Market Share: Where the Figure Actually Comes From

A market share is two numbers with two different histories. The top one is the seller's own sales, counted in the seller's own books and checked by an auditor. The bottom one is a total for the whole field. Nobody counted it, and it usually reaches a reader through a slide deck the company wrote itself. One number was measured. The other was chosen.

What is worth tracing here, and what is already settled?

Two questions about market share look similar and are not. How a share differs from a concentration figure, and how many numbers each one needs before anybody can write it down at all, is covered under Market Concentration vs Market Share: How Each Measure Fails. And the arithmetic itself, dividing one number by another number, was never in doubt for a moment. Nobody has ever been misled by a division.

So set both of those aside and consider what is left. A percentage sits in front of the reader. It exists. Somebody produced it, on a particular day, out of particular materials, and printed it where a reader would find it. A market share is worth tracing backwards rather than computing forwards. The only useful question about a market share is who made it and how, and that question turns out to have a much more interesting answer than the formula does.

Here is why it repays the walk. Ask most readers where a market share comes from and they will say, without much hesitation, that it comes from the market. It does not. A market share comes from a document, the document was written by somebody, and the somebody had reasons. Once the trail has been followed even once, one of these percentages never looks the same again, and it stops being surprising that two of them for the same business in the same year disagree.

What are the two numbers, and where does each one come from?

Write the thing out longhand: a market share is one seller's sales divided by the whole field's sales. Two numbers, one line, one percentage out of the end. Now walk each of them backwards separately. That is where the two halves stop resembling each other.

Start with the top. The numeratorThe number on top of a division. In a share it is the part being measured, and here it is one seller's own sales. is the seller's own sales figure, and it has a paper trail that can be walked in three steps. Step one, it came off the sales ledger. The business keeps that ledger because it has to run itself. Step two, the ledger was rolled into accounts prepared to a published standard, so the figure means the same thing it means at any other business using that standard. Step three, a statutory auditorThe outside accountant appointed to examine a company's accounts and give a formal opinion on them, who carries professional and legal responsibility for that opinion. examined the accounts and signed an opinion on them, and can be sued for that signature. Three steps, and the trail arrives somewhere solid: a count of sales that actually happened.

Now walk the bottom, and watch the walk stop. The denominatorThe number underneath a division, the thing the top number is being measured against. Change it and the answer changes even when the top number does not. is a total for every seller in the field. Step one, it came from adding up every seller. Step two, no seller anywhere is required to tell anybody what it sold. Nobody is appointed to do the adding, so there is no step three. There is no ledger, no standard, no signature and no office. The walk has run out of documents before it has run out of questions.

One half of this figure has a paper trail and the other half has a source. A trail is a sequence of records that can be followed; a source is a person who said so. A trail and a source are not the same kind of thing, and a printed percentage carries no mark anywhere on it to say which half came from which.

The asymmetry is familiar from ordinary life. A person can say to the rupee what they earned last year. A document says so, somebody deducted tax on the strength of it, and the paper could be produced if asked. The earnings of the same person's street are another matter. There is an impression. The impression came from what people mentioned, what was inferred from a new scooter and a new coat of paint, and one figure a neighbour once quoted. Nobody is lying and nobody is careless. The street's total was never written down in the form that person's own income was, and no amount of care with one's own receipts will conjure it.

What does that asymmetry do to everything printed downstream?

Four things rest on that split, and each of them is worth holding on its own.

A ratio inherits the weaker of its two terms, and nothing about the printed result shows which term that was. Divide an exact figure by a guess and what falls out is a guess, printed to two decimal places, wearing the authority of the exact figure that went into it. Inheriting the weaker term is not a subtle statistical point, and it needs no statistics at all. The inheritance is simply the reason a market share can look like the most solid number in a document while being the only number in that document nobody could check.

How much a figure can be trusted is a property of how it was gathered, never of where it is printed. A number that travelled to a stock exchange did not become true on the journey. The question is always which count or which person produced it, and the printing carries no answer to that question unless somebody deliberately wrote the answer down beside it. Printing is a container. Gathering is the thing.

Selection is not dishonesty. A company's own deck is its own account of itself, chosen slide by slide, and everybody in the room knows that and nobody is deceived by the format. The trouble begins later. A figure travels out of the document that framed it and into a document that does not, and the framing was carrying most of the caution.

And the asymmetry itself is the heart of the matter. A business's own sales figure is the most heavily checked number it produces: counted internally, reported externally, signed by somebody who can be sued. The field total is nobody's obligation, nobody's asset and nobody's liability. No person on earth is required to produce it correctly, and no person on earth is punished for producing it wrongly.

Which document does the figure actually live in?

