TAM (total addressable market) is the revenue opportunity available to a product or service if it captured every viable customer in its target market. It's one of the first numbers investors ask about, and one of the first numbers founders get wrong.
Most TAM calculations look precise on a slide and fall apart when someone asks a follow-up question.
The standard approach takes a broad industry number, applies an assumed percentage, and produces a large figure that feels strategic but isn't actually usable. You get a ceiling estimate, but not a model you can filter, stress-test, or update when your market shifts.
The more defensible approach starts from the bottom up. Instead of asking how large the market is in theory, you ask which specific companies match your structural definition of a viable customer. That shift turns TAM from an abstract revenue ceiling into a queryable dataset, and enriched company data is what then makes it possible.
Traditional TAM modeling depends on inputs that flatten real market complexity: analyst reports, NAICS categories, macro revenue estimates, and broad company counts. These inputs assume companies within a labeled industry are roughly comparable and equally addressable, as in static enough to treat as a uniform pool.
They're not. A broad "B2B software" category includes early-stage startups, mature enterprises, consulting-heavy hybrids, and companies actively declining.
Treating these categories as a single unit actually inflates theoretical TAM and hides where actual opportunity lives. Top-down TAM answers a macro question, not an operational one.
A dataset-driven model works differently. Instead of starting with total industry revenue and applying narrowing assumptions, you start with individual companies and filter on structural attributes: employee count, geography, industry classification, founding year, and total funding raised.
With enriched company data, the goal shifts from estimating a ceiling to identifying a defined universe.
Defining your target company profile structurally
Before counting anything, you have to define what qualifies as a realistic customer. "Mid-market SaaS companies" is not a definition. A structural one combines industry classification, employee count, geography, founding year, and total funding raised. Enriched company datasets let you layer these filters precisely rather than approximate them.