Market sizing when your category doesn't exist yet
Guide · Market Growth · 7 min read · last verified 2026-07-24
Market sizing when your category doesn't exist yet: the short answer
When your category has no settled definition, you cannot look up a TAM — you build one from evidence you can defend. Triangulate three views: the adjacent and displaced budgets your product draws from, a bottom-up model from the job customers hire you to do, and any early analyst estimates — then report a range, never a single false-precision number. A new category is defined by disagreement; estimates for it routinely differ by an order of magnitude, and that spread is data, not noise. This builds on what total addressable market means and SAM and SOM; the hard case below is when no comparable exists.
Why the usual TAM sources fail for a new category
(Figures in this section are sourced inline — category numbers attributed to the named analysts, and Magrios’s own numbers labelled derived or hypothetical worked examples.)
(Every market figure in this section is a cited third-party analyst estimate; sources named inline. Forward-looking numbers are labelled hypothesis.)
For an established category you buy a report and quote its number. For a category that barely has a name, the reports either do not exist or contradict each other so badly that any single number is meaningless. Take GEO/AEO — AI-answer optimization — the live example Magrios sizes for its own market. One house puts GEO services at ~$1.01B in 2025 → $17.0B by 2034 at 45.5% CAGR [IntelMarketResearch]; a second at $848M → $19.8B by 2034 at 50.5% CAGR [MarketIntelo]; a third at $7.3B by 2031 at 34% CAGR [Valuates]. Same category, forecasts diverging by an order of magnitude (every forward number here is hypothesis). The disagreement is the signal: the analysts are guessing because the category is new — which is exactly why you size it yourself, from budgets and jobs, and show your work. See the honest market-sizing playbook for the discipline this demands.
Method 1 — Size the adjacent and displaced budgets (the proxy method)
(Figures in this section are sourced inline — category numbers attributed to the named analysts, and Magrios’s own numbers labelled derived or hypothetical worked examples.)
(Every market figure in this section is a cited third-party analyst estimate; sources named inline; the summation is Magrios's own derivation.)
A new category rarely creates spending from nothing; it redirects budget that already exists. So name the adjacent categories your product draws from and the budget it displaces, then size those from cited figures — the proxy / adjacent-budget method. For Magrios three budgets are in reach:
- GEO/AEO — the new category itself. ~$1.0B in 2025 (AEO alone ~$655M, about 34.5% share of the GEO market, ~42.8% CAGR) [MarketIntelo]. Small today, fastest-growing of the three.
- Competitive & market-intelligence software — the budget being displaced. ~$0.6–5B in 2025 depending on how narrowly you scope it, ~15% CAGR toward ~$15B by 2033 [SkyQuest, Fortune Business Insights, Mordor Intelligence].
- SEO software — a large adjacent pool. ~$41–86B in 2025, ~13.5% CAGR [Grand View Research, Precedence Research], of which only the AI-era measurement slice (~5–10%, derived) is genuinely capturable.
Sum the capturable portions and you get a top-down TAM of roughly $8–12B in 2025 (derived), carried toward ~$30–40B by 2030 (derived) by GEO/AEO's growth rate. The discipline of the proxy method: every input is a cited number for a real budget, and the arithmetic joining them is labelled your own derivation — not smuggled in as an analyst's fact.
Method 2 — Build bottom-up from the job-to-be-done
(Figures in this section are sourced inline — category numbers attributed to the named analysts, and Magrios’s own numbers labelled derived or hypothetical worked examples.)
The proxy method can flatter you, so cross-check it from the ground up: who has the job your product does, how many of them exist, and what will they pay? A bottom-up number is a hypothesis by construction — you are modelling a future, not measuring a present — so label it that way.
Worked example (illustrative — Magrios's own model). Start from the buyers who genuinely have the "know and improve my AI and search visibility" job. US mid-market and enterprise companies (over $10M revenue) number ~220,000 [SalesHive, Martal, Close]. Apply a global multiplier (~3.5×, itself a hypothesis) and add roughly 120,000 funded startups and agencies, and you reach ~890,000 addressable B2B organizations globally. Multiply by a blended annual contract value of ~$22K — arithmetic on Magrios's real prices (Pro $9,000/yr, Pro+ $25,164/yr) — and the bottom-up TAM lands near $19.7B (hypothesis).
