Principal vs Agent: Revenue Recognition Explained
Whether your company books $1 billion or $150 million in revenue can come down to a single accounting question: are you the principal or the agent? Under ASC 606 and IFRS 15, that determination drives whether you recognise revenue at the gross amount you receive from customers or only the net fee you retain. For a marketplace with $1 billion in gross merchandise value and a 15% take rate, the difference is $850 million on the top line.
This guide explains what principal and agent mean, why the distinction exists, how the two-step control framework works, and where finance teams most often go wrong.
Key takeaway: A principal controls the specified good or service before it is transferred to the customer and recognises gross revenue. An agent arranges for another party to provide that good or service and recognises only its fee or commission.
What Is the Difference Between a Principal and an Agent?
A principal is the party that controls the specified good or service before transferring it to the customer. Because the principal bears the risks and rewards of that control, it records the full amount receivable from the customer as revenue, then separately records the cost of the good or service provided by any third-party supplier.
An agent, by contrast, never controls the good or service. Its job is to arrange for another party to provide it. The agent records only the net fee or commission it retains, regardless of whether the gross cash flows pass through its bank account first.
The distinction matters because it is not cosmetic. Gross versus net recognition affects:
- Reported revenue and revenue growth rates
- Gross margin percentages (a principal's cost of sales includes the supplier payment; an agent's does not)
- Revenue-based debt covenants
- Analyst expectations and valuation multiples
The FASB codified this framework in ASC 606-10-55-36 through 55-40, and the IASB adopted substantially identical language in IFRS 15 paragraphs B34 through B38. Both standards became effective for annual periods beginning on or after 1 January 2018.
Why the Principal vs Agent Distinction Exists
Before ASC 606 and IFRS 15, preparers applied a patchwork of industry-specific guidance and general concepts to decide whether to report gross or net. The results were inconsistent, particularly as platform and marketplace business models proliferated. The FASB and IASB developed the joint standard to replace that patchwork with a single, principle-based framework anchored in one concept: control.
ASU 2016-08, issued in March 2016 (less than two years after the original ASU 2014-09), clarified the principal vs agent guidance specifically because implementation questions arose so quickly. The FASB-IASB joint Transition Resource Group held multiple meetings between 2014 and 2016 to address the volume and complexity of questions in this area alone.
The speed of those clarifications signals something important: this is one of the most judgment-intensive areas of revenue recognition, and the standard-setters knew it.
The Two-Step Framework: How to Determine Principal vs Agent Status
PwC's Viewpoint guidance summarises the framework precisely: "The principal versus agent assessment is a two-step process that consists of (1) identifying the specified good or service to be provided to the customer and (2) assessing whether the entity controls the specified good or service before it is transferred to the customer."
Neither step is optional. Skipping to step two without completing step one is the most common structural error in practice.
Step 1: Identify the Specified Good or Service
The specified good or service is the unit of account for the control assessment. It can be a physical good, a service, a right to a good or service, or a bundle of goods or services that constitutes a distinct performance obligation.
The "right to a good or service" variant is where many preparers stumble. Consider an online travel agency (OTA) that purchases airline seats in bulk and resells them. The specified good or service the OTA provides to the traveller may be the seat itself (making the OTA a principal) rather than a facilitation service (which would make it an agent). The answer depends on what the OTA actually controls.
In a marketplace model, the specified good or service is often the marketplace's facilitation service, not the underlying product sold by a third-party merchant. Identifying it as the latter without adequate analysis is a common error that inflates revenue.
Step 2: Assess Control Before Transfer
Control means the ability to direct the use of, and obtain substantially all of the remaining benefits from, the specified good or service. This is the same definition of control used elsewhere in ASC 606 and IFRS 15.
As ASC 606-10-55-36 states directly: "An entity is a principal if it controls the specified good or service before that good or service is transferred to a customer."
One critical point that the standard is explicit about: legal title is not determinative. An entity may hold legal title momentarily as a legal formality without controlling the good. Conversely, an entity may control a good without ever holding legal title. Finance teams that equate title transfer with principal status will reach the wrong answer.
