# What agent-readiness means

Agent-readiness is not a search ranking and not a marketing score. It is the answer to
one question: **when an AI shopping agent answers a buying question in your category,
does it name you?**

That splits into two independently measurable halves.

## The two halves

**Are you in the answer?** Ask a model family a real shopper question and read what it
says. Either your brand appears or it does not. This is measured by asking four model
families five tailored questions and having a judge classify every response. It produces
[visibility](../audit/scoring.md#visibility) and
[recommendation rate](../audit/scoring.md#recommendation-rate).

**Can a machine read your site correctly?** Independently of what any model currently
says, does your site expose the structured facts an agent needs — what you sell, what it
costs, whether it is available, what others say about it. This produces
[technical readiness](../audit/scoring.md#technical-readiness) and drives the
missing-signal penalties.

The two halves fail independently, which is why they are reported separately. A brand
can be well-marked-up and still invisible, or widely recommended with an unreadable site.

## Why this is not SEO

Search returns a list; you can be tenth and still get the click. An agent returns *an
answer*, frequently naming two or three options. Position eleven does not exist. The
measurable unit is presence or absence in a generated response, not rank in a list.

The second difference is observability. You can see your search traffic. You cannot see
what a model told a shopper about you last Tuesday — there is no referrer, no log, no
console. The only way to know is to ask the models yourself, repeatedly, which is what
an audit run is.

## AEO and GEO

Both terms are in use, and the reports use them for the two halves:

| Term | Covers | Reported as |
|---|---|---|
| **AEO** — answer engine optimization | Whether you appear in generated answers | visibility, recommendation rate |
| **GEO** — generative engine optimization | Whether your site is machine-legible | technical readiness |

## What the score is for

The score exists to be acted on, not admired. Every point it deducts is attached to a
named cause — a specific question you were absent from, or a specific signal your site
lacks — and every cause becomes a ranked fix. A score with no attached causes would be
a number without a next step.

It should be read with its error bars. See
[Measurement reliability](../audit/reliability.md).