IFRS 15 Variable Consideration: Estimating and Constraining the Transaction Price

What makes consideration variable under IFRS 15?

Variability that is not in the contract

This is the part that gets missed, and it is worth reading slowly.

Three tests flush this out. Compare cash collected against amounts invoiced for the same customer group over two or three years: a persistent gap settled by credit note is a price concession in disguise. Read the sales approval matrix, because standing authority to grant a settlement discount is a customary business practice. And read what the entity says publicly, since a published returns policy is exactly what IFRS 15.52(a) means by published policies or specific statements.

Local FAQs

Is a penalty for late delivery variable consideration or a provision? If it reduces the amount the customer pays under the contract, it is variable consideration under IFRS 15.51 and it reduces the transaction price. If it is a separate damages payment that does not touch the contract price, IAS 37 applies. Read the clause rather than the label.

Does a prompt payment discount make consideration variable? Yes. Whether the customer takes it is contingent on a future event, which is squarely within IFRS 15.51. Most entities estimate take-up from history using expected value under IFRS 15.53(a), because the population of invoices is large and homogeneous.

Potential risks

The dominant risk is completeness. Entities identify the rebate clause in the master agreement and miss the side letter, the growth incentive agreed by email, the shelf-space payment routed through marketing, and the practice of writing off the last invoice of the year. Each of those reduces revenue.

What are the two methods for estimating variable consideration?

One contract can contain more than one uncertainty, and each gets its own method. IFRS 15.54 requires consistency "when estimating the effect of an uncertainty", singular, so applying expected value to a service credit regime and most likely amount to a one-off bonus in the same contract is not inconsistent. What is prohibited is changing the method used for a given uncertainty part way through.

Local FAQs

How many outcomes make a "reasonable number" under IFRS 15.54? Enough that adding another would not move the answer materially. For a tiered rebate the natural set is one outcome per tier. Building a fifty-point distribution over a three-tier scheme adds cost and no accuracy.

Potential risks

The risk is documentation, not arithmetic. Most entities can produce the number. Far fewer can produce the dated schedule of outcomes and probabilities behind it, linked to the pricing model referred to in IFRS 15.54, without which the estimate cannot be re-performed and the disclosure required by IFRS 15.126(a) has no support.

When does each method fit, and what breaks if you choose wrong?

Why expected value suits a population

IFRS 15.53(a) points at a "large number of contracts with similar characteristics", but the real test is whether the outcome behaves like a population. A twelve-month rebate agreement is one contract, yet the outcome is driven by thousands of separate order decisions, so the distribution of final volume behaves like a population. That is why expected value is the standard answer for volume rebates, sales returns, prompt payment discounts and service credits.

Why most likely amount suits a binary outcome

Where the contract pays a bonus of GBP 300,000 or nothing, the expected value of GBP 180,000 at a 60 per cent probability is a number that will never be paid. Recognising it means the entity is guaranteed to record a true-up in one direction or the other. IFRS 15.53(b) accepts that a single-point outcome is better predicted by the single most likely outcome, precisely because there is no population over which the average can be realised.

Practitioners sometimes reach for expected value on a binary bonus because it produces a smaller, apparently more prudent number. That confuses the estimate with the constraint. Prudence in IFRS 15 lives in IFRS 15.56, not in the choice of method. The correct route is most likely amount followed by a constraint assessment that may reduce the included amount to nil.

Local FAQs

Potential risks

Expected value on a binary outcome produces revenue certain to be wrong until the uncertainty resolves, and disguises a live judgement as an average. Most likely amount on a population ignores the tail, understating rebate accruals in a growing business and overstating them in a shrinking one. Both errors show up in the refund liability roll-forward before they show up in revenue.

How does the constraint on variable consideration work?

That last point is the one most often applied incorrectly. The test is run against cumulative revenue recognised on the contract, so the same variable amount can be constrained at inception and unconstrained later without any change in the uncertainty itself. A GBP 300,000 bonus on a contract where GBP 200,000 of revenue has been recognised is a potential reversal of 150 per cent of the amount recorded. On the same contract at 90 per cent completion, with GBP 2,160,000 recognised, the same GBP 300,000 is a potential reversal of 14 per cent. The magnitude limb of the test moves as the contract progresses even when the likelihood limb does not.

