Gross vs Net Revenue: Principal vs Agent Explained

Introduction

Two companies each move $1 billion of merchandise a year from manufacturers to shoppers, and each keeps $100 million of it. The first buys the goods, carries the risk of being stuck with them and sets the shelf price. The second lists other companies' products on its website and charges a 10% fee on every sale.

The first reports $1 billion of revenue; the second reports $100 million. Their gross profit, EBITDA and cash generation can match to the dollar, yet one top line is ten times the other. The only difference is whether each acts as the principal or the agent in the sale.

That distinction decides whether revenue is reported gross or net, and it runs through every comps table that mixes retailers with marketplaces or freight brokers with ride-hailing apps. Margins, EV/Revenue multiples and growth rates all bend around it. The sections below cover the ASC 606 control test, how real companies apply it in their filings, how GMV and take rate connect to revenue, and how bankers put a mixed peer set on one basis.

The two presentations describe one transaction from different seats. A gross reporter books what the customer pays and records the supplier's share as a cost; a net reporter books only the slice it keeps.

The "unchanged" row carries the whole argument. Presentation moves revenue and cost of revenue by the same amount, so profit dollars never shift, while every ratio with revenue in the denominator does.

One P&L, Two Presentations

Extend the opening example into a full income statement. Each company spends $20 million on payment processing and support, booked in cost of revenue, and $40 million on technology, marketing and overhead. Only the principal records the shopper's payment as revenue and the $900 million owed to manufacturers as a cost.

Every dollar of profit matches; the margins differ tenfold only because the denominators do. An 80% and an 8% gross margin here describe identical economics, in figures built to isolate the effect.

Gross vs Net Revenue Reporting

Two ways of presenting the same sale. A company reporting gross, as principal, records the full amount the customer pays as revenue and the supplier's share as cost of revenue. A company reporting net, as agent, records only its fee or commission. Gross profit, EBITDA and cash flow are identical either way; revenue and margin percentages are not.

What Moves and What Stays Put

Presentation is a reclassification, not a change in economics, which is why interviewers like the topic: it tests whether a candidate knows what each income statement line measures.

For a pure agent, then, net revenue behaves much like a principal's gross profit. Booking Holdings said so in its 2018 annual report: when ASC 606 moved its Name Your Own Price business from gross to net, it dropped the cost of revenues line and noted that revenues from 2018 onward were comparable to the gross profit of prior periods.

The Control Test Under ASC 606

US GAAP settles the question inside ASC 606, whose five-step recognition model decides when and at what price revenue is recorded; the principal-versus-agent guidance decides whether that is the full sale or only the fee.

The test has two steps. Identify the specified good or service the customer is buying (the ride, the hotel night, the sofa), then ask whether the company controls it before it reaches the customer. A company that does is a principal and recognizes the gross amount. One whose promise is only to arrange for another party to provide it is an agent and recognizes its fee. Holding legal title for a moment, the standard adds, does not by itself make a company the principal.

Principal vs Agent (Revenue Recognition)

The test that decides whether a company reports revenue gross or net. A principal controls the specified good or service before it is transferred to the customer and records the full amount the customer pays as revenue. An agent arranges for another party to provide it and records only its fee or commission. Under ASC 606 and IFRS 15 the deciding question is control, with primary responsibility, inventory risk and pricing discretion as supporting evidence.

Three Indicators, Not a Checklist

Control is hard to see in a three-party transaction, so the standard lists indicators that a company has it. As amended by FASB's Accounting Standards Update 2016-08, they are:

The 2016 update deleted two older indicators, commission-based consideration and exposure to credit risk, and stressed that the remaining three support the control assessment rather than replace it. Because the test runs per specified good or service, one company can be principal for some products and agent for others.

Judgment is unavoidable: some respondents to the IASB's post-implementation review of IFRS 15 reported inconsistent outcomes for similar fact patterns, notably in e-commerce, internet advertising and fintech. IFRS 15 applies the same converged test (paragraphs B34 to B38), so on this point GAAP versus IFRS comparisons need no adjustment.

Revenue recognition is one corner of the accounting technicals: The 160-page PDF covers accounting, valuation and deal mechanics alongside the technical questions banks actually ask, access the IB Interview Guide and keep the principal and agent logic next to the rest of your accounting preparation.

E-Commerce: First-Party Versus Third-Party Sales

Amazon runs both models, and its 10-K labels them. Online stores covers product sales recorded gross; third-party seller services covers commissions plus fulfilment and shipping fees from merchants whose sales Amazon handles without being the seller of record. In 2025 online stores brought in about $269 billion and third-party seller services about $172 billion, although third-party sellers account for roughly three in five paid units in Amazon's quarterly disclosures.

