Growth Drivers: Volume, Price, Mix and New Markets

Revenue moves for four reasons and they are not four of a kind. Volume is more activity. Price is more revenue from the same activity. Mix is neither: the average rises because the composition of the buyers changed, with nobody charged more and nobody new arriving. A new market is not a growth route at all. Moving a boundary outward changes what gets counted rather than creating anything.

Four things get called growth drivers. Are they four of a kind?

Open almost any account of why a business grew and the same four words appear, set out in a row as though they were four instruments on one shelf: volume, price, mix, new markets. Pulling one moves revenue. The row invites that picture, and the picture is wrong in a way that matters. Two of the four cannot be pulled by anybody, and one of them is not an instrument at all.

Volume and price rest on the two term identity already derived under Revenue Growth vs Monetisation Improvement. Mix is computed below, step by step. And a new market is the odd one out. The fourth is a boundary decision and not a growth route at all.

Feel the difference on something small first. A tuition class took twenty students last term. The class takes twenty five this term: that is volume, and the tutor went and found five people. Or it takes the same twenty and charges each of them more: that is price, and the tutor made a decision and then lived with it. Or it takes the same twenty, charges nobody a rupee more, and five of them move out of the cheap evening batch into the expensive morning one: the term's average fee per student rises, and no meeting anywhere decided that. The third movement is mix, and mix is the one that gets missed. The fourth idea is different in kind: the tutor decides to count the neighbouring town as part of the catchment. Nothing has been taught, nobody has enrolled, and the tutor now has a bigger picture of the class rather than a bigger class.

Volume and price: what is already settled, and where?

The arithmetic underneath the first two legs is derived under Revenue Growth vs Monetisation Improvement and is quoted rather than worked again. A business's revenue is its activity multiplied by its yield: how much business passed through, times how much of that passing business stayed behind. Put the two together and revenue comes out, exactly, every time.

Setu Bazaar, an invented marketplace, makes both terms visible at once. Few businesses do, and that visibility is the only reason the split can be shown at all. The Rs 5,00,00,00,000/- of buyers' money that passed through it in the year is its gross flowThe whole value of what passes through a marketplace in a period, counted before the marketplace takes anything out of it. The flow is emphatically not what the marketplace earns.. The 4.00 per cent of that flow that stayed behind with the marketplace is its take rateThe slice of everything passing through that a marketplace keeps for itself, written as a percentage of the flow. The rest goes to whoever made the sale.. Take that 4.00 per cent of the flow and the revenue line itself falls out: Rs 20,00,00,000/-. The two terms then move one at a time, as they do in the derivation under Revenue Growth vs Monetisation Improvement. Lifting the flow to Rs 550 crore and holding the take rate: revenue reads Rs 22 crore. Holding the flow and lifting the take rate to 4.40 per cent: revenue reads Rs 22 crore again. Moving both together, revenue reads Rs 24.20 crore, and the two separate effects add up to Rs 0.20 crore less than that. The leftover Rs 0.20 crore is a cross termThe corner left over when two things both move and somebody tries to say how much each one contributed on its own. The corner belongs to neither, so it is either shown on its own line or quietly handed to one of them., and it belongs to neither leg.

Valuing each effect at its opening levelThe value a quantity stood at when the period began. Whether a change is measured against the starting level or the finishing one is a choice somebody makes before any arithmetic runs. rather than at the level it finished on is a convention, not a discovery, and the two available conventions were set out by Etienne Laspeyres in 1871 and by Hermann Paasche in 1874. The choice between them is made and explained under Revenue Growth vs Monetisation Improvement, and the convention chosen there is the one used throughout. Activity multiplied by yield returns the revenue line by construction, so the split attributes a change rather than confirming one. That distinction sounds like hair-splitting and is not: the arithmetic can never disagree with the revenue line, so it can never be used as a check on it, and everything it is genuinely worth comes from naming which term the movement came out of.

Why can that split not be run on the second business here at all?

