Market Size: TAM, SAM, SOM and How to Estimate Honestly

A market size is a total, and a total is a convention before it is a measurement. Move where the line goes and the same year's buying adds up to a different figure, with nothing that was counted having stirred. Total addressable market (TAM), serviceable addressable market (SAM) and serviceable obtainable market (SOM) are therefore best read as three questions about where three lines went, and only after that as three numbers somebody produced.

A convention sounds like a quibble until it happens in front of a reader. Counting anything at all takes two decisions before a single item is added up: what counts as one of the things, and who counts as being inside the set. Neither decision is discovered by counting. Both are made by a person, usually in about four seconds, usually without being written down anywhere, and the arithmetic afterwards is faultless whichever way they went. So the number at the bottom of a sizing slide is the easy part, and the two sentences nobody wrote down are the whole of what it means.

Consider a wedding for a moment. The household's payment to the caterer, the caterer's receipt after the tent people took their cut, and the whole event's cost to the household once the jewellery and the travel are counted are three different numbers. Three relatives asked what the wedding cost will give all three, and not one of them is lying. The three relatives are answering three different questions that sound like one question. A market total behaves in precisely the same way, and one marketplace's single year shows it happening.

One marketplace, one year, three totals, and all three right. How?

Setu Bazaar, an invented marketplace, is where small sellers list goods and fifty thousand buyers order them. Over one year, three totals describe that trading.

The buyers paid Rs 5,00,00,00,000/-. Setu Bazaar bills the sellers 4.00 per cent of that flow, so what reaches the sellers is Rs 4,80,00,00,000/-, and the marketplace keeps Rs 20,00,00,000/-. The sellers' total and the marketplace's slice add back to the buyers' total exactly, and no rupee is counted twice. Now draw the chain the other way instead, counting each seller's full sale as that seller's own revenue and the marketplace's billing as a stage of its own on top: the sellers' stage becomes Rs 5,00,00,00,000/- rather than Rs 4,80,00,00,000/-, and the chain totals Rs 5,20,00,00,000/-. The commission gets counted once as the marketplace's revenue and once inside the seller's sale. Neither drawing is wrong, and all three totals are published under profit pools.

Put the arithmetic beside itself. Rs 5,20,00,00,000/- less Rs 4,80,00,00,000/- is Rs 40,00,00,000/-. Against the narrowest of the three that gap is 8.33 per cent; against the middle one it is 8.00 per cent. Eight and a third per cent of a total appeared and disappeared without one buyer changing what they bought, one seller changing what they shipped, or one rupee changing hands differently. Handed the three figures on three slides in three meetings, a reader would assume somebody had made a mistake, and nobody had.

TAM vs SAM vs SOM: so what are the three rings actually asking?

The three letter names are the standard working vocabulary for three nested boundaries, and they get taught almost everywhere as three numbers of decreasing size. Read them instead as three questions and they become genuinely useful.

The widest ring asks: if every buyer who could possibly want this thing bought it, from anybody at all, how much buying would that be? The middle ring asks: of that, how much sits inside the part a seller could actually reach, given where it sells, what it sells, and who it is permitted to sell to? The innermost ring asks: of that, how much did it get, or could it plausibly get? Each name records where somebody drew a line, so the three names describe where the lines went rather than how big anything is.

The three rings are three boundary decisions before they are three figures. The reason is short: a figure for either of the outer two rings has to be built out of a count of other people's buying, and for the two businesses below no such count exists. So the innermost ring is filled from invoices, and the outer two stand empty, with the reason written where the number would have gone.

The asymmetry the whole argument turns on is an asymmetry of evidence. A business's own sales are a fact about its own ledger, made of invoices that exist in a drawer and knowable to the rupee. Every ring outside that is somebody's estimate of somebody else's activity, and the further out the ring, the more of it is estimate. Certainty falls off sharply with distance from the ledger, and nothing in the three letter names says so. The three names look like three measurements of the same kind. In fact the three are one measurement and two constructions.

Addressable Market: addressable to whom, exactly?

The word doing the damage in that phrase is addressable. Addressable means reachable by the seller: its geography, its channel, its permissions, its ability to actually serve an order once it arrives. Addressable is a fact about who is doing the addressing. So two honest businesses in the same trade will report different addressable markets, and both will be telling the truth. A stall outside one office building and a chain with warehouses in nine cities are looking at the same buyers and can reach very different fractions of them.

