Market-level measurement
The Agent-Ready Booking Index
The Agent-Ready Booking Score answers the question for one business. The Index answers it for a market: what share of the restaurants in a city, or the hotels in a country, could an AI assistant actually find, understand and book?
It is the same instrument and the same method version, applied to a fixed sample under identical conditions, so a market can be compared with itself over time and with other markets.
What every index publishes
A figure without its sample is decoration. Each index ships with the conditions that produced it, so anyone can judge the number or reproduce the approach.
Where the list came from, and the date it was collected.
The geographic boundary of the market being measured.
Which businesses are included, and which are excluded and why.
Identical checks, in identical conditions, for every business in the sample.
A share of rows re-checked by hand before any figure is published.
No reservation is ever submitted. The agent stops at the confirmation step.
Where the method stands
v1.0 measures whether the path is readable and completable in principle
The current method inspects the public booking path the way an assistant reads it: is there a findable route to book, are the date, time and party-size fields machine-readable, is the business described in a form a machine can verify, and is there a wall (a captcha, a dead link, an image-only menu) that stops an agent cold. Findings are stated in exactly those terms, never as a completed booking.
v2.0 adds the live agent run, and will be labelled as such
We are building a harness in which an agent walks each booking path in a real browser, fills the fields, reaches the confirmation step and stops. It never submits. To prove that “reached the confirmation step” means “the booking would have gone through”, we validate against venues that have given us permission, including our own booking system. Results measured under v2.0 will say so, and earlier results keep their version label.
Coverage
The first index is Amsterdam restaurants, running now. From there: a second European city, then the first comparative report, then verticals beyond restaurants where the booking path matters as much, such as hotels, clinics and salons.
We would rather publish one index that survives scrutiny than ten that do not. Progress is tracked on the research hub.