Revenue Growth: Quality, Durability and What It Costs to Get

Revenue growth can be stated in rupees or as a rate, and the two tell different stories. Sankalp Industrial Systems Limited, invented, adds exactly Rs 1,20,00,00,000 of revenue in every forecast year, so its growth rate falls from 10.00 per cent to 7.14 per cent while the rupees never move at all. Growth is worth something only when the capital that bought it earns more than that capital costs.

The awkward part of this idea is easier to feel outside a bus stand than inside a spreadsheet. Begin on a pavement. A woman sells tea and vada pav from a cart. In her first year she serves about a hundred cups a day. Then she adds a second flask and a boy to carry trays into the two office buildings across the road, and she starts serving a hundred and fifty. Fifty more cups a day. The next year she adds another flask and another round of buildings, and gets to two hundred. Fifty more cups a day. Then two hundred and fifty. Fifty more cups a day.

Asked how the cart is doing, she says it is doing the same thing every year: fifty more cups a day, every year, without fail. Asked the same question, her accountant says growth went from fifty per cent to thirty three to twenty five to twenty. One of them sounds like a business repeating a feat. The other sounds like a business running out of road. Neither of them is lying, and neither of them has made an arithmetic error.

The gap between what a business did and what its percentage shows is the whole subject of this guide. The gap is not a curiosity. The gap is the most common misreading in the whole of forecasting, it decides what people write in notes and what they assume about the years after a forecast ends, and it takes exactly one extra row of a table to catch.

What is revenue growth, and why do two people describe the same forecast differently?

Revenue growthThe change in a company's top line, stated either in rupees added or as a percentage of the previous year. is the change in the top line, and it can be written down in two entirely different currencies. Growth can be written in rupees, as an amount added. Or growth can be written as a growth rateThe rupees added divided by the previous year's revenue., the same rupees divided by what the company already had.

Most published material picks the second and never mentions the first. Choosing the rate alone is a habit rather than a rule, and the habit has a cost. The two are not the same measurement wearing different clothes. One is an amount. The other is a ratio, and a ratio has a denominator that moves.

Sankalp Industrial Systems Limited, an invented listed maker of industrial valves, precision castings and the aftermarket parts and service that go with them, supplies every figure that follows, all of them from a single five-year forecast. Its last completed year, called Year 0, carried revenue of Rs 12,00,00,00,000. The five forecast years run Rs 13,20,00,00,000, Rs 14,40,00,00,000, Rs 15,60,00,00,000, Rs 16,80,00,00,000 and Rs 18,00,00,00,000.

Now state that forecast the two ways. In rupees added: Rs 1,20,00,00,000, then Rs 1,20,00,00,000, then Rs 1,20,00,00,000, then Rs 1,20,00,00,000, then Rs 1,20,00,00,000. As a rate: 10.00 per cent, then 9.09, then 8.33, then 7.69, then 7.14. The rupees are identical in all five years and the rate falls by nearly three points, and both rows describe exactly the same forecast.

The drawing works in a particular way. The two panels sit apart, one above the other, sharing nothing but the row of years along the bottom. A shared frame would imply that one line explains the other, so the two panels are not laid over each other on a pair of axes. Neither line explains the other. The upper picture and the lower picture come from the same five subtractions. Only the second one has been divided by something.

Why does a growth rate fall when the rupees added never change?

Because a growth rate is a fraction, and this fraction has a numerator that is frozen and a denominator that grows.

Write the two ends of the forecast side by side. Year 1 growth is Rs 1,20,00,00,000 over Rs 12,00,00,00,000, or 10.00 per cent. Year 5 growth is Rs 1,20,00,00,000 over Rs 16,80,00,00,000, or 7.14 per cent. The top of the fraction is the same figure in both. The only thing that has changed between them is that the company is bigger. Being bigger is a consequence of the growth having happened rather than a sign of it stopping.