One substitution does more damage than any other. A supply ceiling is not an addressable market. Anjani Stationers Private Limited, invented for teaching, makes hard bound registers. Its rated capacityWhat a works is built to turn out in a year if the machines run at their stated rate through the time actually available. Rated capacity is an engineering statement about the plant. Whether anybody wants the output is a different question. is 4,00,000 registers a year and it made 2,50,000, which is 62.50 per cent of rated capacity. Put the idle 37.50 per cent of rated capacity through at the same realised priceWhat a seller actually received for each unit once discounts and allowances are taken off, rather than the price printed on a list. The realised price is worked backwards from the money that came in. of Rs 108.00/- and the same cost that moves with each register, and the numbers are these: revenue Rs 4,32,00,000/-, contributionThe part of a sale that survives the costs that sale itself caused. Contribution goes towards the bills that arrive whether anything sells or not, and only what remains after those bills is profit. Rs 1,84,80,000/-, operating result Rs 1,10,80,000/-, with the margin climbing from 15.37 per cent to 25.65 per cent. All of that is worked in full under throughput and capacity utilisation.

And it is still not a market. Rs 4,32,00,000/- is what this business could supply at its own price without buying a machine. Nothing anywhere says one further register would be bought. Two assumptions travel with that figure or the figure does not travel at all: the split between the cost that moves and the cost that stands still is a judgement rather than something the accounts disclosed, and the price is held at Rs 108.00/- straight through a year 60.00 per cent larger than the published one.

Where does the line go, and who drew it?

A substituteA different way for a buyer to get the same thing done. A substitute sells nothing that resembles the seller's product, and for exactly that reason goes missing from a list of rivals. is the reason the outer rings get large so fast. The ruling is settled under rivals and substitutes: a rival is defined by what it sells, and a substitute by what the buyer was trying to achieve, and the second circle is always the larger of the two. Drawn on the trade, one set of buying gets counted. Drawn on the purpose, a bigger set of exactly the same buying gets counted. The purpose in this case is a school's: a record of something that lasts the year, can be written in by hand, can be signed by whoever is responsible for it, and can be produced again a term later when somebody asks.

So the boundary is a ladder rather than a switch, and each rung is a defensible place to stop. The ladder is built upward from the invoices, with each cell filled honestly on the way.

Rung nought is made of invoices, so it is exact: Rs 2,70,00,000/- across 36 accounts, of which the Sunrise Public School group accounts for Rs 81,00,000/-, being 30.00 per cent, with the other thirty five averaging Rs 5,40,000/- each. Rung one is where it starts to thin out. Bhavani Register Works, a second maker of the same registers to the same specification, turns out 1,50,000 registers a year. So that rung carries a floor of 4,00,000 registers and no money whatever.

Rung one carries no rupee value for a reason rather than from a shrug. Nothing published anywhere gives the second maker's price, its cost of paper or its works cost. Filling that gap with a trade average would be an invention wearing the clothes of research. And multiplying the 4,00,000 unit floor by Rs 108.00/- prices a rival's whole year of work at a price the rival never charged anybody.

The everyday version sits on any ordinary lane. There is one chemist on the lane, four in the neighbourhood, some number of shops in the city that sell the same tablet, and then every way at all of stopping a headache, a glass of water and a dark room included. All four are legitimate boundaries. The boundary is what gets reported, and the name comes last, if it is written at all.

Top-Down and Bottom-Up Market Sizing: how is a figure got at all?

Suppose the line is fixed and written down. There are exactly two shapes of route to a number. The difference between them is a difference in what evidence has to be in hand before either can start, so both are worth naming precisely.

Top-Down

The route starts from a total somebody else published and takes a share of it. The total might be a national series, a trade body's estimate or a figure inside a purchased study, and the share is a view of how much of it belongs to the part of the world in question. A top-down run needs one total and one share, and it will return a clean looking number for any pair of them supplied.

Bottom-Up Market Sizing

Start from a count and a price and multiply. How many buyers, how often each buys, what each pays. A bottom-up build needs a count and a price, and if either is missing the build stops where it stands rather than producing anything.

Which route can actually be run on a given business, what it means when two routes agree, and why one of them announces its own failure while the other conceals it are covered separately under Top-Down vs Bottom-Up Market Sizing: Which Route Can Be Run.

Move the control to the right hand position and watch the take rate readout rather than the total. The published take rateThe slice a marketplace bills for carrying a trade, written as a percentage of the money that crossed it. Charge more on the same flow and it rises; carry more flow at the same charge and it does not. of 4.00 per cent is 4.00 per cent at one position only. The rupees the marketplace kept are a fact and do not budge. The percentage is that fact divided by a total, and the total was a choice, so even a ratio has a convention hiding in its denominator. Ratios are exactly the figures people quote when they think they are being careful, so a convention hiding in a denominator is worth more than it looks.

How to Estimate Market Size: what does an honest procedure look like?

Six steps, in order, and the sixth is the one that gets skipped.

An estimate with no stated assumptions is not an estimate, it is an assertion. Assumptions written underneath a figure get separated from it the first time somebody copies the number into another document, so step five is where most sizing work quietly fails. Step six fails more quietly still: a person who can say what would falsify their own figure has thought about it properly, and a person who cannot has produced a number rather than a finding.

