The Revenue Build: Forecasting From Drivers Rather Than Growth Rates

A revenue build states revenue as volume multiplied by realisation, so every rupee of growth has a unit standing behind it. A single growth rate does not. The same rise can be mostly volume or mostly price, and those two versions carry different cost lines, different durability and different margins. The build shows the split. The growth rate hides it.

Here is the awkward thing about a revenue forecast. The forecast looks like one number, so it feels like one decision. One number it may be; one decision it is not. Two decisions are welded together inside it, and once they are welded the join is invisible. Somebody writes twelve per cent in a cell, and inside that cell are a claim about how many units will be sold and a claim about what each will fetch. Neither claim is written anywhere. Neither can be argued with. And when the year arrives and the number is wrong, or worse, right for the wrong reasons, nothing has been recorded that anyone could go back and correct.

The weld comes apart in three steps: what the two drivers are, where each of them comes from, and what happens to the cost line when the split between them changes. Sarvani Coatings Limited, an invented coatings maker, supplies the three completed years the arithmetic runs on.

What is a revenue build, and what is it replacing?

A revenue build is the sentence revenue equals volume multiplied by realisation, taken seriously enough to be arithmetic. Volume is units sold: litres, tonnes, packs, seats, whatever the thing is counted in. Realisation is rupees per unit, being revenue divided by those units. Multiplying the two gives rupees. The identity is the whole of it, and its power is entirely in the units. A quantity with a unit attached can be checked against the world. A percentage cannot.

Realisation is the average price actually achieved rather than any list price, it is derived rather than disclosed, and that is exactly why it is the term people skip. Nobody publishes it. A company prints revenue at the top of the ladder, and with luck it mentions volume somewhere in the commentary, and realisation is what falls out when one is divided by the other. Realisation carries discounts, it carries the free extra litre in the trade scheme, it carries the fact that the industrial customer negotiates harder than the shop on the corner. All of that is inside one derived rupees-per-unit number that no statement ever states.

Think of a vegetable seller with a cart. At the end of the day the cash box has Rs 4,200/- in it. Rs 4,200/- is his revenue line, and the revenue line tells him almost nothing. He wants to know whether he shifted more kilos than yesterday or simply got a better price for the same kilos. Those two days point at completely different tomorrows. More kilos means he should buy more tomorrow morning. A better price on the same kilos means he should not. The cash box will not tell him which. Only counting the kilos will.

Where do volume and realisation actually come from?

The provenance of the two drivers gets glossed over, and it should not be. The honest answer is uncomfortable. Volume is sometimes disclosed and sometimes not. Volume turns up in management commentary as a sentence rather than a table. Volume occasionally appears in a segmentA part of the business reported separately in the statements, usually because it sells different things or sells to different sorts of customers. Which parts of a business must be reported separately was settled in the accounting material. note or in an investor presentation, in a chart with no axis labels. Quite often it appears nowhere at all, and a reader who wants it has to build it from something else or do without.

Realisation is worse. No company publishes it, anywhere, ever. Realisation exists only as revenue divided by volume. So the moment volume is uncertain, realisation is uncertain in exactly the same proportion and in the opposite direction. Overstating volume by two per cent understates realisation by roughly two per cent, and the product of the two still lands on the published revenue, so nothing looks wrong.

The weakest input in the whole build is the one everything else is divided by, and a build that does not say out loud where its volume figure came from has hidden its largest uncertainty. That is why a serious build carries a provenance note next to the volume row: taken from the commentary at the third quarter, or inferred from a capacity utilisation remark, or estimated, with the word estimated left in. The number is not improved by the note. The note tells the next reader which part to distrust.

Here it matters that guidanceWhat management said in advance about the year to come, in a results call, a presentation or a release. Reading what is committed against what is merely hinted is taken up separately. named volume and left realisation alone. The company said something about how many units it expected to shift and said nothing about what each would fetch. The largest single assumption in the year therefore sits entirely with the reader.

How is a past year split into the two, and why does the order change the answer?

The method is best tried on something whose answer is already known. Only a known answer shows whether the method works. Sarvani Coatings went from Rs 2,120 crore of revenue in year two to Rs 2,415 crore in year three. Volume rose 6.0 per cent. Applied first, that gives revenue of Rs 2,247.2 crore with prices unchanged. The published year is Rs 2,415 crore, so realisation must have risen by a factor of 1.07467, or about 7.47 per cent. Multiplying Rs 2,247.2 crore by 1.07467 lands on Rs 2,415 crore, to the rupee. The two drivers rebuild the published year exactly. Run that check before trusting either driver.

Now the interesting part. The rise was Rs 295 crore. How much of it was volume? Take volume first, applying 6.0 per cent to the old revenue base: Rs 127.2 crore. Whatever is left, Rs 167.8 crore, gets called realisation. Now take realisation first, applying 7.47 per cent to the same old base: Rs 158.3 crore. Whatever is left, Rs 136.7 crore, gets called volume. Two answers for the same year, differing by Rs 9.5 crore, and both arithmetically correct.

