Methodology

Sourced, layer-by-layer, defensible.

We don't hand you a single black-box score. The TAM model is a chain of distinct measurements — each one cited, each one stress-tested, and each one replaceable with the merchant's own data as the relationship deepens.

Anchored in published research

Every input traces to publicly available data — direct APIs, published benchmark reports, peer-reviewed studies.

Measured per merchant

Defaults get replaced with the merchant's own measured values as audits run. Real data beats triangulated research every time.

Stress-tested on three scenarios

The softest input is bracketed across conservative, mid, and aggressive — the team picks whichever diagonal they can defend.

Five layers, one chain

Each layer is a multiplication. Output of one is input to the next. Compounding errors are why methodology rigor matters — a small error early in the chain becomes a large one at the end.

1

Search demand

How big is the merchant's category right now?

Real data
2

AI share

What fraction of that demand is shifting to LLM-powered shopping?

Stress-tested
3

Share-of-voice

How often do AI agents mention this brand specifically?

Measured per audit
4

Funnel economics

What converts a citation into a sale and a customer?

Real data
5

Lifetime value

What is one acquired customer worth over the lifetime of the relationship?

Real data

= Annual TAM, bracketed by three scenarios

The chain produces a defensible annual addressable market number per merchant — projected forward across 12, 24, and 48-month horizons, and bracketed by the three sensitivity diagonals so the team picks the one they'll defend.

Three scenarios — never a single number

The softest input gets stressed across three diagonals. We don't hide the uncertainty in a confidence interval — we surface it explicitly so the team lands on whichever scenario they can defend with their own data.

Conservative
Middefensible
Aggressive

Mid scenario × 12-month horizon is the diagonal most merchants anchor on — it sits inside published industry projections and tracks the merchant's own historical growth band.

Three principles to anchor the model

01

The chain compounds

Errors multiply, not add. A small mismeasurement early in the chain becomes a large one at the end. The 3-scenario stress test is the answer — not a hedge.

02

Defaults get replaced with measured data

Anywhere we ship a default, the model is wired to swap in the merchant's own measured value. The measurement infrastructure is the same loop that generates the audit — the model gets sharper every time.

03

Time curve and ROI are independent

Forward projections (12, 24, 48 months) and pilot ROI math ride on top of the chain — they don't inherit the soft layers' uncertainty. The team can stress-test each independently.

The path from default → measured

Each merchant relationship deepens the model. The order of replacement is roughly:

  • Share-of-voice — replaced from the first audit run onward, scoped per merchant, never extrapolated from another tenant's data.
  • Halo / channel mix — replaced when the merchant connects a marketplace presence (e.g. Amazon URL) so the lift ratio comes from real public data instead of a category default.
  • Funnel economics — replaced when the merchant connects their commerce backend so AOV, conversion, and order frequency come from their own books.
  • AI share — replaced as cross-tenant measurement infrastructure accumulates enough data to derive empirical per-category ratios; until then, stress-tested.

Try free auditHomePricing