Most readers have never stopped to ask, so name the document plainly. A market share estimate appears in the company's own slide deck: the pack of charts and headline numbers a company publishes to tell investors its own story. Not in the accounts. Not in the notes to the accounts. In the deck.

Investor Presentation: What It Is For and How It Differs From the Filing establishes four things about that document. The filingThe paperwork a listed business is obliged to send in under the rules. Those rules set the contents, so nobody inside the business gets to pick them. contains what the rules require while the deck contains what the company chooses. The deck is unauditedNot examined by an outside auditor, and carrying no auditor's opinion. The word says nothing about whether the contents are right or wrong, only that nobody outside checked., and it usually says so in small print at the foot. Selection is the format rather than a fault. And, in that same account's list of what a deck can carry that a filing cannot, sitting beside strategy and a photograph of a kiln, are market share estimates.

Put those four together and the central sentence follows, and it is worth reading slowly. The one place a market share figure normally appears is a document nobody checked, chosen slide by slide by the party the figure flatters.

Now stay calm. The next paragraph depends on it. The sentence describes a format rather than an accusation against anybody. A deck is supposed to be the company's own account of itself, in the same way a job application is supposed to be the applicant's own account of themselves, and nobody thinks an application is a fraud because it leads with the good years. The document is behaving exactly as designed. The fault, when something goes wrong, is that a figure has been lifted out of a document everybody read as a company's own account and dropped into a note, a valuation or a lending file that reads it as a fact about the world.

How does a field total get written down at all?

Somebody had to produce that bottom number before it could be divided into anything. There are four ways it usually happens, and the useful thing about walking all four is what they turn out to have in common at the end.

The first route is the company's own build. The company adds up what it believes its rivals sell, from prices it has seen quoted, volumes it has inferred from delivery vehicles and hiring, and what a distributor mentioned over tea. Every input is plausible and the arithmetic is careful. The difficulty is structural rather than moral: this is a construction assembled by the party the answer flatters, and a reconciliation which was constructed cannot confirm anything, as covered separately. When a company builds the total itself and then divides its own sales into it, the figure that comes back is the figure it built.

The second route is a trade body's figure. A trade bodyAn association that sellers in one line of business join voluntarily, which speaks for its members and sometimes gathers figures from them. collects what its own members choose to report and adds it up. The result is real, carefully assembled and honest about what it counted. The result is also a total for the members rather than a total for the field, and the gap between the two is exactly the sellers who did not join. The gap has a shape worth knowing. The sellers who stay out are rarely a random selection of the field: the very small, the very new and the ones who prefer not to be looked at.

The third route is a bought research report. It is cited by name, quoted with confidence, and read in full by almost nobody. Its method sits behind a paywall, or behind a single paragraph, or behind nothing at all. And when somebody does open the method, its own inputs are frequently the first two routes wearing a different cover: a survey of the willing, plus estimates supplied by the companies being estimated.

The fourth route is an earlier estimate repeated, and it is the commonest of the four. A figure appears somewhere, gets quoted, and the quotation gets quoted. Repetition is also the only one of the four routes that leaves no trace. The second printing looks exactly like the first. Nothing in print says this is a copy. Repetition slowly turns a number into furniture: nobody produced it recently, nobody defends it, and everybody uses it.

Every route ends at somebody's judgement, and not one of them ends at a count. The finding is worth sitting with, and it is not a complaint about any of the four. Each one is a reasonable thing to do when the alternative is nothing. At the end of all four walks there is a person rather than a register.

Why does the check that works on the rest of the slide not work on this one?

The next few paragraphs are the ones to slow down for. They change how a slide gets read.

A careful reader already has a discipline for slides, and it is a good one. Investor Presentation: What It Is For and How It Differs From the Filing teaches it in full: the filing comes first, the reported numbers get written down, and the deck is then read against them. Each adjusted figure is carried back through its own footnote to the reported one it started as, and a bridge that refuses to close becomes the first thing raised on the call.

The discipline works, and the reason is easy to miss. A reported counterpart exists. Somebody was obliged to produce it, in a prescribed form, in a document anybody can open. The reconciliationA worked bridge between two versions of the same figure, setting out every item that separates them, so somebody outside can travel from either version to the other. is a bridge, and a bridge needs a far bank.

Now try the same walk on the market share sitting three centimetres away on the same slide. There is no reported market share. No filing carries one. No accounting standard requires one. No auditor gives an opinion on one. Nothing exists to reconcile it to, so there is no footnote and no far bank to build a bridge to.

The one figure on the slide that cannot be checked is the one that looks least like an estimate. The adjusted margin announces itself as adjusted: the word is right there, and it demands work. The market share announces nothing. The share looks like a measurement, sits in the same type size, and asks nothing of anybody at all.

