Why sizing goes wrong
The classic failure is easy to recognise. A slide shows a very large figure lifted from an analyst report, followed by the claim that winning a small slice of it would make the business a success. Nobody has checked how the report defined the market, whether the business can reach those buyers, or why the slice would be won. The figure looks rigorous because it has a source and a currency sign; it is really a hope with a citation.
Three things cause most bad sizes. The first is borrowed definitions: a published figure describes a category as the publisher drew it, which rarely matches what the business sells, to whom, or where. The second is a single method: a top-down estimate never tested against a bottom-up count, so an error in the source passes straight into the plan. The third is false precision: a point estimate where the evidence only supports a range, which invites the reader to treat an assumption as a measurement.
There is also a motive problem. Sizes are often produced to justify a decision already taken, for a funding round, a board paper or a budget request, and the person producing them has an interest in a large answer. The market and competitor research hub covers judging other people's sources for interest and method; the same discipline applies to the sizes a team produces for itself. A useful test is to ask what the figure would be if the analyst wanted it to be small, and whether the method would have stopped them.
TAM, SAM, SOM: definitions that hold
The three-layer model is common because it separates three different questions. Its value depends entirely on writing each layer as a definition someone else could apply and get the same answer.
| Layer | Question it answers | A definition that holds | A definition that does not |
|---|---|---|---|
| Total addressable market (TAM) | How much demand exists for the job the product does, if every buyer bought from us? | Every organisation or household with the problem, counted by a stated rule, valued at what they currently spend solving it | "The global software market" |
| Serviceable available market (SAM) | How much of that demand can our product, channels and geography actually reach? | The TAM filtered by the segments, regions, languages, sizes and price points the business serves today or will serve within the plan | The TAM with a round percentage removed |
| Serviceable obtainable market (SOM) | How much of the SAM can we realistically win within a stated period? | A share justified by sales capacity, conversion evidence, competitor strength and switching behaviour | The SAM multiplied by a share chosen because it sounds modest |
A few rules keep the layers honest. Define the market by the job buyers are trying to get done, not by the product category, so that substitutes are counted; a business selling accounting software competes with spreadsheets and bookkeepers as well as other software. State the unit: buyers, purchases, or spend. State the period, usually a year. And state the boundary conditions, such as the minimum company size, the regions and the price band, because those boundaries are where most disagreements about the figure live.
The obtainable layer is the one that matters for planning and the one most often invented. It should be built from evidence about the business's capacity to sell: how many prospects marketing can reach, how many sales conversations the team can hold, how often those conversations convert, and how quickly customers switch away from the incumbents. That last point is where the competitive intelligence guide feeds in, because a share taken from a strong, watchful competitor is harder to win than one taken from a fragmented field. The five forces model widens that check beyond rivals to the power of buyers and suppliers and the threat of new entrants and substitutes, all of which can shrink the share a business can realistically hold at a healthy margin.
Top-down and bottom-up methods
Top-down starts with a broad figure and narrows it. A published estimate of category spend, an official count of businesses in a sector, or a population statistic is filtered step by step: by region, by segment, by the share of buyers who have the problem, by the share who pay to solve it. Its strengths are speed and a link to a recognised source. Its weakness is that every filter is an assumption, and the assumptions compound: a chain of plausible percentages can produce almost any answer.
Bottom-up starts with the unit of purchase and builds up. Count the buyers who meet the serviceable definition, using official registers, trade directories, customer lists or panel data; estimate how many of them buy, how often and at what value, using the business's own sales data, competitor disclosures or primary research; multiply. Its strengths are that it uses the business's own definitions and evidence, and that each input can be inspected and improved. Its weakness is effort, and a tendency to assume that the buyers the business already knows are typical of those it does not. A business-to-business supplier deciding whether one industry deserves its own programme counts that industry this way, as the guide to marketing to an industry vertical describes.
A third route, sometimes called value-theory sizing, starts from the value the product creates for a buyer and asks what they could rationally pay. It is useful for new categories where nobody yet spends money on the problem, but it estimates potential rather than existing demand, and should be labelled that way.
Run at least two of the three independently, ideally by different people who have not seen each other's working. The point is not to average them but to find out why they differ. A top-down estimate well above the bottom-up one usually means the published category is drawn wider than the business's definition, or that the bottom-up count missed a segment. Either finding improves the size and the understanding of the market behind it.
Proxies: search demand, category data, trade sources
Where direct counts are thin, proxies fill the gaps. Each measures something related to demand rather than demand itself, so each needs a stated reason for why it moves with the market.
Search demand. Query volumes for the problem, the category and its substitutes show expressed demand and how it is distributed across regions and over time. The method is the one used for keyword research: build the full set of terms a buyer would use, group them by intent, and read the volumes as a relative signal. Search data is strongest for comparing sub-markets and reading direction, and weakest as an absolute count, because it only sees people who search and counts searches rather than buyers. Estimating the traffic a business could earn from that demand is a further step; the SEO traffic estimator shows how rankings translate into clicks and makes the gap between total searches and attainable visits visible. Teams that invest in SEOImproving visibility in unpaid search results. often treat the category's query volume as their market, which confuses the size of the shop window with the size of the market.
