Revenue Recognition: When a Sale Is a Sale

Canonical data. Worked examples use Wexford Instruments. The contract illustration is a sub-ledger scenario within Wexford's revenue, expressed in $000; consolidated figures are USD millions and agree with the canonical page.

Revenue is the top line, the input to most growth comparisons, and the number most often cited as evidence that a business is working — and deciding which period it belongs in requires answering questions that frequently have no self-evident answer. This article is about why that judgement is genuinely hard, not about how it gets abused. The abuse cases are real and they belong later in this pillar; the difficulty comes first, because a reader who thinks the answer is obvious will misread the disclosures that exist precisely because it is not.

Why the obvious answers all fail

Four intuitive rules, each wrong. When cash arrives? That is cash accounting, which the previous article showed describes a bank account rather than a business — and it would let a company book three years of subscription income the day a customer prepays. When the contract is signed? Nothing has been delivered; the company has a promise and an obligation, not an earned return. When the goods leave the warehouse? Closer, and still wrong for anything involving installation, acceptance, a right of return, or an ongoing service. When the customer is invoiced? That is a document, and issuing it is entirely within the seller's control. The principle that survives is different: revenue is earned when the company has done what it promised to do. Everything difficult follows from the fact that "what it promised" is often several things, delivered at different times, for one undivided price.

The five-step framework

Both major standard-setters converged on the same model — ASC 606 under US GAAP and IFRS 15 internationally — which is unusual and worth noting, since the later article on GAAP versus IFRS mostly catalogues differences. The model runs in five steps. 1. Identify the contract — an enforceable arrangement with a customer. 2. Identify the performance obligations — the distinct things promised, which may be several inside one agreement. 3. Determine the transaction price — including estimates of variable elements such as discounts, rebates, refunds, and bonuses. 4. Allocate the price to the obligations, normally in proportion to what each would sell for separately. 5. Recognise revenue as each obligation is satisfied — at a point in time for a delivered good, or over time for a service delivered continuously.

Step 2 and step 4 are where the judgement concentrates, and the worked example below shows why: an ordinary equipment sale bundled with a service contract is not one sale but two, priced together and earned apart. Step 3 carries its own difficulty — when part of the price depends on future events, the company must estimate what it will ultimately receive, and estimates of the future are exactly what accrual accounting requires and cash accounting avoids.

Deferred revenue, and the two questions that change everything

When cash arrives before the obligation is satisfied, the company records a liability — deferred revenue — not revenue. It is an unusual liability in that it is generally good news: it represents work already paid for and not yet delivered, which is why a rising deferred-revenue balance often accompanies a healthy order book. Wexford's deferred revenue rose from 58.0 to 70.0 — 12.0 of cash collected ahead of delivery, which appeared in operating cash flow while contributing nothing to revenue. That is the mechanism working correctly, and it is one reason Wexford's cash from operations exceeded its profit.

Two further judgements can change reported revenue by multiples without altering profit by a cent. Principal or agent — gross or net. A company that controls a good before transferring it reports the full amount as revenue and the cost separately; a company that merely arranges the transaction reports only its commission. On the illustration below, the same transaction produces revenue of 300 or revenue of 30 — a tenfold difference — with identical profit of 30 either way. This is the reason revenue-multiple comparisons between businesses with different roles in a transaction chain can be meaningless. Over time or at a point in time. A long construction or engineering contract recognised progressively looks entirely different from the same contract recognised on completion — smooth growth against a lumpy series — and both may be correct depending on the terms.

What a reader should actually do

Read the revenue-recognition policy in the notes before comparing revenue across companies. It states when obligations are treated as satisfied, whether the company reports gross or net, and how variable consideration is estimated — three things that determine what the top line means. Then watch deferred revenue alongside revenue, because the pair together says more than either alone: revenue growing while deferred revenue shrinks means the company is delivering from a backlog it is not replacing, while both growing together indicates demand ahead of delivery. And treat a change in policy or estimate as a question rather than a verdict — standards change, businesses change, and a revised estimate of variable consideration is a normal event that must be disclosed. The reason to know all this is comparability, not suspicion. Two companies with identical economics can report materially different revenue through entirely proper application of the same standard, and an investor comparing them without checking has compared two conventions rather than two businesses.