And now the consequence, the sharpest turn in the whole argument. A reader who runs the discipline properly, and closes every walk that can be closed, finishes the exercise more confident in the figure they could not test than in the figure they tested. The tested one visibly needed work. The untested one visibly did not. The walk on a market share does not fail. The walk never runs, and a check that never runs leaves no record of not having run. Not running feels exactly like a check that passed.

What does an unsettled bottom number do to a settled top one?

Time to watch it happen, with nothing real moving anywhere.

Hold the top number absolutely still. One business, one year, one audited sales figure, nothing about it in question and nothing about it changing. Now suppose two people estimate the same field total for that same unchanged year, drawing the same boundary, and they disagree about the total.

If the second estimate comes out 25.00 per cent lower than the first, the printed figure becomes 1 over 0.75, or 1.3333 of the first reading. The rise is 33.33 per cent. If instead the second estimate comes out 25.00 per cent higher, the printed figure becomes 1 over 1.25, or 0.80 of the first reading, a fall of 20.00 per cent. Nothing about the business changed between those two readings and the printed figure moved by a third.

Take the second finding too. It is free, and most readers walk past it. The two disagreements are the same size. The two moves are not: 33.33 per cent against 20.00 per cent. The direction that flatters moves the figure further, purely because the disputed number is sitting underneath rather than on top. Shrinking a bottom number has more leverage than growing it. Of two equally wrong estimates, the wrong one in the pleasant direction does more damage.

Label this properly, in the same breath as the arithmetic. Both readings demonstrate a property of a ratio rather than describing a field. No trade, no country and no year stands behind them. Both figures are ratios of one printed figure to another printed figure, so neither needs a market total in order to exist. The two readings belong to one unchanged year. A share that genuinely moved between two years is covered under How to Interpret Market Share Changes: Reading a Ratio.

Which shares in these notes can actually be checked?

An account that only takes figures apart teaches distrust, and distrust is not judgement. Plenty of shares are completely solid. Three such shares appear above, and every one can be recomputed on the spot with nothing but the figure in front of the reader.

The Sunrise Public School group takes Rs 81,00,000/- of the Rs 2,70,00,000/- of revenue at Anjani Stationers Private Limited, an invented maker of hard bound registers, and Rs 81,00,000/- out of Rs 2,70,00,000/- is 30.00 per cent. Each of the other thirty five accounts takes Rs 5,40,000/- of that same total, or 2.00 per cent, and the thirty six of them add back to Rs 2,70,00,000/- exactly. And Anjani Stationers made 2,50,000 registers against a rated capacityThe output a works is rated to be able to produce in a period, as recorded by the business itself. A rating is a stated ability rather than a measurement of what was actually made. of 4,00,000, which is 62.50 per cent.

Now ask what those three have that a market share does not, and the answer is one line. Somebody keeps a ledger for the bottom number. A customer book. A revenue line. A capacity rating. In every case the document can be named, the person who maintains it can be named, and it can be asked for.

A share is checkable exactly when its denominator is a ledger somebody keeps, and a market is nobody's ledger. That is the whole rule, and it travels well beyond these notes. How the rule bears once the field has filled up, and what a share of one channel actually establishes, is covered under Market Fragmentation: Share in a Crowded Market.

So what is Anjani Stationers' market share?

Set out everything that is actually known about this business, and watch the wall arrive.

Its own sales are known exactly. Rs 2,70,00,000/- across 36 accounts, being 2,50,000 registers at a realised Rs 108.00/- each, and 2,50,000 multiplied by Rs 108.00/- returns Rs 2,70,00,000/- on the nose. The year before that the same business took Rs 2,40,00,000/-, and what matters here is not the size of the rise between the two but that both figures came off the accounts rather than out of anybody's head. A second maker of hard bound registers, Bhavani Register Works, is known to exist by name. Nine mills are known to be within reach of the one paper mill Anjani Stationers buys from.

Now put that together and see what it is. Two names and a count of mills. No total of anything at all outside Anjani Stationers' own books is known anywhere. Not for the lane, not for the city, not for the trade. No market share for Anjani Stationers can honestly be printed.

Printing nothing is a result rather than a gap, and the difference matters. The bottom number is the hard one, and nobody is appointed to produce it. Producing one anyway, out of two names and a count, would be the precise fault traced above. A count of sellers is not a distribution of sales among them, and that distinction is covered separately.

The same thing holds at household scale. A person can say to the rupee what they spent on groceries last month. The receipts and the bank messages are there. Their fraction of the street's grocery spending cannot be said. The street exists. Roughly how many households are on it is known. A few of the shops they use could even be named. Nobody ever wrote the street's total down, so no amount of care with one's own receipts yields it.

How is a printed share traced backwards on a working morning?

Where do these figures come from?