The Three Indicators: Evidence, Not a Checklist
Under ASC 606-10-55-39 and IFRS 15 paragraph B37, three indicators support (but do not replace) the control assessment:
| Indicator | What It Signals | Common Misread |
|---|---|---|
| Primary responsibility for fulfilment | The customer holds the entity accountable for delivery, not the third-party supplier | Assuming that because a supplier does the work, the entity cannot be primarily responsible |
| Inventory risk | The entity bears the risk of loss before transfer, or after transfer if the customer has a right of return | Thinking inventory risk requires physical possession of goods |
| Pricing discretion | The entity sets the final price charged to the customer | Confusing the ability to set a margin with true pricing discretion |
This is the point where most articles, and many preparers, go wrong. ASU 2016-08 was issued precisely to clarify that the indicators are not a separate or overriding test. They are factors that support the overall control assessment. An entity that scores two out of three indicators is not automatically a principal. The question is always whether, in light of all the facts and circumstances, the entity controls the specified good or service.
KPMG's revenue recognition handbook notes that the "primarily responsible" indicator is often the most persuasive of the three, because it speaks directly to who the customer holds accountable for fulfilment. If the customer looks to the entity, not the supplier, to resolve problems, that strongly supports principal status.
EY's revenue recognition guide highlights that inventory risk is assessed both before and after transfer. An entity that bears the risk of loss if a customer returns a product, even if it never physically held the inventory, may still have inventory risk sufficient to support principal status.
Principal vs Agent: Gross vs Net Revenue Recognition
The financial statement mechanics follow directly from the conclusion:
| Principal | Agent | |
|---|---|---|
| Revenue recognised | Gross amount received from customer | Net fee or commission retained |
| Cost of sales | Includes payment to third-party supplier | Typically minimal |
| Gross margin % | Lower (cost of supplier included) | Higher (only fee retained) |
| Example: $1B GMV, 15% take rate | $1,000,000,000 revenue | $150,000,000 revenue |
The $850 million difference in that example is not hypothetical. For marketplace businesses, SaaS platforms, and financial intermediaries, this determination can be the difference between meeting analyst revenue expectations and missing them by a wide margin.
For a real-world illustration of how this plays out in a high-profile filing, see Finrep's analysis of Anthropic's gross vs net revenue conclusion for its AWS and Google Cloud arrangements.
Can a Company Be Both Principal and Agent?
Yes, and this is common. As Deloitte's Accounting Spotlight notes: "In arrangements involving more than one distinct good or service, an entity could be a principal for certain aspects of a contract with a customer and an agent for other aspects."
The assessment must be performed separately for each specified good or service within a contract. A software company might be a principal for its own software licence (which it controls) and an agent for third-party implementation services it arranges on the customer's behalf (which it does not control). Applying a single conclusion across the whole contract without disaggregating performance obligations is a material error.
IFRS 15 vs ASC 606: Where They Differ
The two standards are substantially converged on principal vs agent, but not identical. The differences matter for multinational companies that report under both frameworks or are transitioning between them.
| ASC 606 (US GAAP) | IFRS 15 | |
|---|---|---|
| Core framework | Two-step: identify + control | Two-step: identify + control |
| Three indicators | Same three indicators (ASC 606-10-55-39) | Same three indicators (IFRS 15.B37) |
| Illustrative examples | More extensive (including financial services, healthcare) | Fewer examples; some gaps filled by IFRIC agenda decisions |
| IFRIC agenda decisions | Not applicable | Active: covers software resellers, cloud computing, SaaS |
| Credit risk | Not listed as an indicator | Explicitly excluded as irrelevant (IASB concluded agents face credit risk too) |
| Scope | Goods, services, rights to goods/services | Same, plus confirmed by IFRIC to include intangible assets and financial instruments used as consideration |
The IFRS Interpretations Committee (IFRIC) has issued agenda decisions that fill gaps in IFRS 15's illustrative examples, particularly for technology arrangements. The May 2022 IFRIC agenda decision on software resellers is essential reading for any IFRS preparer in the software sector.
Grant Thornton's global IFRS team, updated in April 2026, confirms that while IFRS 15's principal-agent guidance is intended to make the assessment easier, it "does not eliminate the need for significant judgement, particularly in complex multi-party arrangements."
Industry-Specific Application: Where This Gets Contested
Marketplace and Platform Businesses
The central question is whether the platform controls the underlying product before the merchant ships it to the buyer. Most marketplace operators do not: they do not take inventory risk, they do not set the final price (the merchant does), and the customer holds the merchant, not the platform, accountable for the product. The platform's specified good or service is typically the facilitation service, making it an agent that recognises only its commission.
SaaS Resellers
For software and SaaS resellers, the analysis turns on whether the reseller can direct the use of the licence and obtain substantially all of the remaining benefits before passing it to the customer. Most resellers cannot: the software vendor controls the licence terms, the reseller cannot modify or redirect the software, and the customer's relationship is ultimately with the vendor. Many SaaS resellers are agents under this analysis, recognising only their margin or referral fee.
