How to size a market in a day (not a week)

Every engagement needs a market size, and every junior analyst dreads the week it normally takes to build one. Here's the bottom-up method that gets you to a defensible number in a day, and the honest list of where the hours actually go.

Why market sizing takes a week when it shouldn't

The task looks simple on a slide: one number, maybe with a range underneath. The week disappears because of everything the number is built from: segment definitions, customer counts, price points, growth rates, and the sheer amount of tabbing between sources that each one requires. Most of that week isn't analysis. It's retrieval.

The good news is that the retrieval is exactly the part a tool can take off your plate. The judgement is the part that stays with you.

The bottom-up method, compressed

1. Define the unit. Is the market "spend on X per year" or "companies that do Y"? Pick one unit and stick to it. Most sizing arguments (and most client disagreements) trace back to two people measuring different things.
2. Top-down anchor, bottom-up check. Find one credible analyst number (public source), then rebuild it from segments to sanity-check it. If your bottom-up build lands within striking distance of the anchor, you have something defensible. If it's off by 10x, one of the two is wrong, and finding which is the whole engagement.
3. Segment by 2–3 variables. Geography × industry × customer size is usually enough. More dimensions = diminishing returns. Every added segmentation multiplies the number of cells you need to fill, and a model full of guesses isn't better than a simple model with three solid assumptions.
4. Build a driver model. Price per unit × volume per customer × customer count. Every input needs a source or an explicit assumption. This is the discipline that keeps the number honest: you can't hide a guess inside a formula if every input is labelled.
5. Show the range. A point estimate is a lie; a band from pessimistic to optimistic is a consultable answer. Clients don't trust a single number anyway. They trust the thinking behind a range, because it tells them what would have to be true for each end of it to happen.

Where the time goes (and what removes it)

The research (finding segment counts, pricing, growth rates) is 80% of the effort. That part is exactly what a market-sizing tool can scaffold for you, so you spend your day on judgement, not tabbing between search results.

The breakdown of a typical "sizing week" looks roughly like:

• Two days finding and reconciling segment and pricing data
• One day building and sanity-checking the driver model
• Half a day writing the assumption notes
• The rest going back to fix the numbers your manager (correctly) questioned

The retrieval days are the compressible ones. The assumption-writing day is not, and honestly, it's the day that produces the most value. A client can re-run your math, but they can't re-find your thinking.

Try it here: free market sizing tool. It builds the structure and lets you focus on the assumptions that actually matter.

A worked example (so the method isn't abstract)

Say you're sizing the market for SME bookkeeping software in the UK. The unit question first: are we measuring software revenue, or all bookkeeping spend? Pick software revenue, because that's what a client sells.

Segment by size and complexity: micro (1–9 employees, self-service tool), small (10–49, tool plus light support), mid (50–249, tool plus managed service). Three segments, not twelve.

Driver model per segment: number of firms × % that outsource bookkeeping × % that use software instead of a spreadsheet × price per year. Now every cell is a number with a story. The analyst count comes from a public registry. The outsource rate comes from a survey you can name. The spreadsheet-to-software ratio is your assumption, and you've written it down, so when the number gets challenged you defend the assumption, not the arithmetic.

Run the range: pessimistic uses a low software-adoption rate, optimistic uses the survey's upper bound. The answer lands as "£X–Y million, with the spread driven by adoption, which the client can influence." That's a slide a board can act on.

The whole exercise took a morning once the sources were gathered: not because the model did the thinking, but because the structure meant you never redid work or hunted for a number twice.

The sources that make or break it

The quality of the answer is the quality of the inputs. Three source types carry most of the load:

• Government and registry data for firm counts: stable, citable, and free.
• Industry surveys for adoption and price points: usually the weakest link, so triangulate two surveys before trusting one.
• Competitor pricing for the price cell: scraped or collected directly, and usually the most defensible number in the model because it's visible to everyone.

If a cell has no credible source and no stated assumption, the cell should not exist. Empty is more honest than invented.

The one habit that protects you

Write the assumption next to every number. "I assumed 40% of SMEs outsource this" is worth more to a client than a third-decimal precision that came from nowhere. It's also the thing that survives the review: when someone asks "where did this come from," the answer is already on the page.

Three follow-ups worth internalising:

• Source everything you reuse. If a number came from a previous deck, the source should come with it. Reused numbers that lost their provenance are how errors get laundered across engagements.
• Label estimates vs. facts. A fact has a source you can open. An estimate has an assumption you can state. Both are fine: the crime is mixing them so the client can't tell which is which.
• Know your sensitivity. If the price assumption changes 20%, does the number move 5% or 40%? Know which inputs move the answer most, because those are the ones worth your verification effort.

When to refuse the number

There's a moment every analyst hits where the honest answer is "we can't size this reliably." The professional move isn't to pad the estimate: it's to say so, and offer what you can do instead: a directional range, a benchmark comparison, or a sizing based on the one segment that matters.

A client would rather hear "this is uncertain, here's the range and here's what would move it" than be handed a precise-looking number that falls apart in their board meeting.

Want to see the whole deliverable assembled? Watch the live demo build a strategy project end-to-end: including the market chapter.