The fall has a name worth knowing, and the name disposes of an argument in three words. The name is a base effectThe fall in a growth rate caused only by the denominator rising.: the fall in a rate caused only by the denominator getting larger. A base effect is pure arithmetic and carries no information whatever about the business underneath it. The cart selling fifty more cups a day meets exactly the same wall. So does a household whose savings rise by the same amount each year: put aside sixty thousand rupees on top of six lakh and the pile grows ten per cent, put aside the same sixty thousand on top of twelve lakh and it grows five, and the household has not become worse at saving.

What does the five-year table look like when both rows are printed?

The discipline that catches all of this costs one row. The rupees belong beside the rate. Always. Not instead of the rate, and not as a footnote.

Two rows, one forecast. A reader given only the third row watches a business decelerate for five straight years. A reader given only the second row watches a business repeat itself five times. The second row is the one almost nobody prints, and it is the one that settles the question in a glance. If the rupees are flat, nothing decelerated, whatever the percentage row is doing.

The middle row is worth being precise about. The row is not a claim that adding the same amount every year is easy, or that any company will do it. The row is a description of what this invented forecast assumes. The point is only that the description in rupees and the description in per cent are both faithful, and that reading one without the other loses information that cannot be recovered.

What does a compound annual growth rate hide?

Sooner or later somebody compresses the five years into one figure, and the figure they reach for is the compound annual growth rateThe single constant rate that would take the first year's revenue to the last year's over the same period.. A compound annual growth rate is the constant rate that would have carried the first figure to the last one over the same number of years. The working is short and the result is instructive.

Revenue rises from Rs 12,00,00,00,000 to Rs 18,00,00,00,000, a multiple of 1.50 times. Five years elapse, so the rate wanted is the fifth root of 1.50, less one. The fifth root of 1.50, less one, is 8.447177 per cent before rounding, printed here as 8.45 per cent. The figure is derived from the two locked revenue endpoints rather than read off a record, and every later use of it is computed on the unrounded 8.447177 rather than on the printed 8.45.

Now put 8.45 per cent beside the five years it claims to summarise. The years grow 10.00, 9.09, 8.33, 7.69 and 7.14 per cent. The compound annual growth rate is a constructed average rather than an observation, so it matches none of them and was never going to. It sits 0.11 points above Year 3 and 0.64 points below Year 2, in a gap where no actual year lives.

The second thing the summary hides is the shape of the path, and this one is easier to see than to say. A compound rate of 8.447177 per cent applied to Rs 12,00,00,00,000 gives a smooth curve that touches the actual forecast at exactly two places, the start and the finish, and runs below it everywhere in between. At Year 3 the smooth path reads Rs 15,30,50,94,008 against the forecast's Rs 15,60,00,00,000, a gap of Rs 29,49,05,992. The gap of Rs 29,49,05,992 is not an error in either figure. The gap is the price of describing a straight-line addition with a compounding rate.

The last line in the drawing carries a small lesson of its own. Rebuild the compound path using the printed 8.45 per cent rather than the unrounded 8.447177, and Year 5 finishes at Rs 18,00,23,42,822 instead of Rs 18,00,00,00,000. Rs 23,42,822 has appeared out of nothing but a rounding decision. Round once, at the end, from the full value, and never rebuild one figure out of another figure's printed form.

What does a rupee of new revenue actually cost to buy?

So far the work has only rearranged one row of a forecast. Now for the part that decides whether any of it is worth having, and it starts with a question the percentage row cannot answer: what did the company have to spend to add that Rs 1,20,00,00,000?

Revenue is the one line in a forecast that can be raised without anybody having to explain how. Nobody has to buy anything to type a larger number into a revenue row. Revenue is therefore the assumption made most carelessly, and the useful discipline is to insist that every rupee of extra revenue is attached to the capital that bought it.

For this company the attachment is exact. Net new invested capitalCapital expenditure less depreciation plus the movement in net working capital. is capital expenditure less depreciation, plus the movement in net working capital. How that figure is built and why it is measured net rather than gross is covered separately; take it here as given. In Year 1 the company spends Rs 1,34,80,00,000 of capital expenditure against a depreciation charge of Rs 52,80,00,000, a difference of Rs 82,00,00,000, and adds Rs 18,00,00,000 of working capital, for a total of Rs 1,00,00,00,000.