Online Travel Agencies
Most OTAs are agents. They do not take inventory risk on hotel rooms or airline seats (unsold inventory returns to the supplier), and the final price is typically set by the hotel or airline. The OTA earns a commission. The exception is an OTA that purchases inventory in bulk at a fixed price and bears the risk of unsold rooms, which may support principal status.
Pharmacy Benefit Managers (PBMs)
PBMs must assess whether they control the drug before dispensing or are merely arranging for a pharmacy to dispense it. The analysis is highly fact-specific and depends on whether the PBM sets the price, bears inventory risk, and is primarily responsible to the plan sponsor for drug delivery. The SEC has issued comment letters on PBM revenue recognition.
Common Mistakes Finance Teams Make
- Treating the three indicators as a checklist. Two out of three does not equal principal. The control assessment is the test; the indicators are evidence.
- Misidentifying the specified good or service. Skipping step one and jumping to the control assessment produces unreliable conclusions.
- Equating legal title with control. The standard is explicit: title is not determinative.
- Applying a single conclusion to a mixed contract. Each performance obligation requires a separate assessment.
- Assuming a third-party fulfiller means agent status. Using a subcontractor does not automatically make an entity an agent. If the entity controls the output and is primarily responsible to the customer, it may still be the principal.
- Insufficient documentation. A conclusion without a documented control assessment is not auditable.
What the SEC Expects on Principal vs Agent Disclosures
The SEC's Division of Corporation Finance has been active in challenging gross revenue recognition, particularly in e-commerce, marketplace, and platform businesses. Staff have focused on whether entities truly bear inventory risk and are primarily responsible for fulfilment, or whether they are effectively arranging for third-party sellers to fulfil the promise.
The SEC's revenue recognition interpretive guidance makes clear that boilerplate disclosure is insufficient. Staff expect entity-specific explanation of how the control test was applied to the company's actual arrangements, not a recitation of the standard's language. Companies that disclose only that they "evaluated whether they are the principal or agent" without explaining the conclusion and the key judgements involved are likely to receive a comment letter.
For public companies, the documentation standard for auditors and the SEC is the same: a memo that walks through the two-step framework, identifies the specified good or service, applies the control test with reference to the specific facts of the arrangement, and explains why the three indicators, taken together, support (or do not override) the control conclusion.
FAQ
What is the difference between a principal and an agent in accounting? A principal controls the specified good or service before transferring it to the customer and recognises gross revenue. An agent arranges for another party to provide the good or service without taking control, and recognises only its net fee. The distinction is governed by ASC 606-10-55-36 (US GAAP) and IFRS 15 paragraph B34 (IFRS).
What is the difference between a principal and an agent in VAT? For VAT purposes, the principal vs agent distinction follows similar logic but is governed by tax law, not accounting standards. An agent acting in its own name may be treated as the principal for VAT, meaning it accounts for VAT on the full transaction value. An agent acting in the name of the principal typically accounts for VAT only on its commission. The rules vary by jurisdiction and are separate from ASC 606 / IFRS 15.
Can you give an example of a principal and agent relationship? An online marketplace that connects buyers with third-party sellers, earns a 15% commission, does not hold inventory, and does not set the final price is typically an agent. It recognises $150 million in revenue on $1 billion of gross merchandise value. A retailer that buys goods from a supplier, holds them in its own warehouse, sets the retail price, and is responsible to the customer for delivery is a principal. It recognises the full sale price as revenue.
Are the three indicators a pass/fail test? No. The three indicators (primary responsibility, inventory risk, pricing discretion) support the control assessment but do not replace it. An entity that satisfies all three indicators is likely a principal, but the overarching question is always whether it controls the specified good or service before transfer.
What happens if a company changes from a reseller model to a marketplace model? A change in business model may trigger a reassessment of principal vs agent conclusions. If the conclusion changes, the entity must determine whether the change is a change in accounting estimate or policy, apply the appropriate transition method, and provide enhanced disclosure. This is a live area of SEC scrutiny for companies that have launched marketplace features alongside their core reseller operations.
What documentation do auditors expect for a principal vs agent conclusion? Auditors expect a written memo that: (1) identifies the specified good or service for each performance obligation; (2) applies the control test with reference to the specific contractual and operational facts; (3) evaluates each of the three indicators and explains their weight in the overall assessment; and (4) reaches a documented conclusion with supporting rationale. A generic policy statement is not sufficient.
