How to Calculate Market Share in Competitive Analysis

Market share is the share of a category’s spending that ends up with one company over one stated period. In competitive work it is almost never requested for its own sake. It is requested because somebody wants to know whether a named competitor is pulling ahead, and share is the most respectable-sounding number available to answer that.

A number you measure divided by a number you assume

The two halves of this fraction are not the same kind of object. Your own revenue is measured: it exists in a system, somebody reconciles it, and an auditor has looked at it. The market total is an estimate produced by someone with their own definition of where the category starts and stops, or it is your own estimate, built from counts and assumptions you chose.

That asymmetry is why two competent analysts can compute different shares for the same company in the same year without either making an arithmetic error, and it is the reason almost every argument about a share figure turns out to be an argument about the denominator. Most of what follows is about what the denominator does to the answer.

What the request is usually standing in for

When a founder or a board asks for market share, the underlying question is normally competitive: are we gaining on them, or are they gaining on us. Share can answer that, and it answers it slowly and at low resolution, for reasons that are structural rather than fixable. Two cheaper measurements answer the same question faster: your share of what the category actually decided this year, and your win rate against that competitor in the deals you were both in.

None of this makes the percentage worthless. It is the number that says how much of a category you hold, it is the one a board and an investor both already understand without explanation, and no conversation about concentration or about which segment to defend can start without it. It is simply not a performance report, and most of the trouble this metric causes comes from being read as one.

Same period and same basis on both sides

The two numbers have to describe the same window and the same thing. A trailing-twelve-month revenue figure over a calendar-year market total is a mismatch that nobody notices and that can move the answer by more than the competitive change you were trying to detect. So is subscription revenue over a total that includes implementation and services, and so is bookings over a total built from recognised revenue.

Revenue, units, seats and customers all produce legitimate share figures and they produce different ones. Counting subscribers rewards whoever sells cheaply to many buyers; counting money rewards whoever sells dearly to few. The same two companies can therefore change places depending on which is used, with neither figure being wrong. Choose the basis that matches the decision, then keep it for the life of the series.

The definition has to travel with the number

The usual failure is not that nobody defined the market. It is that somebody did, carefully, in an analysis that ran to several pages, and then the percentage went into a slide on its own and was quoted for two years without the sentence that made it meaningful.

Treat the definition as part of the figure rather than as documentation of it. Written out in full, the unit is not “8.5%” but “8.5% of practice management software sold to independent clinics in 2025, on a total published by a named source”. It is longer, it survives being repeated, and it is the version anybody outside the room can actually check.

Both numbers in this fraction move every year, which makes the percentage a statement about two growth rates rather than about your own performance. The relation is exact: your share next period is your share this period, multiplied by one plus your growth, divided by one plus the market’s growth. Anything you grow more slowly than costs you share.

Revenue grows by sixty-four per cent across those four years. Every absolute number on the company’s own dashboard improves, every quarter is a record, and the share figure falls by nearly a point and a half. Nobody has done anything wrong and nothing competitive has necessarily changed. The category simply grew faster, which usually means new buyers arrived and somebody else served them.

The same arithmetic running the other way

In a contracting category the effect reverses and is more dangerous, because it flatters. A company holding revenue exactly flat while the market falls ten per cent a year gains roughly a point of share every year and has achieved nothing at all. Share rising during a downturn is the single most over-interpreted reading of this metric, and the test for it is one subtraction: compare your growth rate with the category’s before deciding the movement means anything.

Two extra numbers make the series readable

Publish your growth rate and the market’s alongside the share, in the same row. With those two present, a reader can see immediately which of four situations they are looking at, and the share stops being a figure people argue about from memory.

Dividing by an estimate produces an estimate. A share is your revenue over the market total, so a proportional error in that total passes almost one for one into the answer: a denominator good to plus or minus twenty per cent gives you a share good to roughly plus or minus twenty per cent, and no amount of care in the numerator improves that.

The range is not symmetric either, and the direction is worth knowing. Working the example above, a total that could be as low as $134m puts the share at 10.6%, and a total as high as $202m puts it at 7.0%. The distance above the point estimate is larger than the distance below it, because dividing by a smaller number moves the answer further. A share quoted with a plus-or-minus attached is therefore already slightly understating its own upside.

Rounding is a claim about your evidence

One decimal place says you can tell 8.5% from 8.6%. On an estimated denominator you cannot, and the decimal is doing rhetorical work rather than reporting anything. Round to whole points when the total is an estimate, and to the nearest five points when the estimate is weak. The loss of apparent precision is the whole benefit: a figure stated as “roughly a tenth of the market” cannot be quoted back at you as though it were measured.

The numerator is less solid than it looks

Your revenue is measured, but your revenue inside the defined market often is not. A company selling three products into two segments has to decide which revenue belongs to the category it is claiming share of, and that allocation is a judgement made by the person building the analysis. Include an adjacent product line and the share rises without a single additional sale, which is the numerator version of drawing a flattering boundary.

