Buyability Index
Every venue we've scanned, ranked.
We scan real venues' own websites the way an AI agent would read them, then publish what we found. No opinion, no placement, no fee.
Leaderboards by category
Every venue in a category is tested on the same checks, so the ranking means what it says. Hospitality and ecommerce are scored on the same four pillars but ranked apart, because a booking flow and a checkout are not the same job.
- Venues tested
- 235
- Not yet agent-ready
- 235
- Closed to agents
- 70
- Categories live
- 3
- Checks per scan
- 23
How a score is built
Nothing here is modelled, weighted or bought. A venue's score is a count of what the scan found, and every finding on a venue's page quotes what was actually on their site.
- 01
We scan the public site
The same scan Selfe runs on itself. No login, no manual review, nothing paid for.
- 02
Roughly twenty checks, four pillars
Found, Fit, Trusted, Paid. Each one a real, quoted finding, not an estimate.
- 03
Ranked against the category
Every venue tested on the same checks, so the leaderboard means what it says.
Some sites an agent cannot open at all
Of 384 sites we tried, 70 would not let an automated reader in on two separate runs. Some answer with a block page, a captcha or a challenge; others present a certificate a browser would warn about, and a person clicks through that warning where an agent simply stops.
We count a site as closed only when it turned us away twice, because one failed request can just as easily be a rate limit or a dropped connection. Those venues are not scored here either: we cannot see what is behind the door, and a score we did not measure would be worse than no score.
No fee changes a score
The Index is not a directory listing and cannot be bought. Claiming a page tells us a venue wants a conversation about closing its gaps; it does not change the number, and a rescan is the only thing that ever will.
Choose what suits you
Start as you're meant to go on.
What the scan looks for
Four pillars, roughly twenty checks. Each one reads something specific off the venue's own public site, and every finding on a venue's page quotes what was actually there.
Found
Can an agent discover it at all?
Robots access, sitemaps, DNS records and the agent-facing manifests a registry reads before it ever fetches a page.
Fit
Can it tell what is on offer?
Structured data, machine-readable content and the tool surfaces an assistant connects to, rather than a page it has to scrape.
Trusted
Would it vouch for a booking?
A resolvable identity, signed claims and a documented way to ask permission, not just a name on a page.
Paid
Can it actually complete one?
A real, machine-reachable path from found to paid for: a booking engine an agent could drive and settle against.