When Revenue Multiples Apply, and When They Don't
Most of the damage is done before anyone argues about the number itself, in three questions nobody asked.
Summary
- Revenue multiples are a fallback for when earnings cannot carry the valuation, not a starting point. There is one real precondition: whether you can form a defensible view of steady-state margin.
- Gross margin, growth, revenue quality and retention do not qualify a business for the method. They set the level of the multiple, and a low-margin business growing slowly earns a low one.
- Most errors are denominator errors. Trailing twelve months for anything transactional, net revenue rather than GMV, and no valuation built on users, downloads or disbursements.
Revenue multiples are the most misused number in Indian valuation conversations.
They get applied to businesses that have no business being valued that way. They get quoted at scales where the multiple was derived backwards from a price someone had already decided. And they get attached to revenue lines that overstate what the company actually earns.
The multiple itself is rarely the problem. Three prior questions went unanswered. Should this business be on a revenue multiple at all? Has it reached a scale where the multiple means anything? And which revenue figure is being multiplied?
This is the third piece in a series on benchmarking. The first covered how to frame a benchmarking exercise; the second, the metrics that carry a diagnosis.
There is a hierarchy, and revenue sits low in it
Revenue multiples are what you use when earnings cannot carry the valuation. They are a fallback rather than a preference, and treating them as a starting point is how most valuation arguments lose credibility with a mid-market buyer.
Earnings multiples stop working in four situations.
Margins are thin. The arithmetic explains why. EV/Revenue equals EV/ multiplied by margin. At a low margin, a small move in margin swings the implied earnings multiple substantially, and the multiple becomes hypersensitive to noise that has nothing to do with the business.
Earnings are too small to be stable. Below a certain absolute size, EBITDA moves with a single hire, a single contract or a single provision. The buyer universe also thins to the point where comparable multiples are not observable anyway.
Earnings are not converting to cash. Where operating cash flow consistently runs well below EBITDA, pricing the earnings means pricing an number. This is a quality problem, and switching to revenue does not solve it.
Margins are volatile across the cycle. Commodity-linked and project-lumpy businesses produce an EBITDA figure that depends heavily on which year you pick. The answer here is normalised or mid-cycle earnings, not a switch to revenue. A volatile earnings profile is an argument for normalising earnings, not for abandoning them.
What actually determines whether a revenue multiple applies
It is tempting to write down a set of qualification tests: a gross margin floor, a growth floor, a recurring revenue share. Then treat a business as eligible only if it clears all of them. That framing is wrong, and it is worth being explicit about why, because it is common.
A revenue multiple is an earnings multiple in disguise. EV/Revenue equals EV/EBITDA multiplied by margin, always. Applying a revenue multiple is therefore an implicit assertion about the margin this revenue will eventually produce. There is only one real precondition.
Can you form a defensible view of steady-state margin?
Where you can, because the model is well understood, gross margin is visible and stable, and the cost structure at scale is predictable from comparables, a revenue multiple works. Where you cannot, no revenue multiple is defensible at any level, however attractive the growth looks.
Everything else that gets written up as a threshold is not a gate. It is what determines the level of the multiple.
Six drivers that set the level of a revenue multiple. None of them is a gate on the method.
| Driver | Pushes the multiple down | Pushes it up |
|---|---|---|
| Gross margin | Low, or falling with scale | High, and stable or improving |
| Growth | Slow, or decelerating sharply | Fast and sustained across several periods |
| Revenue quality | Project, one-off, transactional | Contracted, recurring |
| Retention | Net revenue retention below par | Expansion exceeding churn |
| Cost behaviour | Cost base scales in lockstep with revenue | Margin expands as the business grows |
| Predictability | Volatile, cyclical, concentrated | Visible, diversified |
Read this way, a low-margin business growing slowly does not fail a test. It earns a low multiple, while a high-margin business growing quickly earns a high one. Both are valid applications of the same method. The mistake is treating the drivers of the level as conditions of the method.
Two corollaries follow.
Low-margin businesses do get revenue multiples. Staffing, distribution and logistics businesses transact on revenue multiples routinely, at low levels. The point is not that they are ineligible. It is that at those levels the multiple carries almost no information: it is a restatement of margin, and the real drivers of value sit in capital employed and returns. Use it as a cross-check, never as the primary basis.
EV/gross profit is not a consolation prize. It is the better presentation whenever gross margin varies widely across the peer set. It strips out the single largest source of false variance between business models. That applies to a high-margin peer set with dispersion just as much as to a low-margin one. Treating it as the fallback after failing a margin test understates its usefulness, and it is badly underused in the Indian mid-market generally.
The one genuine condition beyond steady-state margin applies only to loss-making businesses: is the loss discretionary? A company losing money because it is choosing to spend on acquisition is one proposition. A company losing money because the unit itself does not work is another. Positive contribution with a period the business can fund is the usual evidence. Where the unit is negative, there is no steady-state margin to assert, and the first condition fails anyway.
How big does a business have to be before the multiple means anything?
Steady-state margin can be estimable and the multiple still be meaningless, because the business has not reached a scale at which any multiple is observable. Below that point, valuation is set by round dynamics: norms, round size, investor ownership targets. Any multiple quoted afterwards is reverse-engineered from a price, and it is not a valuation conclusion.