Working the five factors

(a) Susceptibility to factors outside the entity's influence. The examples given are market volatility, the judgement or actions of third parties, weather and obsolescence. In practice this is the strongest of the five. A bonus dependent on a regulator's certification, a price linked to a commodity index, a milestone requiring a customer's own subcontractor to finish first: in each case the entity's own performance does not determine the outcome. Contrast a bonus for delivering on a date the entity controls with its own resources, which is influenced almost entirely by the entity.

(b) A long period before the uncertainty resolves. Time is corrosive because it multiplies the number of things that can go wrong and because it removes the ability to correct. A rebate resolving in eleven weeks is a different risk from a bonus resolving in three years, even at identical probabilities today. This factor is also why the constraint tends to relax as a contract approaches completion, without any deliberate change in judgement.

(c) Limited experience, or experience with limited predictive value. Note the second limb. Long experience of a discontinued product line has limited predictive value for its replacement. A retailer with fifteen years of store returns data has limited predictive value for its first year of online returns, because the return rate differs by an order of magnitude. Entities frequently claim experience they hold for a different fact pattern.

(d) A practice of offering broad price concessions or changing payment terms. This factor closes the loop with IFRS 15.52(b). If the entity habitually renegotiates, the contract price is a weak predictor of entitlement, and an estimate built on it should be constrained. It is also self-diagnosing: the credit note file tells the answer.

(e) A large number and broad range of possible amounts. Range matters more than count. Ten outcomes clustered between GBP 95 and GBP 100 is a narrow range and a well-behaved estimate. Three outcomes at nil, GBP 500,000 and GBP 5,000,000 is a broad range, and any point estimate drawn from it carries real reversal risk.

How much of the estimate survives

Because IFRS 15.56 says "some or all", the useful way to apply it is to walk down the distribution of outcomes and ask at each level whether cumulative revenue at that level is highly probable not to reverse. Take a rebate scheme where the expected per-unit price is GBP 94.50 but the entity is close to certain that at minimum the middle tier will be reached, giving GBP 95, and the only real question is whether the top tier at GBP 90 is hit. Here the constraint runs the other way: the risk is that the price falls further, so the entity would include no more than GBP 90 if the top tier were highly probable, and would sit at its expected value only where the distribution genuinely supports it. The direction of the risk drives which end of the range is the safe one.

Where the variable amount is an upside, such as a bonus, the constrained amount is the highest amount within the range that clears the highly probable threshold, which is often nil at inception. Where the variable amount is a downside, such as a rebate or a return, the constrained amount is the transaction price after assuming the least favourable outcome that is reasonably supportable. IFRS 15.B23 makes this explicit for returns, requiring the entity to determine the consideration it expects to be entitled to "ie excluding the products expected to be returned", with paragraphs 56 to 58 applied.

Local FAQs

Is "highly probable" a percentage? IFRS 5 Appendix A defines it as significantly more likely than probable, a meaning IFRS 15 borrows without repeating it in its own nine-term Appendix A, and does not put a number on it. Practice tends to sit around 90 per cent or higher, but a numerical policy applied mechanically will produce the wrong answer on contracts where the magnitude limb, not the likelihood limb, is what matters.

Can an amount be constrained to nil forever? Yes, and it will be if the uncertainty never resolves in the entity's favour. What cannot happen is a permanent constraint that is never revisited. IFRS 15.59 requires the assessment at every reporting date.

Does the constraint apply to the fixed part of the price? No. IFRS 15.56 applies only to "an amount of variable consideration estimated in accordance with paragraph 53". Collectability of the fixed amount is dealt with at Step 1 under IFRS 15.9(e) and thereafter under IFRS 9.

Potential risks

The two failure modes are opposite and both are common. Over-constraining hides revenue the entity has earned and creates a lumpy release later that has no operational explanation, which distorts trend analysis and invites questions about earnings management in reverse. Under-constraining recognises revenue that reverses, and the reversal lands in a later period with no offsetting cost, which is exactly the outcome IFRS 15.56 exists to prevent. The audit evidence for either position is the same: a documented walk through the IFRS 15.57 factors, refreshed at the reporting date, that a reviewer can disagree with on the facts rather than on the absence of reasoning.