The split needs four numbers and there is no fifth. Both terms in the earlier period, then both terms again in the later one. The whole method is a comparison of two products. Miss any single cell and there is nothing to compute at all.

Now turn to Anjani Stationers Private Limited, an invented maker of school registers, and count what is actually published. Its revenue is published for both years: Rs 2,40,00,000/- and then Rs 2,70,00,000/-, a rise of Rs 30,00,000/- or 12.50 per cent. Its second year is published all the way down: 2,50,000 registers left the works at a realised priceThe average a buyer actually handed over, once discounts and allowances have been taken off. A list price is what was asked for; this is what arrived. of Rs 108.00/-, and those two multiply back to Rs 2,70,00,000/- to the rupee. So the second year gives both of its cells. The first year gives neither. Nothing published carries a register count for that first year, and nothing carries a price for it.

Two of the four numbers exist, so the split cannot be run, and the 12.50 per cent does not say which leg moved. Rs 2,70,00,000/- against Rs 2,40,00,000/- is consistent with more registers at the same price, with the same registers at a higher price, and with any number of combinations in between, including ones where the register count fell. The honest output is the rise, plus the name of the evidence that would settle it.

Then the discipline, stated flatly because this is the point most likely to slip. The 12.50 per cent is a description of two years that happened. The 12.50 per cent is entirely legitimate as a description, it may be quoted, and it may be argued about. The one thing it may not do is travel into a third year. Nobody has observed a third year, and a rate lifted off two observed periods and dropped onto an unobserved one stops being a reading and becomes a projection with a reading's credentials.

So what is mix, exactly?

Setu Bazaar's 50,000 buyers do not all buy the same amount, and how they differ is published. The split is published against revenue, so working it on revenue means nothing at all has to be assumed. Take the two ends of it. Of the 50,000, some 5,000 leave Rs 16,000/- apiece across the year. Multiply out and that band alone accounts for Rs 8,00,00,000/-, exactly 40.00 per cent of the published revenue. The other 45,000 leave Rs 12,00,00,000/- between them, an average of Rs 2,666.67/- each and the remaining 60.00 per cent. Add the two bands and Rs 20,00,00,000/- comes back, the only check available on them and the only one they need.

Two readings fall straight out. The blended averageOne figure covering a whole population, worked by dividing a total by a head count. The figure describes the group, and it is under no obligation to describe anybody in it. across all 50,000 is Rs 4,000/- of revenue a buyer. A heavy band buyer is 6.00 times one of the others and 4.00 times that blended figure. And 10.00 per cent of the buyers produce 40.00 per cent of the revenue. Write the basis into every one of those sentences. The same share measured against the flow would be a different claim about a different quantity, and not the claim published.

One honest note before going further: the other band is really two published groups pushed together, so its Rs 2,666.67/- is an average of two figures rather than anybody's own. Collapsing them loses detail and loses not one rupee. Every band boundary makes the same trade.

Mix is the composition of the activity, meaning which kinds of buyer made it up, and it is a fact about the same buying rather than a third thing standing alongside it. Picture a canteen with two thalis, one at Rs 60/- and one at Rs 110/-. Nothing on the board changes all week. On Thursday more people happen to want the expensive one, and the average bill for the day goes up. The canteen did not raise a price and did not serve more people. The only thing that changed was who ordered what.

The same lesson one level down: what the two published sets hold between them

A second reading of the same three segments is published in goods rather than revenue: a heavy buyer carries Rs 4,00,000/- of goods across the year, a regular one Rs 1,20,000/- and a light one Rs 40,000/-. The three rebuild the published Rs 5,00,00,00,000/- of flow exactly, so the two sets agree with each other. Dividing one set by the other, segment by segment: Rs 16,000/- over Rs 4,00,000/- is 4.00 per cent. Rs 4,800/- over Rs 1,20,000/- is 4.00 per cent. Rs 1,600/- over Rs 40,000/- is 4.00 per cent.