Wayfair shows that inventory on the balance sheet is not the test. Most of its products ship straight from suppliers' warehouses, yet it reports product sales gross, because it is responsible to the customer for fulfilment and acceptability, bears inventory risk from shipment through delivery, sets prices and selects the suppliers. Control, not possession, decides it.

Ride-Hailing and Travel

Uber lands on both sides. Where it simply provides the platform drivers and merchants use, it says it does not control the service and records Mobility and Delivery revenue net; in markets where it agrees to provide those services to end users for a fee, it reports gross. Freight revenue is mostly gross, because Uber is primarily responsible for moving shippers' goods and has pricing discretion over both what shippers pay and what carriers receive.

Airbnb is the cleanest agent among the large platforms: its 10-K says it does not control the right to use the properties, fulfil rental promises, bear inventory risk or set prices. Booking Holdings reports net even in its merchant model, where it collects the traveler's payment, because its contracts never hand it responsibility for delivering the travel.

Payments: Merchant of Record and Pass-Through Costs

Much of what a merchant pays on a card transaction flows on to issuers and networks. Block treats Square transaction fees as gross revenue: as merchant of record, it says it is the principal in delivering payment services to sellers, and it books processing costs in cost of revenue. A processor that is not the principal for those pass-through fees nets them out of revenue instead, so equal volume can produce very different top lines.

Marketplaces publish volume metrics because net revenue understates their activity. Gross merchandise value (GMV) is the e-commerce term and gross bookings the travel and mobility one; both measure total transaction value. Net revenue is the slice the platform keeps, and the ratio of the two is the take rate:

In the opening example, $100 million of revenue on $1 billion of GMV is a 10% take rate. The principal's equivalent is the 10% spread between what shoppers pay and what manufacturers receive: the same economics, appearing as a take rate at one company and as the margin over supplier cost at the other, before processing costs take both to 8% of GMV in gross profit. The TMT guide covers marketplace take rates and unit economics in depth, and Alibaba's 2014 IPO shows a platform whose merchandise volume dwarfed its revenue because it never owned the goods.

Take Rate

The share of a platform's transaction volume that it keeps as revenue, calculated as net revenue divided by GMV or gross bookings. A marketplace that processes $10 billion of bookings and reports $1.5 billion of revenue has a 15% take rate. For a company reporting gross, revenue already equals transaction value, so the equivalent measure is gross margin.

Definitions are company-specific: Uber's gross bookings include taxes, tolls, driver earnings and its gross-reported Freight revenue at face value, so more Freight lifts the blended take rate without any price change.

Margins and Multiples in a Mixed Peer Set

Benchmark the principal against the agent and the agent looks ten times more profitable, with a 40% EBITDA margin against 4%. Margin percentages only compare between companies on the same basis; EBITDA per dollar of GMV gives both 4 cents, which is the honest answer.

Multiples break the same way. At 10x EBITDA each company is worth $400 million, so the principal trades at 0.4x revenue and the agent at 4.0x. Apply the agent's multiple to the principal's $1 billion of revenue and you get $4 billion, ten times what its earnings support. EV/Gross Profit (5.0x for both) and EV/EBITDA (10x for both) are immune. Where peers are unprofitable and revenue multiples are unavoidable, restate everyone to net revenue first, part of the peer-selection discipline in how to build a comparable company analysis.

When Presentation Changes, Reported Growth Changes

Move half the principal's volume onto a third-party marketplace at the same 10% fee and revenue falls from $1,000 million to $550 million, down 45%, with gross profit and EBITDA untouched. Reported growth can swing either way on presentation alone, as filings show:

Normalising a Peer Set

A comps table that mixes gross and net reporters needs a deliberate fix before anyone reads a multiple off it. The work runs in order:

The adjusted table is less tidy, but it compares businesses rather than accounting choices.

Interviewers use the topic to test whether a candidate can see through the revenue line. Expect variations on:

Gross versus net revenue questions sit where accounting meets valuation: Work through revenue recognition, comps and multiples questions with worked answers on the practice platform, start practicing interview questions for free and see which explanations still need tightening before a superday.

Most errors treat the revenue line as a fact about the business rather than a presentation choice. Reporting gross is not wrong in itself, but an unexplained shift toward it belongs alongside other accounting red flags that signal weak earnings quality.

Gross versus net is one of the few accounting topics that changes how big a company looks without changing what it earns. That makes it dangerous exactly where bankers spend their time: comps tables, margin benchmarks, revenue multiples and the growth narrative in a pitch book.

The protective habits are simple. Read the revenue recognition note before trusting a revenue figure. Pair every marketplace's revenue with its GMV or gross bookings, and every gross reporter's revenue with its gross profit. When a peer's growth rate jumps or collapses, look for a presentation change before looking for a story.

In an interview, the question rewards candidates who start with control and move quickly to the consequences for margins and multiples. That shows an understanding of what the revenue line actually measures, which is the judgment the question is designed to test.