Worked example: how do you account for a volume rebate using expected value?

Before the numbers, it is worth seeing where the transaction price actually ends up relative to the list price on a typical consumer goods contract. Very little of Step 3 is about the headline price.

The facts

This is a constructed example, not a company's numbers. A components supplier signs a twelve-month agreement with a distributor. List price is GBP 100 per unit, invoiced on shipment. A retrospective volume rebate applies to all units purchased in the year, on this scale:

The supplier has more than 300 similar distributor agreements and eight years of settlement history, so expected value is the better predictor under IFRS 15.53(a). Every unit is a separate performance obligation satisfied at a point in time on delivery, and the contract is a single performance obligation of a series under IFRS 15.22(b) for the purposes of allocation. On the identification question see performance obligations under IFRS 15.

Step 1: estimate the expected per-unit price

At inception, management assesses the probability of each tier from the distributor's own forecast and from history for comparable accounts:

The estimate is a price. That is the entire point, and it is where the previous version of this article went wrong.

Step 2: apply the constraint

Run IFRS 15.57. The outcome depends on the distributor's ordering, which the supplier influences through service levels but does not control, so factor (a) is present but mild. The uncertainty resolves within twelve months, so factor (b) is weak. Experience is extensive and directly comparable, so factor (c) is absent. The supplier does not have a practice of granting concessions outside the scale, so factor (d) is absent. The range runs from GBP 100 to GBP 90, a spread of 10 per cent across three defined outcomes, so factor (e) is weak. On these facts it is highly probable that recognising revenue at GBP 94.50 per unit will not cause a significant reversal of cumulative revenue. The estimate is not further constrained. Transaction price per unit is GBP 94.50.

Step 3: the journals across the year

Cost of sales is ignored in this example to keep the revenue mechanics visible. All amounts in GBP.

Cumulative position at the half year: revenue GBP 661,500, refund liability GBP 38,500, receivable GBP 700,000. The liability equals 7,000 units at GBP 5.50, the expected rebate per unit.

Step 4: the estimate changes, and IFRS 15.88 does the work

In quarter 3 the distributor ships 5,000 units, taking cumulative volume to 12,000. The bottom tier is now impossible. Management reassesses under IFRS 15.59 and reweights: 40 per cent chance the year closes between 10,000 and 14,999 units at GBP 95, 60 per cent chance it exceeds 15,000 units at GBP 90.

Revised expected price = (0.40 × 95.00) + (0.60 × 90.00) = 38.00 + 54.00 = GBP 92.00.

IFRS 15.88 requires the change to be allocated on the same basis as at inception and recognised as revenue, or a reduction of revenue, in the period in which the transaction price changes. So the new price is applied to cumulative units and the difference falls into quarter 3:

Refund liability now stands at 38,500 + 57,500 = GBP 96,000, which is 12,000 units invoiced at GBP 100 less cumulative revenue of GBP 1,104,000. The liability reconciles to the balance sheet without a plug, which is the test that the catch-up has been booked correctly.

Step 5: the uncertainty resolves

In quarter 4 the distributor takes 4,000 units, closing the year at 16,000 and triggering the 10 per cent tier. Price is now known at GBP 90.00.

Refund liability closes at nil: 96,000 + 64,000 = 160,000, extinguished by the credit note. Revenue for the year totals 283,500 + 378,000 + 442,500 + 336,000 = GBP 1,440,000, which is 16,000 units at GBP 90.00. The full-year answer is exactly right, and it was approached in steps rather than in one jump at the year end.

Local FAQs

Should the refund liability be netted against the receivable? No. They are different items with different counterparties in substance, and IAS 32 offsetting conditions are rarely met before the credit note is issued. Presenting them gross also preserves the disclosure of the returns and refunds obligation contemplated by IFRS 15.126(d).

Is the rebate accrual a provision under IAS 37? No. IAS 37.5 scopes out items covered by another standard, and a refund liability arising from a contract with a customer is measured under IFRS 15.55. Presenting it within provisions obscures the link to revenue and breaks the roll-forward.