The two sets of published figures hold the take rate identical on every segment. An order is the same size whoever placed it and the same slice is kept off each of them, and that reason is given outright under Revenue Growth vs Monetisation Improvement rather than left to be inferred. The identical take rate matters for one reason only, and the reason comes back once the two term identity is read twice below. A heavy buyer being charged exactly the same fraction as a light one is a real claim about a marketplace, it can be checked in one division, and on a business that did not publish both sets it would have to be assumed instead. Naming it now costs a sentence. Leaning on it silently later costs a conclusion.

What happens when nobody is charged more and nobody new arrives?

Here is the arithmetic of mix, worked one step at a time and open to a check at every step. The construction moves one percentage pointThe plain gap between two percentages. Going from 10 per cent to 11 per cent is one percentage point, and it is also a tenth in relative terms. The two readings are different numbers and both are correct. of the buyers into the heavy band. The heavy band goes from 5,000 to 5,500, or 11.00 per cent of the buyers. The other band goes from 45,000 to 44,500. Now hold everything else absolutely still: the heavy band figure stays at Rs 16,000/-, the other band figure stays at Rs 2,666.67/-, and the total count stays at 50,000.

Now the new blended figure. 5,500 at Rs 16,000/- apiece comes to Rs 8,80,00,000/-. 44,500 at Rs 2,666.67/- apiece comes to Rs 11,86,66,666.67/-. The two add to Rs 20,66,66,666.67/-, and across 50,000 buyers that is Rs 4,133.33/- each. Against the published Rs 4,000/- that is a rise of 3.33 per cent. The head count never moved, so the year's revenue reads Rs 20.67 crore against Rs 20.00 crore, the same 3.33 per cent.

Nobody was charged more, nobody new arrived, and the average rose. That is the whole of mix in one line, and it is worth staying with for a moment because agreeing with the sentence is easy and reading a rising average correctly afterwards is not. An average is a weighted thing. The weights can move while every single figure being weighted stands perfectly still.

Now the label, and it travels with this construction wherever the construction appears. The construction just worked is arithmetic demonstrating a property of an average. The arithmetic has no name, no trade, no country and no year. Nothing published anywhere says Setu Bazaar's composition moved between any two dates. The claim is narrower and more useful: if a composition like this moved, the reported figures would do exactly what is shown above.

One note on where that line stops. Carried one point further, to 13.00 per cent, the same construction makes the year read exactly Rs 22 crore. The two term split reaches that same figure by both of its routes, and the figure has nothing whatever to do with composition. A panel that quietly reproduces a number from somewhere else hands a reader a confirmation that was never earned, and an arithmetic coincidence is the most dangerous kind of error precisely because it arrives looking like agreement. The control therefore stops at 12.00 per cent.

Why can a two term identity not see mix at all?

Take the construction back to the two term split and read it twice. The two readings look symmetrical and they are not, and the difference is what each one costs.

Read one. Call the activity the count of buyers, and the yield the revenue a buyer. The count did not move: 50,000 before and 50,000 after. So the activity effect is nothing, the whole 3.33 per cent lands in the yield term, and the year reads as a pure yield year. Read one assumes nothing at all. Both of its terms are published, and they are published in the same basis the 40.00 per cent sits in.

Read two. Call the activity the flow, and the yield the take rate. The flow rose 3.33 per cent, the take rate held at 4.00 per cent, and the year reads as a pure volume year. Read two costs something, and here is exactly what: it needs the take rate to be the same on a heavy buyer as on a light one. On this marketplace that condition holds and is published, and the segment-by-segment division above was worked to establish exactly that. On a business that published only one of the two sets, the same reading would be resting on an assumption nobody had checked.

One event, two correct decompositions, opposite answers, and the only thing that separated them was a definition somebody chose before the arithmetic started. Not a disagreement about a number. Not an error. A choice of what the word activity was going to mean, made silently, usually by whoever built the spreadsheet first.

Now the sharper half, and the one that matters most. Back at read one a further question arises: what else could have produced Rs 4,133.33/- a buyer? With the flow held at Rs 5,00,00,00,000/- and the head count held at 50,000, the take rate alone is allowed to rise. Set at 31 parts in 750, or 4.1333 per cent with the three running on forever, it brings the revenue to Rs 20,66,66,666.67/-, and across 50,000 buyers that is Rs 4,133.33/- each. The identical figure.