Category data. Official statistics on business populations, household spending and trade flows, and regulator market studies where they exist, give counts and values with documented methods. They often lag the market and use category definitions that need mapping to the business's own. The tutorial on running a desk research pass logs each such source with its definition and dates before any figure is used, so the mapping is visible.
Trade and industry sources. Trade associations, trade press, industry events and the published accounts of listed competitors give spend estimates, customer counts and distribution data. They are informed and frequently self-interested. The disclosures of competitors are especially useful for bottom-up checks: a customer count or a segment revenue figure sets a floor under a market estimate. Where the published figures leave a gap, the guide to interviewing industry experts and the trade shows how distributors, installers and consultants can test a bottom-up assumption without crossing into competitors' confidential information.
Media and audience data. Readership, membership and audience ReachUnique people who saw an ad. figures for the publications and communities a segment uses can bound how many buyers exist, as long as the overlap between sources is handled. A PESTEL scan alongside the proxies helps judge whether the forces acting on the market will grow or shrink the figures over the plan period.
Triangulation and ranges
Triangulation means estimating the same quantity by routes that do not share assumptions, then reading the agreement and the gap. Where routes agree, confidence rises. Where they disagree, the disagreement points at the assumption to investigate before anyone relies on the figure.
The output is a range, not a point. Build it by identifying the few assumptions the estimate is most sensitive to, usually the share of buyers with the problem, the purchase value, and the obtainable share, and running the estimate at a low and a high value for each. Report the low and high ends, the assumption that drives most of the spread, and the evidence for each end. Where the range is too wide to support the decision, the next research effort goes into the assumption that drives the width, which is a far better use of budget than a larger report.
Sizing a market for refurbished office chairs
Pellam Seating refurbishes used office chairs and sells them to small firms in England. The team sizes the market three ways. Top-down: a trade body puts annual UK office seating spend at about £600 million; filtering to England (roughly 85%), firms under 50 staff (about 40% of spend) and the share open to refurbished goods (the team assumes 15% to 25%) gives £31 million to £51 million. Bottom-up: official statistics list around 230,000 firms with 5 to 49 staff in England; Pellam's own sales data shows a buying firm replaces about six chairs a year at an average refurbished price of £140, and a customer survey suggests 10% to 18% would consider refurbished, giving £19 million to £35 million. Search proxy: query volume for refurbished office chairs in England is about a fifth of the volume for new office chairs, consistent with the lower half of both ranges. The team reports a serviceable market of £20 million to £35 million, names the openness-to-refurbished share as the assumption driving the spread, and commissions a short survey to narrow it. Its obtainable target, based on sales capacity, is £1.2 million to £1.8 million within three years.
Presenting a market size honestly
The presentation decides whether a size helps the decision or distorts it. Lead with the definition, not the number: who is counted, where, in what unit and over what period. Give the range and the method for each end. Show the agreement or disagreement between routes, because a reader who sees that two independent methods land close together trusts the figure for the right reason. Name the assumption that drives most of the uncertainty and what it would take to narrow it.
Keep the layers separate on the page. Readers drift from the total to the obtainable figure without noticing, and plans built on a total addressable market as if it were obtainable are among the most common failures in go-to-market planning. The obtainable figure should connect directly to targets: the number of customers it implies, the sales and marketing capacity needed to win them, and the KPIA metric tied directly to an objective. the plan will track. If the obtainable figure implies more sales conversations than the team can hold, it is not obtainable. The market sizing worksheet records each estimate, assumption and source in this order, so someone who did not build the range can audit it. When a size has already been questioned by a board or an investor, the diagnosis for a market size estimate that has been challenged rebuilds it layer by layer from these definitions.
Date the estimate and say when it should be refreshed. Markets move, sources are revised, and an estimate built for one decision is often reused for another without anyone checking that the definitions still fit. For businesses comparing several countries, the same discipline applies market by market, and the market entry and prioritisation guide covers how sizes combine with access, cost and risk to rank them. Finally, connect size to position: Share of VoiceYour share of market visibility. in the category, tracked over time, is a leading indicator of whether the business is gaining on the obtainable share it planned for. The share the business already holds is a separate estimate measured against these same definitions, and the guide to estimating your market share works it out by value or by volume from the business's own sales and from its competitors.
A segment close to its natural size is one of the ceilings diagnosed in marketing growth that has plateaued, together with saturated channels, churn and the absence of new offers.
A market size says how many buyers could exist, not whether they want a particular product; the tutorial on testing demand before a launch asks that second question with an honest page and a commitment such as a refundable deposit.
Common mistakes
- Quoting the total as the opportunity. The total addressable market presented as if the business could win it, with the serviceable and obtainable layers skipped or blurred.
- Borrowing a category definition. A published figure used without checking that the publisher drew the market the same way the business does.
- One method only. A top-down chain of percentages never tested against a bottom-up count.
- A share chosen for how it sounds. An obtainable share picked because it seems modest, not derived from sales capacity and conversion evidence.
- Search volume as market size. Treating queries as buyers and ignoring everyone who buys without searching.
- A point estimate with no range. False precision that hides the assumption doing the work.
- No date and no refresh. An old estimate reused for a new decision without rechecking definitions or sources.