What if the rebate is prospective rather than retrospective? A prospective rebate that applies only to future units usually does not make past units variable. It is more often a material right to a discount on future purchases, a separate performance obligation under IFRS 15.B40, which is allocation rather than estimation. That analysis sits in allocating the transaction price.

Potential risks

Two things reliably break in practice. The first is that the estimate is refreshed only annually, so quarters one to three are recognised at a price everyone knows is stale, and the whole correction lands in quarter four. IFRS 15.59 requires an update at the end of each reporting period. The second is that the rebate is computed on invoiced volume in the rebate system and on shipped volume in the revenue system, so the refund liability and the revenue deduction never agree and the difference is written off as an unreconciled balance. That difference is usually cut-off, and it is usually revenue.

Worked example: how do you apply the constraint to a performance bonus?

The facts

Constructed example. An engineering contractor agrees to design, build and commission a bespoke plant control system. The fee is GBP 2,400,000 fixed, plus a single bonus of GBP 300,000 payable only if the system passes the independent regulator's site acceptance test at the first attempt. There is one performance obligation satisfied over time under IFRS 15.35(b), because the contractor's performance creates an asset the customer controls as it is created. Progress is measured on an input basis using costs incurred against total expected costs of GBP 1,800,000. Billing is on milestones, so the debit is a contract asset rather than a receivable until the right to consideration becomes unconditional. On that distinction see contract assets and contract liabilities.

The bonus is either earned in full or not at all. Two outcomes, so most likely amount is the better predictor under IFRS 15.53(b). Management's assessment at inception is that a first-time pass is more likely than not, at roughly 60 per cent. Most likely amount is therefore GBP 300,000.

The constraint at inception

The estimate is GBP 300,000. The question under IFRS 15.56 is a different one: how much of that GBP 300,000 is it highly probable will not have to reverse.

Add the magnitude limb of IFRS 15.57. In year one, cumulative revenue on this contract will be around GBP 1.2 million. A reversal of GBP 300,000 is 25 per cent of that. That is significant on any reading. The conclusion is that no part of the bonus is highly probable not to reverse. Constrained amount: nil. Transaction price at inception is GBP 2,400,000.

Year 1

Costs incurred GBP 900,000 of GBP 1,800,000 expected, so progress is 50 per cent.

Year 2: the constraint releases

By the year 2 reporting date the factory acceptance test has been passed with no exceptions, the regulator has issued a clean pre-assessment on the design dossier, and both comparable systems delivered by the contractor since inception have passed first time, taking the relevant experience to four. The uncertainty now resolves within four months rather than twenty. Reassessing under IFRS 15.59: factors (a), (b) and (c) have all weakened materially, and cumulative revenue on the contract is now large enough that GBP 300,000 is a much smaller proportion of it. Management concludes it is highly probable that including the full bonus will not cause a significant reversal. The full GBP 300,000 enters the transaction price.

Costs incurred to date are GBP 1,620,000 of GBP 1,800,000, so progress is 90 per cent.

The GBP 1,230,000 splits into two pieces, and being able to explain the split is what separates a supportable number from a plug. The fixed fee contributes (90 per cent less 50 per cent) × 2,400,000 = GBP 960,000, which is ordinary progress. The bonus contributes 90 per cent × 300,000 = GBP 270,000, all of it in year 2 because none was recognised in year 1. Of that GBP 270,000, the portion relating to the first 50 per cent of progress, GBP 150,000, is the cumulative catch-up required by IFRS 15.88 on the part of the obligation already satisfied. It lands in year 2 because that is the period in which the transaction price changed, not in year 1 by restatement.

Year 3: completion

The system is commissioned, the regulator's site acceptance test is passed at the first attempt and the bonus is invoiced. Progress reaches 100 per cent.

Total revenue across the three years is 1,200,000 + 1,230,000 + 270,000 = GBP 2,700,000, which is the fixed fee plus the bonus. Note the shape: the bonus never appeared as a single lump at the moment of payment. It was drawn into revenue from the reporting date on which the constraint released, spread across the progress made by then, which is the point of IFRS 15.88.