Both halves of the problem are worth recording rather than resolving. Finding a competitor’s market share covers how to grade a published total as measured, estimated or claimed, which is the question to settle before any of this arithmetic is worth doing.

Where the grading comes back saying nobody has measured the category at all, which is the normal outcome in business software, the comparison that still works is with one company rather than with a market. Relative market share divides your revenue by the largest competitor’s, so no total appears in the arithmetic and none has to be defended. The trade is that everything then rests on a single revenue estimate for a single company.

For any business built on renewing contracts, this year’s category revenue is mostly the result of decisions taken in earlier years. A share of that revenue therefore reports on the installed base: an average across several years of purchasing, weighted towards whichever choices proved durable. It is a sound description of how the category arrived where it is. It is a poor description of what is happening now, and the reason is arithmetic rather than reporting.

Only a portion of the category is genuinely in play in a given year. New buyers entering the market, contracts that actually went out to comparison, and accounts that switched. A renewal nobody contested is not in play, however large it is. Call that portion the flow, and everything else the base, and the movement of the headline percentage is fixed by the size of the flow.

This vendor takes thirty per cent of every decision the category makes, every year, from a starting position of eight and a half. That is a dominant competitive performance, and it still needs four uninterrupted years to move the headline figure to twenty. Run the same arithmetic on a category where only five per cent of spend comes into play each year and the identical performance takes about fifteen.

Why the base share cannot be an operating metric

A number that moves this slowly cannot be used to judge a quarter, a campaign or a repositioning, because the honest expected movement over any of those horizons is smaller than the error bar on the denominator. Teams that set share targets for the year are usually setting a target that their own market cannot reach even if everything goes right, and the same damping works in reverse: a competitive position that has genuinely deteriorated will hold the percentage up for years while the flow underneath it turns.

A win rate is not a share of new business

The natural move at this point is to reach for the win rate, and it does not measure what is needed here. A win rate counts the deals you were in. Its denominator excludes every purchase decided by a buyer who never put you on the list, and in most categories that group is larger than the contested one. Competitive win rate tells you how you do once you are invited; a share of the flow has to account for the invitations that never arrived.

The practical version is an estimate with its working shown. Take the deals you saw, add a reasoned allowance for the ones you did not, and sense-check the total against what your competitors publicly announced they won. Finding a competitor’s customers sets out where those announcements appear. The result is rough, it moves within a quarter, and it reports on the present in a way the headline share cannot.

The row, and every column it needs

One row per period in a single sheet, with eleven columns: the period, the market definition written out, the basis, our revenue, how that revenue was allocated to the market, the market total, who published it, when, the accuracy they stated, the resulting share, and the range. Add a twelfth for what changed since the last row. Anything missing from that list is a question somebody will ask in the meeting where the figure is presented.

The revenue column comes out of the finance system as an export by product and period, and the allocation rule belongs beside it in writing rather than in the head of whoever ran the filter. The total comes from a document, so file the document: a share figure whose source has been lost is unusable the following year, because nobody can reproduce it or update it on the same basis.

Where the artefact and the sourcing are worked through

Two adjacent pieces of work make this one easier. The market share analysis template carries the tables for boundary decisions, two-method sizing, share by segment and confidence per figure, which is where the findings usually are. The TAM, SAM and SOM template covers the sizing itself when no published total exists and you have to build one.

Connecting it to a monitoring platform

The percentage is recomputed once a year in most categories, so there is nothing to automate in the arithmetic. What earns a connection is the column next to it. Where a competitive intelligence platform feeds dated competitor events into the same workbook, each period’s share movement sits beside a list of what actually happened in that period, and the explanation stops being reconstructed from memory six months later.

The hardest input to this metric is neither number in the division. It is the list of companies you decided belonged inside the market, and that list is maintained by your competitors rather than by you. One adds a module and is now selling into your category. An adjacent suite bundles your function as a checkbox and starts taking buyers who never reach your pipeline at all. A vendor you counted last year quietly retires the product line.

A share series computed on a definition written eighteen months ago is comparing periods whose boundaries were different, and the difference shows up as movement that gets attributed to performance. It is a quiet failure: nothing in the spreadsheet indicates that the membership of the category changed, and the person reading the trend has no way to know.

Keeping that membership list current is ordinary competitive monitoring rather than anything exotic. Flares follows what each company ships, prices and claims in public and records the date, so an entry into your category or a withdrawal from it is visible as it happens rather than inferred from a share figure that has already absorbed it.

What no amount of monitoring produces is the denominator. Change detection can tell you that the set of companies competing for a buyer has changed and it cannot tell you how much that set of companies sells, because most of them do not publish it. The total remains an estimate you build and label as one, and the value of watching the boundary is that you are at least estimating the right thing.