The Rs 9.5 crore is not a rounding artefact and it is not an error, it is the part of the rise produced by both drivers moving at once, and a decompositionSplitting a change into named contributions that add back to the whole change. A decomposition that does not add back exactly has left something unnamed. presented without it has quietly handed it to whichever driver went first. Where it comes from is direct: Rs 2,120 crore multiplied by 6.0 per cent multiplied by 7.47 per cent is Rs 9.5 crore. The extra units were sold at the higher price. The honest presentation names three parts, not two: Rs 127.2 crore of volume, Rs 158.3 crore of realisation, and Rs 9.5 crore where they overlap. The three parts add to Rs 295.0 crore with nothing left over.

The household version is a shopping bill. More items were bought and the items cost more. Some of the extra spend is the extra items at last year's prices, some is last year's items at this year's prices, and a small remainder is the extra items at this year's prices, belonging to neither on its own. Nobody argues about that at the till. In a forecast people argue about it constantly. Whoever puts their driver first collects the overlap.

Can two completely different builds print the same revenue?

Two builds can, and that is the strongest argument against treating a growth rate as a forecast. Run two builds forward from the published year three revenue of Rs 2,415 crore. Build A assumes volume up 10.0 per cent and realisation up 1.8 per cent. Build B assumes volume up 2.0 per cent and realisation up 9.8 per cent. Build A gives Rs 2,704.3 crore of revenue. Build B gives Rs 2,704.7 crore. The two builds differ by Rs 0.4 crore. Both round to Rs 2,704 crore.

Now attach a cost line to each. The cost of materialsThe money spent on the physical inputs that go into what is sold: resins, pigments, solvents, packaging. Materials is the first cost line under revenue, and it is taken up properly under cost forecasting. follows units and input prices, not rupees of revenue. Hold the input cost per unitWhat the physical inputs for one unit of output cost, being the materials bill divided by units produced. Input cost per unit moves with commodity prices and with how efficiently the plant runs. rise at 4.0 per cent in both, so the only thing separating them is the volume assumption. Build A: Rs 1,304 crore multiplied by 1.10 multiplied by 1.04 is Rs 1,491.8 crore. Build B: Rs 1,304 crore multiplied by 1.02 multiplied by 1.04 is Rs 1,383.3 crore.

The two builds are indistinguishable at the revenue line and four points of gross marginGross profit as a percentage of revenue, being what is left after the cost of materials and before every other cost. How it is forecast honestly is set out under margin forecasting. apart underneath it, and a forecast stated as a growth rate cannot say which of the two is in hand. Build A comes out at Rs 1,212.5 crore of gross profit, a margin of 44.8 per cent. Build B comes out at Rs 1,321.4 crore, a margin of 48.9 per cent. Rs 0.4 crore of difference at the top. Rs 108.9 crore of difference one line down.

The panel's frozen numbers all come off the published ladder. In year two the materials bill was Rs 1,187 crore on Rs 2,120 crore of revenue, or 56.0 per cent, leaving Rs 933 crore of gross profit and a margin of 44.0 per cent. In year three the bill was Rs 1,304 crore on Rs 2,415 crore, or 54.0 per cent, leaving Rs 1,111 crore and 46.0 per cent. The panel does not touch either revenue figure. The slider asks only how much of the gap between those two years was units rather than price.

Sweep it and read the two ends in words. The finding should survive the panel being switched off. Hold revenue at the published Rs 2,415 crore. If none of the rise was volume, the materials bill would have been Rs 1,230.2 crore and the margin 49.1 per cent. If all of it and more had been volume, at twelve per cent, materials would have been Rs 1,377.8 crore and the margin 42.9 per cent. The span is about six points and Rs 148 crore of gross profit, sitting entirely underneath a revenue line that never changed by a single rupee.

How is revenue built by part of the business, and why is the blend not an average?

Sarvani Coatings reports two lines: decorative and industrial. In year two they were Rs 1,610 crore and Rs 510 crore. In year three, Rs 1,811 crore and Rs 604 crore. Build each separately. Decorative volume rose about 5.0 per cent, so realisation must have risen about 7.1 per cent to reach Rs 1,811 crore. Industrial volume rose about 9.2 per cent, so realisation must have risen about 8.5 per cent to reach Rs 604 crore. The two rebuilt lines sum to Rs 2,415 crore, and that sum is the check.

Now the trap. Decorative volume grew 5.0 per cent and industrial grew 9.2 per cent. Averaging those two gives 7.1 per cent. The published company volume growth is 6.0 per cent. The average is wrong by more than a point, and it is wrong for a reason worth internalising.

A blendA weighted average, where each part counts in proportion to its size rather than counting once. Averaging without weights treats a tiny part and a huge one as equals. is weighted by revenue, not by counting each part once, so a fast growing quarter of the business barely moves the total. Industrial was Rs 510 crore of a Rs 2,120 crore base, or 24.06 per cent. Weighting the two growth rates by those shares gives 6.01 per cent, the published 6.0 per cent rounded. The faster segment, growing 4.2 points quicker, lifted the whole company's volume growth by about 1.01 points above the decorative rate. A quarter of the business can do no more than that.