Local FAQs

Should the bonus be a separate performance obligation? No. A bonus is consideration, not a promised good or service. Whether it can be allocated to one obligation is a separate question answered by IFRS 15.85, covered below.

If the probability of the bonus is 51 per cent, is most likely amount still GBP 300,000? Yes. Most likely amount takes the single most likely outcome, and at 51 per cent that is the bonus. The thin margin is not reflected in the estimate. It is reflected in the constraint, which at 51 per cent would almost certainly hold the entire amount back.

Does a penalty work the same way in reverse? Broadly yes. A penalty reduces the transaction price, so the reversal risk points upward: the risk is that the entity recognised too much by assuming no penalty. The entity estimates the penalty under IFRS 15.53 and reduces the transaction price by the amount that is not highly probable to be avoided.

Potential risks

The release of a constraint is the highest-risk journal in Step 3, because it produces revenue with no corresponding activity in the period and no invoice. It should be supported by a dated reassessment identifying what specifically changed in the IFRS 15.57 factors. A release that coincides with a period in which the entity narrowly meets a target, and is supported only by an updated probability with no underlying event, is the classic pattern that turns a technical judgement into a management override question.

How is variable consideration reassessed at each reporting date?

Change in estimate, not error

A revised estimate that reflects information arising after the previous reporting date is a change in accounting estimate under IAS 8, recognised prospectively. It is not a prior period error and it does not restate comparatives. The distinction turns on what management knew, or could reasonably have known, at the earlier date. If the rebate accrual was understated because the entity used a rate it already knew was superseded, that is an error. If it was understated because the customer's ordering pattern changed in the following quarter, that is an estimate.

The evidential difference is a dated file. An estimate refreshed at each reporting date, with the inputs available at that date on record, is defensible as a change in estimate. An estimate refreshed once a year and back-solved to the settled amount is not, because there is no record of what was known when.

The interaction with modifications catches people out. A rebate scale renegotiated mid-year is a modification, not a change in estimate, and it goes through IFRS 15.18 to 21 first. Only a change in the transaction price occurring after the modification is allocated under IFRS 15.87 to 89. The sequence matters because a modification can create a separate contract under IFRS 15.20, in which case the old estimate stays with the old contract untouched. The mechanics are set out in contract modifications under IFRS 15.

Local FAQs

Does the reassessment apply at interim reporting dates? Yes. IAS 34.28 requires the same recognition and measurement principles as in the annual statements, so a half-year or quarterly reporting date is the end of a reporting period for IFRS 15.59.

Can a catch-up be spread forward instead? No. IFRS 15.88 says the amount allocated to a satisfied obligation is recognised in the period in which the transaction price changes. Smoothing it over remaining periods has no basis in the standard.

Potential risks

The pattern to watch is a refund liability that only ever moves in one direction, or that moves in a straight line as a fixed percentage of revenue. Both suggest the balance is being accrued by formula rather than estimated. The second pattern to watch is a catch-up recorded in the same period every year, which usually means the estimate is only genuinely refreshed at the year end and the interim reporting dates carry a rolled-forward number.

Why do sales-based and usage-based royalties escape the constraint?

The "later of" test has two limbs and both must be cleared. A licensee's sales in January do not produce revenue if the licence itself has not yet been made available, because limb (b) fails. Equally, a right-to-use licence transferred in full in January produces no revenue until the licensee sells something, because limb (a) fails. In practice limb (a) usually governs for a right-to-use licence and limb (b) usually governs for a right-to-access licence in its early months.

That last row is the one worth reading twice. A minimum guarantee is not variable, so it does not get the exception and it does not get constrained either. It is recognised as the licence obligation is satisfied, over time for a right-to-access licence and at a point in time for a right-to-use licence. Only the amount that exceeds the guarantee is a sales-based royalty within IFRS 15.B63. The full analysis of licence types sits in licensing intellectual property under IFRS 15.

Local FAQs

Does the exception apply to a royalty on a patent licence bundled with ongoing research services? Only if the licence is predominant under IFRS 15.B63A. Where the customer is really buying the research programme, the royalty is estimated and constrained under IFRS 15.50 to 59 like any other variable amount.