Run the same three lines for the other four years and the answer does not move. Capital expenditure rises to Rs 1,39,60,00,000, Rs 1,44,40,00,000, Rs 1,49,20,00,000 and Rs 1,54,00,00,000. Depreciation rises to Rs 57,60,00,000, Rs 62,40,00,000, Rs 67,20,00,000 and Rs 72,00,00,000. Capital expenditure has been set at a falling share of a rising revenue, 10.21 per cent then 9.69, 9.26, 8.88 and 8.56, exactly so that the gap over depreciation holds still. The difference is Rs 82,00,00,000 in every single one of the five years. Add the flat Rs 18,00,00,000 of working capital and Rs 1,00,00,00,000 falls out five times.

Rs 1,00,00,00,000 of net new capital goes into the ground and Rs 1,20,00,00,000 of new revenue comes out, so a rupee of new revenue costs about 83 paise of capital, and the capital is spent before the revenue exists. Read the exchange the other way and it is cleaner still: a rupee of new capital buys Rs 1.20 of new revenue.

Turns are something a reader can argue with, so the 1.20 is worth stating as a number of turns rather than as a vague claim that growth costs money. The existing business carries Rs 12,00,00,00,000 of invested capital against Rs 12,00,00,00,000 of revenue, or 1.00 turn. The new capital is assumed to spin 1.20 times. The forecast is therefore assuming that new capacity turns capital into revenue twenty per cent more efficiently than the capacity already installed, and no evidence for that is recorded anywhere.

An assumption of 1.20 turns is a much more specific statement than saying new capital does better, and far easier to interrogate. What would have to be true for 1.20 turns to hold? Newer machines running at higher utilisation, or a product going out of the door with less inventory sitting behind it, or a customer who pays sooner. Each of those is checkable in principle. None of them is recorded here, so the correct thing to write is that the forecast assumes it and that the assumption is unsupported.

How much of new revenue is consumed by working capital before anything else?

Part of what growth costs never looks like growth at all, and it never appears on a capital expenditure line. Part of the cost is stock on a shelf and an invoice nobody has paid yet.

Net working capitalReceivables plus inventory less payables, the money tied up in running the business. is receivables plus inventory less payables, the money tied up simply in running the business day to day. Sankalp holds it at a flat 15.0 per cent of revenue throughout the forecast, so its working capital intensityNet working capital as a share of revenue, here held flat at 15.0 per cent. never changes. The flat ratio has an immediate consequence: every extra rupee of revenue drags fifteen paise of working capital along behind it. Rs 1,20,00,00,000 of extra revenue therefore needs Rs 18,00,00,000 of extra working capital, precisely the movement in the forecast.

The tea cart shows this more plainly than any balance sheet. To serve fifty more cups a day she has to buy more milk, more sugar and more leaves, and that stock sits in her cart overnight whether or not the fifty extra customers turn up. When the offices across the road start settling weekly rather than paying at the counter, more of her sales stop being cash at the moment she makes them. Both of those are money she has already parted with, funded out of what she had earned before.

Rs 18,00,00,000 of the Rs 1,00,00,00,000 that growth costs each year is working capital rather than fixed assets, eighteen paise in the rupee, and it is spent before the revenue arrives rather than after. That timing is the part people miss. Inventory sits on the floor and the receivable sits outstanding whatever the growth turns out to be.

What separates growth that adds value from growth that adds none?

The sentence that settles all of this is not about growth at all.

Koller, Goedhart and Wessels put growth beside the return on invested capital, in one expression, for a specific reason: growth on its own carries no information about whether a business is becoming more valuable. Two companies can print the identical revenue line and be entirely different objects. The quality of growthWhether the capital that produced the growth earns more than that capital costs. is a question about the capital that bought the growth, never about the growth itself.

Stated as a test: does the capital that produced this revenue earn more than that capital costs? If it does, adding revenue adds to value. If it does not, adding revenue subtracts from it, and the faster the company grows the more it subtracts. The revenue line looks precisely the same in both cases, and the test can therefore never be applied to the revenue line.