Now run it on Anjani Stationers Private Limited and watch where it stops. The purpose is published, so step one is easy. The ladder above is exactly that decision written out five ways, so step two is easy. Step three is where it ends. Nobody has published a total for this field, so the first route cannot start. There is no count of anything outside the business's own ledger, so the second route cannot start either. The output is the name of the missing evidence. The name is worth handing to somebody, and a number built on top of the gap is not.

Market Concentration: what shape is the field inside the line?

A total states how much buying happens inside a boundary. The total says nothing about how that buying is arranged, and the arrangement decides whether any of it is available to a given seller. A total with no shape is a number, and the shape of the denominator decides whether the total means anything to a given seller at all.

Two fields of identical size are not the same opportunity. One is ten thousand tiny buyers, each ordering a little, none of them worth a trip. The other is four buyers who between them do nearly all the spending, every one of whom already signed a three year contract with somebody else. Same total. In the first, a slice can be won by being cheap and available; in the second there is nothing to win until one of the four contracts comes up.

Shape is the whole of what concentration adds to a total. How a concentration figure is built, what its scale means and why the shares get squared is covered separately under Concentration Risk: How Exposure Clusters and How It Is Measured. The effect of a badly drawn boundary on such a figure is covered separately under Herfindahl-Hirschman Index: Which Market Are You Measuring? How many sellers a field carries and what happens as one consolidates is covered separately under Consolidation and Fragmentation. To a market total, concentration is the shape of the denominator, and nothing else.

How Market Size Affects Company Strategy: what does the answer actually buy?

Why size matters has no general answer worth having. Two businesses answer it better: one that needs the figure and cannot get it, and one that does not need it at all.

Setu Bazaar bills a buyer Rs 4,000/- across a year, spends Rs 2,000/- serving that buyer, and keeps Rs 2,000/- of contribution. Against that sits a standing costA cost that arrives at the same size whether the business trades hard or stands quiet: the rent, the insurance, the pay of people who are there either way. of Rs 12,50,00,000/- for the year. So its break-even volumeThe number of units, or of buyers, at which what a business takes in exactly covers what it spends. Money goes out below that count and stays above it. is Rs 12,50,00,000/- divided by Rs 2,000/-, which is 62,500 buyers, being Rs 25,00,00,000/- of gross merchandise valueEverything traded across a marketplace in a period, added up at what the buyers paid. Gross merchandise value says how much business crossed the place and nothing at all about how much the place kept. billed onward as revenue at its own rate. Setu Bazaar has 50,000. The shortfall is 12,500 buyers, 20.00 per cent below the line, and closing it needs 25.00 per cent more buyers than the marketplace currently has.

So: are there 62,500 buyers? Sit with that question rather than moving past it. The question is a market size question, asked in the units a market is actually made of, on a business that has to answer it in order to survive. Nothing Setu Bazaar has published answers it in either direction, and nothing it has published would let anybody build the answer honestly. The refusal has a price, and the price is that the one question this business must answer stays open.

The mirror image makes it sharper. Anjani Stationers Private Limited's binding question is not a market size at all. Its break-even level sits at 1,60,173.16 registers, being Rs 1,72,98,701.30/- at Rs 108.00/-. The level is 35.93 per cent below the Rs 2,70,00,000/- the business actually took, and the business sells 89,826.84 registers more than the level requires. Anjani also runs at 62.50 per cent of rated capacity, so more than a third of its machines stand quiet. Both of those come off its own accounts, so one business needs a market size and cannot get one, and the other needs its own ledger and holds it.

What gets printed when three of the five rungs are blank?

The answer to the practical question is not the one most people expect. The card goes out with the blanks in it, and beside each blank stands what would have to exist for it to be filled.

A card with three empty rows and a reason beside each one is a finding, and the same card with three plausible numbers in it is a fabrication that nobody will ever check. The claim is a strong one, and the reason it holds is this. The blank card can be handed to somebody who knows the trade, and they can say at once whether the missing thing exists, who publishes it and what it would cost to buy. The filled card has nothing in it to disagree about, so it cannot be argued with: every row carries a number, every number looks like every other number, and nothing on the face of it separates the row built from invoices from the row somebody guessed on a Tuesday afternoon.

What four lines travel with any market total?

Four lines, in this order, travel beside any market total before it leaves the desk, and a working analyst, a lender or a person writing the slide can use them tomorrow.

A market total with all four lines blank is a number rather than a finding. And notice how cheap the first two are. Written on each of the three opening totals, lines one and two alone would have separated them on sight, without anybody recounting a single transaction, and the meeting that argued about which figure was right would never have happened.