Is a milestone payment in a pharmaceutical licence within IFRS 15.B63? No. A development or regulatory milestone is not based on the customer's sales or usage. It is ordinary variable consideration, usually estimated at most likely amount and heavily constrained under IFRS 15.57(a) because approval turns on a regulator.

What about a sales-based royalty on a licence recognised at a point in time? Limb (b) is satisfied on transfer, so revenue follows limb (a): recognised as the licensee's sales occur, which can be years after the licence was transferred.

Potential risks

The main risk is scoping. Applying IFRS 15.B63 to a royalty where the licence is not predominant defers revenue that should have been estimated and recognised, and the error compounds across every period of the arrangement. The second risk is reporting lag: licensees typically report sales one to two months in arrears, so the licensor is estimating the most recent period's royalties. That estimate is not caught by IFRS 15.B63, which addresses whether the sale has occurred, not whether it has been reported. Estimating an amount for sales that have already happened is measurement, and it is required.

How do refund liabilities and rights of return actually work?

Two further paragraphs shape the analysis. IFRS 15.B22 confirms that the entity's promise to stand ready to accept a return is not a separate performance obligation, so no part of the transaction price is allocated to it. IFRS 15.B26 removes like-for-like exchanges from the analysis entirely: swapping one colour or size for another of the same type, quality, condition and price is not a return. And IFRS 15.B27 sends a customer's right to return a defective product to the warranty guidance in IFRS 15.B28 to B33, where the question is whether the warranty is assurance-type or service-type. Those two regimes are covered in warranties, returns and customer options.

Worked example: a right of return with all four legs

Constructed example. An online retailer sells 10,000 garments in December for cash at GBP 60 each. Cost per garment is GBP 24. History over four comparable seasons shows a return rate of 20 per cent within the 30-day window. Returned garments come back saleable, but inspection, repackaging and restocking costs GBP 1.50 per unit. No further write-down is expected on this line.

The GBP 195,000 charged to cost of sales is the GBP 192,000 cost of the 8,000 units expected to be kept plus the GBP 3,000 of expected recovery costs on the 2,000 units expected back. Those recovery costs are expensed at the point of sale because IFRS 15.B25 deducts them in measuring the asset, not when they are later incurred.

Gross margin recognised on the sale is 480,000 less 195,000 = GBP 285,000. Sense check: 8,000 units at a margin of GBP 36 each is GBP 288,000, less the GBP 3,000 of expected recovery costs, which is GBP 285,000. Correct.

Settlement in January

1,850 garments are actually returned. Refunds of 1,850 × GBP 60.00 = GBP 111,000 are paid. Restocking costs of 1,850 × GBP 1.50 = GBP 2,775 are incurred in cash. The returned garments go back into inventory at their former carrying amount of GBP 24.00 each, a total of GBP 44,400.

The GBP 3,375 charge is not a plug. It is 150 units at GBP 22.50, the carrying amount of the return asset attributable to units the entity expected back and did not get. Those units were sold, and their cost belongs in cost of sales.

Both lines agree to the actual outcome, which is the test that the four legs were set up correctly in December.

Local FAQs

Can the return asset be measured at selling price? No. IFRS 15.B25 measures it by reference to the former carrying amount of the product, which is cost, less expected recovery costs and expected value decreases.

What happens if returned goods cannot be resold? The expected decrease in value comes off the asset at the point of sale under IFRS 15.B25. Where returns are expected to be scrapped entirely, the asset is nil and the whole cost goes to cost of sales immediately, while the refund liability is unaffected.

Is a right of return a performance obligation? No. IFRS 15.B22 says the promise to stand ready to accept a return is not a performance obligation in addition to the obligation to provide a refund, so no transaction price is allocated to it.

Does the constraint apply to returns? Yes. IFRS 15.B23 expressly applies paragraphs 47 to 72, including the constraint in 56 to 58, in determining the amount the entity expects to be entitled to.