Readers overstate a mix shift at exactly this point. Industrial went from 24.06 per cent of revenue to 25.01 per cent, a mix shiftA change in which parts of the business make up the sales, so the company's average behaviour drifts even when nothing inside any part has changed. of 0.95 percentage points in a year. Written up as the industrial business growing nearly twice as fast, it sounds transformative. Measured on the size discipline established when competitive position was settled, it is under a point of the sales base. Both sentences are true. Only one of them is proportionate.

What is actually inside a single revenue growth number?

Two assumptions, unwritten. The whole answer is those two, and everything unpleasant about a growth rate forecast follows from them. Typing twelve per cent makes a claim about how many units will move and a claim about what each will fetch. The analyst has simply declined to say which claim is which, or how much of the twelve belongs to each.

Writing a revenue growth number is writing a volume assumption and a price assumption without stating either, neither can be checked afterwards, and when the forecast turns out wrong there is nothing recorded to correct. Compare that with the build. A build says volume up 6.0 per cent, realisation up 7.47 per cent, and both of those are sentences somebody can disagree with today and measure against something twelve months from now. Nothing inside a growth rate is stated, so there is nothing in it to disagree with. A growth rate is a conclusion with its reasoning deleted.

What outside the company can test a build?

Insisting on units pays off here. A realisation assumption is a claim about pricing, so what tests it is the pricing evidence for the whole field: what makers as a group did with price over the period, and what the peer set reported. A volume assumption is a claim about demand, so what tests it is the field's volume record and the demand drivers behind it. Both of those live outside the company, and that is the entire point.

A company cannot be its own control, so the company's own statements test neither assumption. An assumption that realisation rises 7 per cent, checked against the company's past realisation, compares the forecast with the thing it was extracted from. Nothing has been tested. The peer set matters here for exactly this reason: Nandivarman Paints Limited is almost entirely decorative and is the volume leader, Kesaria Surface Solutions Limited is industrial heavy, and Thottam Chemicals Limited sits one step up the chain supplying resins and additives. Each of those is a place to look for evidence the company itself cannot supply.

And here is the honest limit. Even with that evidence, the reason Sarvani Coatings' margin improved cannot be pinned down. A pricing environment across the whole field, this company's own pricing power, and a mix shift towards industrial would all produce what the statements show, and the statements separate none of them. The segment data does narrow it. The mix shift was 0.95 points against a gross margin gain of 2.0 points in the same year, and a gain that size out of a shift that small would need an impossible gap between the two segment margins. Mix is ruled out as the main explanation. Field pricing against company pricing remains, and nothing published settles which of them did the work. Narrowing a question without closing it is what real evidence usually does.

How does an analyst use this on a real morning?

Meghna Iyer covers coatings. Her first move on results day is not to look at the revenue line at all. She goes hunting for volume. Volume is the input that is sometimes given and never derivable, and everything else in her build waits on it. If the release does not carry it, she goes to the presentation. If the presentation does not carry it, she writes the word estimated next to her own figure and leaves it there permanently, so nobody downstream mistakes an inference for a disclosure.

Her rule is that a revenue row never leaves her sheet without a volume assumption and a realisation assumption stated on separate lines above it, each with its unit written out. The revenue growth cell in her model contains a formula, never a number. If somebody asks her why she is at twelve per cent she can answer in two sentences about units and price, both of which can be argued with, and neither of which is a defence of the twelve.

The same habit pays off wherever somebody is lending or committing money against a sales line. Inventory and receivables scale with volume rather than with rupees of revenue, so a lender sizing a working capital limit cares about units. A price led rise reverses when the pricing environment does and a volume led rise generally does not, so an investor comparing this year's rise with last year's cares about which driver did the work. Even a household deciding whether the shop can support a second hand is asking the same thing: are more people walking in, or are the same people simply paying more.

The same habit changes how she reads the consensusThe aggregated expectation of the analysts publishing on a company, usually a mean and a range. How it is assembled and where it is blind is taken up separately. number for the coming year. A mean revenue growth figure across nine estimates is nine welded numbers averaged into one welded number, and the average conceals the split just as thoroughly as each of the nine did. She wants the two or three published notes that state a volume assumption. Only those can be compared with her own.

Where does a revenue build stop?

A revenue build stops one line down. The build produces a revenue figure and a list of assumptions with units attached, and that is the entire deliverable. The build does not produce a cost line. Cost forecasting starts one rung lower and has its own drivers. No margin comes out of a revenue build. Nothing about the shares comes out of one either.

The output of a build is one line and its assumptions, and every temptation to carry it further is a temptation to hide how much is still assumed. The restraint sounds like modesty. It is closer to hygiene. The moment a revenue line is carried into a valuation without its assumption list travelling with it, the assumptions stop being visible and start being structural, and the further downstream the model goes the more confident it looks and the less anybody can check.