So two entirely different events land on one set of readings. In the first, buyers rearranged themselves between bands and nobody's terms moved a rupee. In the second, nobody moved anywhere and the marketplace took a larger slice of everything. The revenue line, the head count and the revenue a buyer are the same under both, and the arithmetic cannot tell them apart.

Say plainly what this is and what it is not. The limit is not a mistake in the work that built the identity. An identity with two terms in it has exactly two terms, mix is a third fact about the same buying, and a third fact has nowhere to sit. The separation was never encoded in the arithmetic to begin with, so mix arrives inside the yield term, indistinguishable from a charge rise, and no amount of care with the arithmetic will separate them. Naming that limit is the point. Thinking the built split is broken is the wrong lesson. Knowing which question the split cannot answer is the right one.

Is entering a new market a growth driver?

Take the idea seriously for a paragraph. Dismissing it in a clause teaches nothing, and everybody has sat in the meeting where it gets said. Somebody proposes that the business will grow by selling to a set of buyers it does not sell to today: a neighbouring district, a second kind of customer, a category next to the one it already handles. Selling to new buyers is a perfectly sensible thing for a business to do. The arithmetic asks something narrower: which two things must exist before a figure can sit in a column beside that sentence?

Exactly two things. A total for the buying that happens outside the current line. And a share of that total. Without both there is nothing to take a share of and nothing to be a share, so the sentence can be believed but it cannot be counted.

Now check the two against what is published for both businesses. Anjani Stationers sells to 36 accounts and there is no total anywhere for register buying beyond them. Setu Bazaar's flow is a total of its own marketplace and of nothing wider. Neither cell exists. A new market is a boundary decision and not a growth route, and moving the boundary outward does not create revenue, it changes what gets counted.

The honest treatment is to send the question back to the boundary rather than to fill the column. A ladder was built under market size estimation for exactly this: the served set at the bottom, filled, at Rs 2,70,00,000/- across 36 accounts; the rung above it holding a count of 4,00,000 registers as a floor, with no rupee figure against it; and two rungs above that carrying nothing whatever. The two blank rungs are precisely where a new market leg would have to go for its total, and the rungs are blank because nobody has measured them, not because nobody has bothered to write them down.

The professional consequence lands in one line. A plan that lists a new market as its fourth growth driver has put a boundary decision and two measured effects into one column, and the column now adds up to a number nobody can check. Worse, it adds up neatly, and a neat total is exactly what stops anybody checking.

How to Assess Market Growth Drivers

Six steps, in order, and the first one is the one that gets skipped by almost everybody who is otherwise doing this carefully.

One. The analyst writes down what counts as activity, in units of the thing, before touching a single number. Registers, buyers, orders, rupees of flow. One event has just read two opposite ways depending on that choice, so making the choice in writing is not bureaucracy, it is the step that fixes the answer.

Two. Fill all four cells, meaning both terms in the earlier period and both again in the later one. Where a cell stays empty, treat that emptiness as information rather than as an obstacle to be worked around with an assumption.

Three. The analyst computes both effects at opening levels and puts the residual on its own line, naming the convention used. A residual named is a residual somebody can argue with; a residual quietly folded into one of the two effects is a decision nobody knows was made.

Four. Ask separately whether the composition of the activity moved. Step four is the insertion. A composition change arrives inside the yield term and is invisible from there, so it has to be asked as a separate question.

Five. The analyst says which of the four legs the evidence supports and which it does not, item by item, rather than as one general caveat at the bottom. A caveat covering everything protects nothing.

Six. The analyst stops at the period observed. A growth leg is a description of a change that happened and never a rate to carry forward. Step four is the one that has to be inserted, and the reason is worth noticing: the other five are already standard practice, a careful analyst runs every one of them, and all five together still miss mix entirely.

Where does the split of a revenue change stop, and what carries on elsewhere?