Potential risks

The most common defect is the missing return asset, which leaves revenue correct and margin wrong, and which is invisible in a revenue-focused review. The second is a return rate applied uniformly across channels or product categories that behave differently. The third is a refund liability that is presented net of the return asset, contrary to the final sentence of IFRS 15.B25, which removes the reader's ability to see either balance. The fourth is a return period that straddles the year end without the reporting date estimate being cut properly, so December sales carry a return rate estimated on a full-year average rather than on the specific December mix.

When is variable consideration allocated to one performance obligation instead of pro rata?

Worked through on a two-obligation contract: a software vendor licenses a platform for GBP 400,000 and provides two years of hosting for GBP 200,000, with a GBP 50,000 bonus payable if platform uptime exceeds 99.95 per cent in year one. Uptime is a function of the hosting service, not the licence. Criterion (a) is met: the bonus relates specifically to a specific outcome from satisfying the hosting obligation. Criterion (b) holds, because GBP 250,000 for two years of hosting is consistent with what the vendor would charge for a premium-availability service. So the whole GBP 50,000, once it clears the constraint, is allocated to hosting and recognised over the hosting period.

Change one fact. Suppose the bonus was GBP 250,000 and the stated hosting fee was GBP 50,000, with the licence at GBP 550,000. Criterion (a) is still met on the drafting. Criterion (b) now fails, because allocating GBP 300,000 to a two-year hosting service whose stand-alone selling price is nowhere near that amount does not depict the consideration expected for that obligation. The bonus goes into the pool and is allocated on relative stand-alone selling prices under IFRS 15.74. The full allocation mechanics, including how discounts are allocated under IFRS 15.81 to 83, are set out in allocating the transaction price under IFRS 15.

Local FAQs

Can a change in a variable amount be allocated to one obligation? Yes, but only under the same test. IFRS 15.89 permits a change in the transaction price to be allocated entirely to one or more, but not all, obligations only if the criteria in IFRS 15.85 are met.

Do both criteria really have to be met? Yes. The paragraph says "if both of the following criteria are met". A conclusion resting on criterion (a) alone is incomplete.

Does IFRS 15.85 apply to a single-obligation contract? IFRS 15.75 says paragraphs 76 to 86 do not apply where a contract has only one performance obligation, but adds that paragraphs 84 to 86 may still apply where the entity promises a series of distinct goods or services identified as a single obligation under IFRS 15.22(b) and the consideration includes variable amounts.

Potential risks

The risk is a documented conclusion under criterion (a) with silence on criterion (b), which is not a conclusion at all. The second risk is allocating a variable amount to an obligation that has already been satisfied, which converts a future uncertainty into immediate revenue. That is permissible where IFRS 15.85 is genuinely met, but it deserves scrutiny, because it is the highest-value outcome available from the paragraph and therefore the one most likely to be reached for the wrong reasons.

When does a contract contain a significant financing component?

IFRS 15.61 sets the objective: recognise revenue at the cash selling price, the amount the customer would have paid had it paid cash when the goods or services transferred. It also gives the two indicators of significance: the difference between the promised consideration and the cash selling price, and the combined effect of the expected length of the payment gap and prevailing market interest rates.

IFRS 15.64 sets the rate: the rate that would be reflected in a separate financing transaction between the entity and its customer at inception, reflecting the credit characteristics of whichever party is receiving the financing. It is a customer-specific borrowing rate where the entity is financing the customer, and an entity-specific rate where the customer has paid in advance. Once set, IFRS 15.64 prohibits updating it for later changes in interest rates or credit assessment. IFRS 15.65 then requires the financing effect, whether interest income or interest expense, to be presented separately from revenue.

Local FAQs

Does a customer deposit always create a financing component? No. IFRS 15.62(a) excludes advance payments where the timing of transfer is at the customer's discretion, which covers most gift cards, stored value and prepaid credit arrangements.

Is a retention on a construction contract discounted? Usually not. IFRS 15.62(c) applies where the deferral protects the customer against non-performance rather than providing finance, which is the purpose of a retention.

Is the practical expedient an accounting policy choice? It is applied by contract, and IFRS 15.129 requires disclosure of the practical expedients used, so an entity applying it should say so.