Sankalp carries a weighted average cost of capital of 12.00 per cent. The 12.00 per cent is the company's own locked figure, used here as a given. How a cost of capital is built, and what a weighted average of it means, is covered separately. The measures that answer the test, being the return on the capital already in the ground and the single figure that nets a capital charge off profit, are also covered separately. The question stands here, and the answer does not.

One line about the relationship, and then it is handed on. Growth, reinvestment and what new capital earns are tied together by a single identity: how much a business puts back in, multiplied by what that capital earns, is what it grows. Computing that identity, checking it and using it are set out under the growth equation.

What makes revenue growth durable rather than a single strong year?

The other half of what people mean by quality is durabilityHow many years a growth rate can be repeated rather than how high it is in one year., and it is a different question from how large the number is. Durability asks how many years a rate can be repeated, not how high it climbed once.

A stall that gets a wedding order one December has a spectacular December. A stall that has signed a contract to feed a factory canteen every day has something else entirely, and the difference between them is invisible in a single year's revenue figure. The everyday version is a household with one salary against a household with three. The annual total can be identical, and the two are still not remotely the same object. One of them stops entirely if a single thing goes wrong.

So durability is a set of questions put to a growth figure. Every one of the questions below is answerable in principle from a company's own disclosure, and not one of them is answerable from the record behind this guide. The absence is not a defect in the questions. Recording the absence is the honest position, and saying so is more useful than filling the gap.

The right-hand column refuses. The record behind this guide locks the rupees of revenue and locks nothing about who pays them or on what terms, and an answer to any of those five rows would be inventing a fact into a forecast that several other calculations rest on. Writing that a figure is not recorded is a real finding, and it travels better than a confident guess.

What does a forecast have to say about where the extra revenue comes from?

The same refusal, applied to the sharpest question of all. Rs 1,20,00,00,000 of extra revenue arrives in Year 1. Where does it come from?

There are only four broad answers a forecast can give. More units at the same price. The same units at a higher price. A different mixThe share of revenue coming from each part of a business, which can move a total without any part moving., meaning more of the revenue coming from the parts that carry a higher price without any individual price moving. Or revenue bought by acquiring another business rather than built by this one. Each has completely different consequences for what happens to profit, to capital and to whether next year looks like this one.

The record locks none of them. There is no volume figure in it, no price assumption, no split by product and nothing about an acquisition. A forecast that cannot say which of the four it is has stated a number rather than made a case. So the correct thing to write is that the source of the growth is not recorded, naming the four possibilities and picking none of them.

What does a growth assumption look like beside other people's?

One last comparison, and then the subject is handed on. A growth assumption is not a fact about a company. A growth assumption is a position somebody has taken, and the cheapest way to make the position visible is to print it beside the positions other people have taken.

The record carries six invented industrial companies with forecast revenue growth from 4.0 per cent to 18.0 per cent. Sankalp's own 10.00 per cent sits inside that range, above five of the six and below one. The position in that range is the entire statement. No assumption becomes reasonable, aggressive, conservative or achievable by sitting where it sits on a line, because none of those verdicts is knowable from a position among other numbers. How such a set of companies is chosen in the first place, and what their multiples imply about anything, is covered separately.

Where does reading a growth line go wrong in practice?

Almost never through carelessness. The mistake below is made by people being careful, working from a correctly built table, and it produces a note that reads perfectly well.

What does none of this establish?

The honest boundary of the subject is narrower than it might feel by now.

Everything above shows that a growth rate and an amount of growth are different measurements and that one of them can fall while the other holds still. The arithmetic shows what this particular invented forecast assumes about the capital behind its revenue, and an assumption is not an outcome. Rs 1,20,00,00,000 a year for five years is a typed assumption, and printing it in two forms rather than one makes the assumption clearer without making it more likely.

Nor does the exchange rate of 1.20 turns establish anything about efficiency. The 1.20 turns is read off the forecast's own arithmetic, so the forecast asserts the figure rather than demonstrating it. A set of accounts can show what the capital already installed produced. No accounts can show what capital not yet spent will produce, and every forecast in the world has to guess at that.

One reading rule does stand: a falling growth rate is not by itself evidence of anything, and the row that settles what it is evidence of costs one line of a table.