The threshold is not universal, and the principle behind it is simple: it rises as the gap widens between reported revenue and durable economics.
Where a rupee of revenue is close to a rupee of value-bearing revenue, the multiple becomes meaningful early. Subscription businesses clear first, because recurring revenue is self-evidencing and gross margin is stable from the outset.
Where a rupee of reported revenue is a much smaller amount of contribution, and that ratio is still moving with scale, the threshold is substantially higher. Consumer and D2C businesses sit here: freight, returns, discounting and acquisition cost all improve with volume, so early-stage margin is not representative of anything. Marketplaces are similar, because and liquidity stabilise only with volume. Product businesses with lumpy, non-recurring revenue need several cycles before a run-rate means anything at all.
Tech-enabled services rarely justify a revenue multiple at any scale. The test is whether gross margin is visibly rising as the business grows. Where it is not, it is a services business and should be valued as one, on gross profit or earnings.
And for some models the question of scale does not arise, because a revenue multiple is simply the wrong measure.
Three models where no revenue scale makes a revenue multiple appropriate.
| Model | Why a revenue multiple does not apply |
|---|---|
| Manufacturing | Value sits in capital employed and returns, not in turnover |
| Distribution and trading | The model earns through turns, not margin; a revenue multiple prices the wrong thing |
| Lending | Priced on book value against return on equity |
These are not high-threshold cases. They are wrong-metric cases. No revenue scale makes a revenue multiple appropriate for a business whose economics live in the balance sheet.
Which revenue figure is being multiplied?
This is where most errors actually occur, and they are errors of denominator rather than multiple.
Forward versus trailing. Forward ARR is acceptable where it is contracted. Exit run-rate works for subscription and not for transactional models. For anything transactional, use trailing twelve months, never a strong quarter annualised. A run-rate needs several consecutive periods of consistency before it is credible, and a company that annualises a festive-season spike will not defend that number in diligence.
Net, not gross. Net of returns and discounts for consumer businesses. Take rate rather than GMV for marketplaces. Excluding one-off implementation and hardware revenue in software. Net interest income and fees, not disbursement, in lending.
GMV, users, downloads and disbursements are not valuation denominators. They are narrative, they occasionally price a round, and they do not survive a diligence process.
Which line does the multiple attach to, by business model?
Startups are valued on multiples more often than mature businesses, which makes the denominator question sharper rather than softer. Two things vary by archetype: which line the multiple attaches to, and what evidence supports a steady-state margin view. The first is close to non-negotiable; the second sets the level.
Which revenue line a multiple attaches to by archetype, the evidence it rests on, and the fallback basis.
| Archetype | Multiple attaches to | Evidence for steady-state margin | Better basis where that evidence is absent |
|---|---|---|---|
| Subscription / SaaS | ARR | Gross margin, net revenue retention, CAC payback | EV/gross profit |
| Marketplace | Net revenue, never GMV | Take rate stability, liquidity, repeat rate | Contribution multiple |
| D2C / consumer | Revenue net of returns | Contribution trajectory, repeat rate, cohort payback | Contribution or EBITDA |
| Tech-enabled services | Revenue | Gross margin rising with scale | Gross profit multiple |
| Lending / fintech | Net interest income and fees | Credit cost, margin, capital adequacy | Book value against ROE |
| Healthcare / clinics | Revenue | Mature-unit economics, occupancy ramp | EBITDA per bed or centre |
| Hardware / deeptech | Revenue | Order book, demonstrated path to gross margin | Asset or milestone-based |
Three things break this in practice.
Gross margin has to be computed honestly. Indian consumer businesses routinely report margin before delivery, packaging and payment costs, and correcting for those changes the margin view materially. It is the same arithmetic as why a 45% margin D2C product loses money on COD, and it changes the multiple that follows from it.
The reference points circulating in the market derive largely from US benchmarks. Indian companies transact well below them at equivalent metrics, so those reference points indicate direction rather than level, and should never be quoted as though they set a price.
And a round that was priced on a metric does not make that metric a valuation basis. Pricing and valuation are different exercises, and conflating them is how a narrative number becomes an anchor nobody can dislodge later.
What to use when no multiple applies
Where none of the above holds, the business is out of multiples entirely. That is not a problem, because it means a different method applies.
- DCF or normalised forward earnings where visibility exists
- VC method or last-round reference for early stage
- Asset or replacement value for asset-heavy and distressed situations
One India-specific note is worth ending on. Mid-market buyers, lenders and promoters anchor hard on EBITDA multiples. Presenting a revenue multiple to that audience without first establishing why earnings fail tends to read as advocacy rather than analysis. And once a valuation argument reads as advocacy, the number stops doing any work at all. Which is why, in deal advisory work, the earnings case gets built first and the revenue multiple is introduced only after it has failed.
The last piece in this series covers why MCA filings are a good starting point for benchmarking, and the work that turns them into comparable numbers.
How useful was this article?
One tap. It tells us what to write more of.
The next one
Get what we publish next, by email.
Working notes on raising, borrowing, protecting, growing and structuring capital in India. One email a week at most, and you can leave any time.
We use your address only to send this. See our privacy policy.
We store your address to send you these emails and nothing else. See our privacy policy.
Related reading