Potential risks

The recurring issue is advance payments on multi-year contracts, where entities apply the expedient by looking at the contract term rather than at the gap between payment and transfer. On a three-year licence paid in full upfront, the gap for the final year's service is close to three years, and the expedient does not apply to it. The second issue is applying the entity's own weighted average cost of capital as the discount rate rather than the customer-specific rate required by IFRS 15.64, which is usually materially different.

When is consideration payable to a customer a reduction of revenue?

This is where slotting fees, listing fees, co-operative advertising allowances and shelf-space payments live. The commercial framing is a marketing cost. The accounting question is narrower: did the entity receive a distinct good or service in exchange, judged against the criteria in IFRS 15.26 to 30? A payment for shelf space at the front of a store generally does not, because the retailer is not transferring anything the supplier could have bought from a third party in the ordinary course. A payment for a defined advertising campaign, with placement, dates and audience specified, and at a rate the supplier could obtain from a media agency, generally does.

Local FAQs

Does a payment to a customer that is also a supplier get netted? Only if the payment is genuinely for a distinct good or service from that party in its supplier capacity, at fair value, under IFRS 15.71. Where the entity trades with a customer in both directions, the analysis is done separately for each leg.

Is an upfront payment to win a contract a revenue deduction or a contract cost? If it is paid to the customer, IFRS 15.70 applies and it reduces revenue, generally spread across the related revenue under IFRS 15.72. Costs paid to third parties to obtain a contract are a different question under IFRS 15.91, covered in contract costs and commissions.

Can the deduction be recognised before the related revenue? No. IFRS 15.72 uses a later-of test, so the deduction cannot precede recognition of the related revenue.

Potential risks

The dominant risk is classification within the income statement rather than measurement. Payments to customers sitting in marketing or distribution costs overstate revenue and overstate expenses by the same amount, leaving profit correct and gross margin and revenue growth wrong. Because the profit line is unaffected, the error can persist for years. The second risk is fair value support for payments claimed as distinct services. IFRS 15.71 puts the burden on the entity, and a rate card produced by the customer is not, by itself, evidence of fair value.

How does non-cash consideration interact with the constraint?

IFRS 15.69 completes the picture for contributed goods or services: where a customer contributes materials, equipment or labour to help the entity fulfil the contract, the entity assesses whether it obtains control of them, and if so accounts for them as non-cash consideration received. This is common on construction and infrastructure contracts where the customer supplies free-issue materials, and it increases both revenue and cost.

Local FAQs

At what date is the fair value measured? IFRS 15 does not specify a single date and practice generally measures at contract inception. What IFRS 15.68 does settle is which subsequent movements go through the constraint and which do not.

Are advertising barter arrangements within IFRS 15? They are, and they are measured at fair value under IFRS 15.66, but IFRS 15.5(d) excludes non-monetary exchanges between entities in the same line of business to facilitate sales to customers, which removes most like-for-like media swaps.

Potential risks

The risk is measuring non-cash consideration by reference to the entity's own cost or to the carrying amount in the customer's books rather than at fair value. Where free-issue materials are involved, the second risk is omitting them entirely, which understates both revenue and cost of sales and distorts any input-method progress measure that runs on costs incurred.

What has to be disclosed about the judgement?

IAS 1 applies until IFRS 18 Presentation and Disclosure in Financial Statements takes effect for annual reporting periods beginning on or after 1 January 2027, at which point the equivalent requirements on sources of estimation uncertainty sit in IAS 8 as amended.

Local FAQs

Does the refund liability need a roll-forward? IFRS 15 does not require one in terms, but IFRS 15.126(d) requires information about how obligations for returns and refunds are measured, and a movement table is the most efficient way to provide it. It is also the disclosure users ask for most often.

Is a boilerplate policy note enough? Not for IFRS 15.126(b). A conclusion that variable consideration is constrained "where appropriate" describes the standard rather than the entity, and it gives a reader nothing to test.

Potential risks

The largest disclosure risk is a note that describes the accounting policy and never reaches the judgement. Regulators reviewing revenue disclosures return to this repeatedly: the requirement is entity-specific information about the estimates actually made, not a restatement of the paragraphs.