The standard tools for sizing a market — analyst reports, industry surveys, comparable company revenues — arrive years after the window of earliest opportunity has already opened.
That lag is not a flaw in the research process; it is a structural feature of how markets work. Clean data accumulates after economic activity does. By the time third-party research firms have enough evidence to publish a defensible number, the earliest and often most favorable positions have already been taken. Understanding how to size a market before that data exists is therefore not a niche skill — it is a prerequisite for acting at the frontier.
Why early sizing is hard
Early-stage markets resist conventional sizing for several compounding reasons. The customer segment may not yet recognize itself as a segment. The product category may not have a name. Willingness-to-pay is speculative because buyers have not yet experienced the alternative. And the enabling conditions — infrastructure, regulation, complementary technologies — may not be fully in place.
The error most analysts make at this stage is reaching for precision the evidence cannot support. A point estimate — a single figure stated with decimal-point confidence — implies more certainty than any early-market analysis can honestly claim. It tends to anchor decision-making around a number that was always a guess dressed as a fact.
The more durable instinct is to ask a different question: not "how big is this market?" but "what would have to be true for this market to reach meaningful scale — and what signals would tell me those conditions are forming?"
Signals that proxy for future demand
Before revenue data exists, several observable phenomena serve as forward-looking proxies. None is definitive on its own; together they begin to sketch the contours of a future market.
Research density and acceleration tell you where scientific attention is concentrating. A sustained increase in publication volume in a technical area — especially when it crosses from basic science into applied research — often precedes commercial activity by a predictable interval.
Patent activity reveals organizational intent. When multiple organizations begin filing in overlapping technical areas, it signals that they expect those areas to have commercial value worth protecting. The geography and assignee diversity of filings can indicate whether one player is moving to dominate or whether an ecosystem is forming.
Capital concentration provides a conviction signal from investors who have done their own diligence. Early-stage funding rounds in a nascent category do not confirm a market exists, but they do confirm that well-resourced parties believe one will. The pace of follow-on rounds and the profile of investors entering at later stages add resolution to that picture.
Regulatory attention is often underweighted as a market signal. Governments and standard-setting bodies tend to move toward technologies they expect to matter. A rulemaking process, a parliamentary inquiry, or a new standards committee is evidence that institutions are taking a technology seriously — and that the addressable market will eventually operate within a defined regulatory envelope, which is what large buyers typically require before committing.
Early adopter behavior is the most direct signal of all. Even a small number of initial deployments — a pilot, a procurement, a first commercial contract — provides actual willingness-to-pay data and begins to define who the real buyer is, as distinct from who analysts assumed the buyer would be.
Bottom-up vs top-down reasoning
Two complementary approaches exist for constructing a size estimate, and the discipline lies in using them together rather than treating either as sufficient.
Bottom-up reasoning starts with the unit of demand. Who specifically will buy this, and why? How many of those buyers exist in the relevant geography or industry segment? At roughly what price point, given what they currently spend on the problem it solves? How frequently will they buy or renew? Multiplying defensible answers to these questions produces an estimate grounded in observable facts about the world rather than extrapolated from aggregate industry projections.
The bottom-up method forces the analyst to make assumptions explicit and therefore testable. Each assumption — about buyer count, price sensitivity, adoption pace — can be revised as new evidence arrives. That is a feature, not a limitation.
Top-down reasoning works in the opposite direction. It begins with the size of the adjacent or legacy market that the new technology is displacing or expanding, then applies a penetration assumption. If a technology captures a defined share of an existing spend category, the implied revenue is a rough ceiling. Used alone, this approach tends to produce numbers that are too large and too confident — it is easy to assume high penetration without testing whether the enabling conditions support it. Used as a sanity check on a bottom-up estimate, it is genuinely useful: if the two approaches produce figures that are orders of magnitude apart, something in one of them is wrong.
Bounding uncertainty with ranges
The right output of early market sizing is not a number — it is a range, with each boundary tied to a named set of assumptions. A low scenario should reflect slower-than-expected adoption, a key enabling condition that takes longer to materialize, or regulatory friction that delays commercial deployment. A high scenario should reflect a faster-than-expected adoption curve, a platform dynamic that expands the addressable pool of buyers, or a favorable regulatory ruling that opens new segments.
Geoffrey Moore's work on technology adoption — and the concept of "crossing the chasm" from early adopters into mainstream markets — is relevant here. The distribution of adoption across Rogers' diffusion curve implies that market size grows non-linearly. Early estimates tend to undercount the eventual scale of markets that cross the chasm and overcount the scale of those that do not. Expressing the estimate as a range with an explicit adoption-pace assumption is the honest way to hold both possibilities.
Ranges also protect the analyst from the most common failure mode in this work: anchoring. A point estimate becomes a target. People argue about whether the real number is slightly above or below it, rather than asking whether the underlying assumption set is correct. A stated range, by contrast, keeps the conversation on the assumptions — which is where the real analytical work lives.
Updating the estimate as signals accumulate
An early market-size estimate is a hypothesis, not a finding. Its value is that it makes assumptions explicit enough to be tested. The practice of updating it systematically as new signals arrive is what distinguishes rigorous foresight from one-time guesswork.
When a first commercial deployment is announced, update the buyer-identity assumption. When a major player enters through acquisition, update the competition-intensity assumption and potentially the price-point assumption. When a regulator publishes a final rule, update the deployment-timeline assumption. Each update should narrow the range — or, in some cases, widen it if the signal reveals a variable that was not previously in the model.
NASA's Technology Readiness Level framework is a useful reference here: each increment in TRL corresponds to a reduction in technical risk and a corresponding increase in the credibility of commercial projections. Anchoring market-size updates to observable TRL progression — alongside the signal types described above — creates a principled cadence for revision rather than a reactive one.
The analysts who size emerging markets most accurately are not those who produce the most confident initial estimates. They are those who build the discipline of continuous revision into their process from the start — treating every new signal as evidence that should either confirm or adjust the model.
Keep exploring: return to the Practice pillar for the full collection, read how to evaluate an emerging technology for the complementary technical-assessment framework, and see how CanaryIQ puts these methods to work on the venture capital solutions page.