Market Sizing and Category Economics: Definition and Executive Primer

Market sizing estimates the amount of demand that fits a declared category, buyer, geography, period, and revenue unit. Category economics asks how buyer value, vendor revenue, delivery cost, supplier cost, and competitive pressure move through that boundary. Executives need both views because a wide value pool can coexist with a narrow, expensive, or slow-to-adopt vendor opportunity. The useful definition therefore includes the choice the model is supposed to improve.

A board discussing capital allocation may need a different boundary from a product leader selecting a first workflow. The board may examine several revenue pools and long-term adoption conditions. The product leader may need the number of organizations with one fundable problem, accessible data, implementation readiness, and an accountable budget owner. Calling both answers a market size without naming the decision makes the figures appear comparable when they are not.

The first object is buyer value: time released, risk reduced, throughput added, or revenue opportunity influenced. The second is addressable vendor revenue: the amount buyers might pay for products and services. The third is provider economics: revenue after model, data, infrastructure, support, implementation, and review costs. The fourth is obtainable revenue: the portion a particular company could reach and serve within its channel, product, capital, and delivery constraints.

These objects can inform one another, but they cannot be substituted. Labor expense is not automatically software revenue, an economic-impact estimate is not a sales forecast, and an addressable market is not a promise of market share. A disciplined primer labels each object before presenting a calculation. That simple practice prevents a strategy discussion from moving between value, spending, and company revenue as if the terms described the same pool.

Write a one-sentence category rule that a researcher can apply consistently. For an agentic workflow category, the rule might require a system to pursue a defined business objective through more than one authorized step, use approved context or tools, and produce an inspectable disposition. The rule should also state exclusions, such as stand-alone drafting assistance, general cloud infrastructure, or consulting revenue that is already included inside an implementation price.

Next define the eligible buyer using observable conditions rather than aspiration. Conditions may include workflow volume, system access, process ownership, data quality, risk posture, procurement capacity, and a reason to change the current approach. An organization can recognize a costly problem yet remain outside the serviceable market because its process is undocumented or its authority model is unresolved. Eligibility is a filter, not a judgment about the importance of its need.

Core revenue comes from offers purchased primarily for the defined job. Adjacent revenue comes from products that solve part of it. Substitutes include internal development, incumbent software, outsourced labor, process redesign, and the option to do nothing. Enabling layers include models, data, infrastructure, identity, observability, and other inputs used to deliver the offer. The map should show commercial relationships rather than add every layer into one impressive total.

Overlap is especially important when revenue flows through several companies. A customer may pay an application vendor, which pays a model provider and a data supplier. Adding all three receipts as end-customer category demand counts part of the same economic event more than once. An executive model can still analyze each layer, but it should label gross ecosystem flows separately from final buyer expenditure and provider net revenue.

Observed data may include government business counts, regulatory filings, procurement records, disclosed contract information, invoices, usage records, or a documented sample of buyer decisions. Each source has limits. A company count does not reveal workflow readiness, a survey response does not equal a funded purchase, and a vendor disclosure may combine several products. Source authority matters, but fit to the exact question matters just as much.

Create a source register with publication date, measurement period, geography, covered population, currency, original definition, and known exclusions. Record whether the model uses the source directly, derives a value from it, or treats it as an upper bound. A secondary summary can help discover evidence, but a material input should return to the original source whenever possible. Missing methodology should reduce confidence rather than invite a convenient interpretation.

Inference begins when the analyst maps an observed population to an eligible segment, applies a readiness rate, estimates adoption, or translates workload into annual revenue. Those steps are legitimate if they are visible. Every transformed value should name its formula and rationale. A workbook that colors observed inputs, calculated outputs, and judgment assumptions differently is easier to challenge than a polished chart that conceals their origins.

Confidence should follow the weakest material dependency, not the authority of the best source in the file. If the buyer count is well observed but willingness to pay is based on a few exploratory conversations, the revenue estimate remains sensitive to pricing evidence. State that limitation directly. The model becomes more useful when it identifies which uncertain input deserves the next research dollar instead of averaging uncertainty into a falsely precise total.

