media spend is the classic small-agency trap. A client approves a USD 20,000 media plan, the agency places the buy with publishers, and the invoice to the client shows USD 20,000 of media plus a USD 3,000 management fee. Reporting USD 23,000 of revenue feels natural because USD 23,000 of cash arrived. Under ASC 606 it is usually wrong.

If the client approves the placements, the agency cannot redirect the inventory to another client, and the publisher — not the agency — is responsible for running the ads, the agency never controlled the media. It is the agent for the media stream and the principal only for its own service. Revenue is the USD 3,000 fee, with the USD 20,000 recorded as a pass-through payable to publishers.

The answer flips when the agency bundles a deliverable it controls: a fixed-price campaign where it chooses channels, guarantees impressions, and absorbs overruns. Then it is responsible for the outcome and bears the cost risk, supporting gross reporting. The contract terms and approval trail are the evidence either way, so keep them.

### Resellers and distributors: usually the principal

Buy low, sell high, set your own price, stock the goods, and handle returns under your own name, and you are the principal. A reseller that buys at USD 10 and sells at USD 25 controls the good before transfer — it can direct units to any customer, bears the loss if they do not sell, and sets the resale price. Revenue is USD 25 and cost of sales is USD 10.

The exception is the pure referral deal where the vendor still controls the service: the end customer contracts with the vendor, which provisions and supports the seats, while the reseller earns a referral fee. Contract labels like reseller or partner do not override that substance. Conversely, a value-added reseller that owns implementation, support, and the customer relationship often stays the principal for the bundle even when the underlying license passes through.

### SaaS sold through app marketplaces: ask who your customer is

For a software vendor selling through a cloud marketplace, the same analysis decides who the customer even is. If the end user contracts with you while you provision and support the subscription, the end user is your customer: report the end-user price as revenue and the marketplace fee as a selling expense. If the marketplace controls the subscription before transfer — setting the bundle price and supporting the buyer — then it may be your customer and your revenue is only what it pays you. Most vendors conclude the first way and report gross, but enterprise private-offer terms can differ from self-serve terms, so assess each channel on its own contract.

## The Journal Entries: Same Cash, Different Income Statement

Bookkeeping is where an abstract conclusion becomes a concrete income statement. Suppose your platform collects USD 1,000 from a buyer, owes the seller USD 850, and keeps USD 150.

As the principal, the full buyer payment is your revenue and the seller payout is your cost:

```beancount
2026-01-01 open Assets:Bank:Checking USD
2026-01-01 open Income:Marketplace-Sales USD
2026-01-01 open Expenses:COGS:Seller-Payouts USD

2026-11-01 * "Buyer" "Marketplace sale collected - principal, gross"
  Assets:Bank:Checking       1000.00 USD
  Income:Marketplace-Sales  -1000.00 USD

2026-11-02 * "Seller" "Seller payout for principal sale"
  Expenses:COGS:Seller-Payouts  850.00 USD
  Assets:Bank:Checking          -850.00 USD
```

As the agent, the buyer payment creates a liability to the seller, and only your USD 150 commission ever touches revenue:

```beancount
2026-01-01 open Assets:Bank:Checking USD
2026-01-01 open Liabilities:Seller-Payable USD
2026-01-01 open Income:Marketplace-Commissions USD

2026-11-01 * "Buyer" "Buyer payment collected as agent"
  Assets:Bank:Checking         1000.00 USD
  Liabilities:Seller-Payable  -1000.00 USD

2026-11-02 * "Seller" "Remit seller share, retain commission"
  Liabilities:Seller-Payable          1000.00 USD
  Assets:Bank:Checking                 -850.00 USD
  Income:Marketplace-Commissions       -150.00 USD
```

Notice what the agent entries protect you from. The USD 1,000 never appears in `Income`, so a revenue report over `Income:Marketplace-Commissions` cannot accidentally include pass-through cash. The seller obligation sits in `Liabilities:Seller-Payable` where an aging report can track it, ideally with one subaccount per seller so payouts reconcile the way bank accounts do. If you run both models, keep the gross stream under a sales income account and the agency stream under a commissions account from day one — separating them after a year of commingled bookings is the expensive version of this lesson. The [Beancount documentation on open directives and booking](https://beancount.io/docs/) and the [Fava reports](https://beancount.io/fava/) for income-statement views make these flows easy to verify once the accounts exist.

## Five Mistakes That Misstate Revenue

The most common mistake in marketplace accounting is treating the buyer gross payment as platform revenue. It usually starts innocently — the payment processor settles the full amount into your bank account, so the bookkeeper books the deposit as sales. From there the error compounds: seller payouts get booked as refunds or generic expenses, revenue is overstated by a multiple, and the margin looks like a distressed retailer instead of a software-enabled service. Fix the booking at the bank feed, not at year end.

The second mistake is equating payment processing with control. Collecting the card payment, holding funds in escrow, issuing the invoice, or filing marketplace-facilitator sales tax does not mean you controlled the underlying goods. Each of those is an agency-compatible function, and auditors will ask what evidence beyond touching the cash supports a principal conclusion.

The third is counting credit risk as an indicator. Bearing the loss when a buyer defaults feels like ownership of the transaction, but the standard excludes credit risk from the control assessment deliberately. Plenty of agents guarantee collection or advance payouts to sellers against future settlement. That is a financing or credit decision layered on top of an agency relationship.

The fourth is letting pricing discretion carry the whole analysis. The ability to set or adjust the price supports control, but agents can have pricing latitude too, and the standard says so directly. A marketplace that lets sellers run discount campaigns within guardrails has not necessarily handed control to anyone. Read pricing together with fulfillment responsibility and inventory risk instead of letting it decide alone.

The fifth is assessing the company once instead of each revenue stream continuously. Contracts change: a marketplace adds first-party inventory, an agency moves from pass-through media to guaranteed campaigns, a reseller shifts from stocked goods to referral deals. Each change reopens the question for that stream. A one-page memo per stream — specified good or service, control conclusion, indicator evidence, and the contract clauses that support it — turns a future audit inquiry into a file retrieval instead of a reconstruction project.

## A Practical Checklist for Your Next Revenue Stream

Work through these steps before the first invoice for any stream where another party touches fulfillment, and revisit them whenever the contract terms change materially.

First, name the specified good or service the customer contracted to receive, in the customer's words. Second, ask the control question directly: did you control that item before it transferred to the customer, with the ability to direct its use and capture its benefits? Third, walk the three indicators honestly — fulfillment responsibility, inventory or capacity risk, and genuine pricing discretion — and write down the evidence for each, including the contract sections and the operational facts. Fourth, book it the way you concluded from the first transaction: gross revenue with cost of sales for a principal stream, a seller or vendor payable with commission revenue for an agency stream, in separate accounts that never mix. Fifth, align everything downstream — covenant calculations, board-deck growth rates, sales tax positions, and information-return reporting — with the same conclusion, because a revenue number that means different things in different filings is the inconsistency examiners notice first.

Done this way, the principal-versus-agent call stops being a year-end scramble and becomes part of how you launch: each new channel gets its memo, its accounts, and its reconciliation before it gets its first customer.
