Machine equity
Machine equity is the business asset that arises from AI systems being able to recognize, understand, recall, recommend — and, on a user's behalf, transact with — a brand. Composed of three layers: rented reach (retrieval visibility), owned memory (trained model recall), and machine buyability (agent-side transactability). Measured honestly only as a portfolio of ranged, per-layer metrics — never as a single blended score.
Updated ·
The three layers
Each of the three layers of machine equity has its own name, and each has to be measured separately. The definitions below are our published vocabulary entries, verbatim. The concept itself, the buyer behaviour behind it and the way the layers reinforce one another are covered in the Machine equity essay.
Rented reach
The retrieval-side component of machine equity: whether AI systems find and name a brand in live answers. Re-decided at every query — valuable but never owned, like a well-placed rented billboard. Built and measured on the GEO/AEO tier.
Owned memory
The parametric component of machine equity: what a model knows about a brand from trained memory alone, with no internet access — measured as unaided recall per model and language, always with confidence ranges. Built and measured on the KEO tier; changes only on training-generation cycles (4–16 month lag).
Machine buyability
The transaction-side component of machine equity: whether an AI agent can parse, compare, and complete a purchase from a brand's catalog and checkout — structured product data, agent-payment protocol readiness, machine-readable offers. The AI-transactions (B2M) tier.
How we measure it
Each layer needs a different instrument. Rented reach is measured on live engine answers: we ask the same buyer questions repeatedly and count how often the brand is named. Owned memory is measured closed-book, with the model answering from trained memory alone, without search. Machine buyability is measured on how readable the catalog and the checkout are to a machine.
We do not report a single combined score, because the three layers move at different speeds. Rented reach shifts in weeks, owned memory in model generations. One number would hide that difference and promise a precision that does not exist. Every measurement therefore comes with a range.