The precise figure matters less than where it lands: the same order of magnitude as the top-down $8–20B. Two independent methods agreeing on the order is the strongest signal available when no report can settle it — and how pricing shapes market size explains why the ACV assumption does so much of the work.
Method 3 — Triangulate, then report a range not a point
(Figures in this section are sourced inline — category numbers attributed to the named analysts, and Magrios’s own numbers labelled derived or hypothetical worked examples.)
(Figures below are the derived and hypothesis outputs of the two methods above; category inputs remain the cited estimates named earlier.)
Now stack the two views and read them together:
| Layer | Figure (2025) | Basis |
| --- | --- | --- |
| The category — GEO/AEO | ~$1.0B | cited analysts (measured) |
| TAM — adjacent + displaced budgets | ~$8–20B | top-down + bottom-up (derived) |
| SAM — reachable, marketing-mature B2B | ~$3–5B | ~30–40% share of TAM (derived) |
| SOM — realistic 3-year obtainable | ~$5–20M ARR | bottom-up on real pricing (hypothesis) |
SAM narrows TAM to what you can actually serve — for Magrios the B2B, English-first, marketing-mature slice, roughly 150K–250K organizations (hypothesis) at ~$18–22K ACV, giving ~$3–5B today tracking to ~$12–15B by 2030 (derived). SOM is the honest floor, sized next. The output is a defensible band — TAM ≈ $10–20B, SAM ≈ $3–5B, SOM ≈ $5–20M ARR — never a single decimal-place figure.
A worked example end-to-end: the three SOM scenarios
(ARR figures are arithmetic on Magrios's real, published prices; the customer counts are the hypothesis.)
SOM is where founders lie to themselves, so model it explicitly and show the arithmetic. Magrios's prices are public: Pro $750/user/mo ($9,000/yr), Pro+ $2,097/user/mo ($25,164/yr), Enterprise custom. Blended ACV depends on plan mix (all hypothesis): conservative ~$13K, base ~$22K, aggressive ~$34K. Run three-year scenarios and the obtainable market falls out:
| Scenario | Blended ACV | Year-3 customers | Year-3 ARR |
| --- | --- | --- | --- |
| Conservative | $13K | 400 | ~$5.2M |
| Base | $22K | 900 | ~$19.9M |
| Aggressive | $34K | 2,000 | ~$67.5M |
Read together, that is a realistic 3-year SOM of ~$5–20M ARR (hypothesis), with genuine upside near ~$65M if product-led growth and an upmarket motion both fire. Against a category heading to $10–20B, even the base case is a rounding error — the correct shape for an early category-definer, not a flaw to hide.
How to be honest about the uncertainty
(Figures in this section are sourced inline — category numbers attributed to the named analysts, and Magrios’s own numbers labelled derived or hypothetical worked examples.)
The credibility of a no-comparable market size comes entirely from how you label it. Apply a three-level confidence taxonomy to every number:
- Measured — a cited third-party figure for a real, existing budget (e.g. "AEO ~$655M in 2025 [MarketIntelo]").
- Derived — your own arithmetic joining measured inputs (e.g. the ~$8–12B TAM summation).
- Hypothesis — any forward-looking or bottom-up figure that models a future: customer counts, blended ACV, 2030 projections.
Never write "assumed"; write which of the three it is. Show ranges, not false precision. And treat analyst disagreement as evidence about the category's youth rather than something to average away — the spread is the finding. The same standard runs through Magrios's product: size from public evidence, cite the source beside the claim, and separate measured from modelled. See the confidence taxonomy.
Common mistakes when sizing a category that doesn't exist
- Quoting one analyst's number as fact. When estimates diverge 10–100×, a single figure is a guess wearing a decimal point.
- Top-down only. A big adjacent-budget number with no bottom-up cross-check is a story, not a size.
- Confusing TAM with SOM. "Any business could use this" is your TAM; your obtainable market is a narrow, reachable, marketing-mature wedge.
- Hiding the assumptions. The global multiplier, the blended ACV, the capturable slice — surface each and label it hypothesis, or the model is unfalsifiable.
- Averaging away disagreement. In a young category the spread is the finding; report it, do not smooth it — which is what makes category creation so hard to size, and so valuable to own.