Hypothetical model: assume a researcher starts with 12,000 organizations in a defined region. Treat 35 percent as having the required workflow volume, 50 percent of that group as technically and operationally ready, and 20 percent of the ready group as plausibly purchasing within the chosen period. The modeled purchasing population is 420 organizations: 12,000 multiplied by 0.35, 0.50, and 0.20. None of those inputs is observed unless a cited project later supplies evidence.

Assume, again only for illustration, an annual contract range of 40 to 70 labeled currency units. Multiplying 420 modeled buyers by that range yields 16,800 to 29,400 units of annual vendor revenue. The model should not round this into a market claim. Its value is diagnostic: the analyst can see whether buyer eligibility, readiness, purchase timing, or price has the greatest effect and can design research to replace the hypothetical inputs.

Suppose the hypothetical provider can onboard 24 customers in the period and its reachable channels expose it to 180 eligible prospects. If a modeled 15 percent of those prospects purchase, obtainable customer count is 27, but delivery capacity reduces the operational ceiling to 24. At a hypothetical 50-unit contract, obtainable annual revenue is 1,200 units. This is not market share; it is a company constraint model based on declared inputs.

Now assume each contract consumes 18 units of variable delivery and supplier cost plus 9 units of allocated implementation and support. The illustrative contribution before other company costs is 23 units per contract. This second calculation prevents the demand model from implying that every revenue unit is equally attractive. Change the exception rate, supplier route, or support burden and the economics can move even when the addressable buyer count remains unchanged.

Run low, base, and high cases, but do not move every assumption together. Adoption may rise while support burden also rises. Lower supplier cost may encourage more usage. Stronger governance may lengthen initial deployment while improving acceptance later. Change one material input at a time and rank its effect on revenue, contribution, and capacity. The largest sensitivity becomes a research priority and a monitoring signal.

Use explicit failure conditions. The thesis may weaken if paid conversion remains low after qualified evaluation, if retained use falls after initial pilots, if exceptions require disproportionate human review, or if an incumbent substitute solves the job at lower switching cost. These conditions are not predictions. They are predeclared reasons to narrow, pause, or redesign the opportunity before sunk cost turns an assumption into organizational commitment.

A market model cannot guarantee category growth, customer adoption, attainable price, product reliability, regulatory acceptance, or company execution. It cannot convert stated buyer interest into signed demand or prove that released employee time becomes cash. Forecast horizons increase uncertainty because product definitions, provider costs, substitute behavior, and buyer controls can change. The model should carry a refresh date and a change log for these reasons.

The analysis also does not replace legal, accounting, investment, or procurement judgment. Currency treatment, revenue recognition, data rights, competitive claims, and regulated use require qualified review where material. An executive can still act under uncertainty, but the decision record should identify what is observed, what is inferred, what is modeled, and what remains unresolved. Honest limits make the range more actionable, not less ambitious.

Executives can use the primer to test whether a category narrative supports capital allocation. Investors can use it to separate ecosystem value from vendor revenue and company reach. Product and strategy leaders can use it to choose a segment or workflow. Required evidence includes a boundary statement, source register, buyer filters, pricing rationale, workload unit, delivery assumptions, sensitivity analysis, failure conditions, and a date for refreshing the decision.

The first deliverable should be a reviewable model rather than a promotional chart. A reviewer should be able to trace every material output to an observed source, a calculation, or a labeled assumption. Conflicting evidence should remain visible. If no source supports a critical input, the answer is not to hide the gap; it is to lower confidence, create a research plan, or restrict the decision until better evidence exists.

Within an applicable and verified configuration, OmegaOS can connect a market question to Hermes research, source references, confidence labels, decision ownership, Forge work, economic observations, and later learning. That operating chain can help a team preserve why an assumption was adopted and what evidence later changed it. It does not make an inferred market size observed, and it does not authorize publication of a claim that lacks support.

A proportionate bridge begins with one decision and one evidence packet. The team records the category boundary, model version, review owner, stop conditions, and next signal. Later observations can be compared with the prediction without rewriting history. Current product access, connectors, data sources, and automation posture must be verified for the intended use. The accountable executive, finance owner, and specialist reviewers still determine what the evidence means and what action follows.