Agent Readiness Compare

How to read an Ora score

What the 0 to 100 number, the letter grade and the four layer scores mean, and what they leave out.

In short

An Ora score is a 0 to 100 number built from four weighted layers: Discovery 20, Access 30, Usability 40 and Payments 10. The letter grade is a band on that number. Layers that do not apply to a site leave the calculation, so read the layer scores and the grade together.

The number is four layers weighted by how much each matters

Ora says its score follows an agent through four layers: find you, access you, use you, pay you. The layer weights are Discovery 20, Access 30, Usability 40 and Payments 10. Usability carries the most weight, so a site that agents reach but cannot use loses more than a site that is hard to find.

Layer maxima are not fixed at those weights per site. For claytonfarms.com the Ora API returned Discovery 6 of 7, Access 32 of 44, Usability 12 of 17 and Payments 0 of 0, for a final 75 (B), from a scan stored on 2 November 2026. Read each layer as a share of the checks that applied to that site, not as points out of 20, 30, 40 or 10.

The grade is a band, and the bands are uneven

The grade bands below are Ora's own, read from its methodology and API on 2026-11-04.

Ora grade bands (Ora methodology, read 2026-11-04)
GradeScoreOra label
A+95 to 100Leading
A86 to 94Agent-Ready
B70 to 85Competitive
C48 to 69Needs Work
D28 to 47At Risk
F0 to 27Unusable

The bands are not equal width. B spans 16 points and A spans 9. A move from 69 to 70 changes the grade, while a move from 49 to 68 does not. Compare scores, not only letters, when you track change.

Layers that do not apply leave the denominator

Ora excludes checks that do not apply, and bonus checks that a site did not earn, from the denominator. A free tool is therefore not marked down for missing payment rails. The visible sign is a layer that reads 0 of 0. On 2026-11-04 ora.ai's own Payments layer showed 0/0: excluded, not failed. If a layer on your report reads 0 of 0, do not chase it.

Wait for the final reading too. The API field analysisStatus is "partial" while deep checks run, and Ora says final scores need "complete". A screenshot taken during a partial run can understate a site.

What to do with it, by reader

  • Head of web: use the grade as a ranking against peers, then ask which layer drags it. Ora's leaderboard average was 43 and 6 percent of sites were at B or above on 2026-11-04, so a C is not an outlier. Those are Ora's own figures for sites that were scanned with Ora.
  • Technical marketer: ask for the layer scores, not the headline. A site at 70 from strong Discovery and weak Usability needs different work from one at 70 the other way round.
  • Platform engineer: store the full response, the scan date and analysisStatus. Cached scores move within a day, and the Concordance kept the stored scan date for each Ora cell.

Where the number is weaker than it looks

Ora does not rank itself first on its own leaderboard. It scored ora.ai 84 (B) on 2026-11-04, below Telnyx at 100 and Vercel at 89. Its methodology page lists 126 checks while its API catalogue listed 132 the same day, six more in Usability. Ask which catalogue a given score used.

The Concordance sample is not the leaderboard. Its 60 domains were spread across grades on purpose, and 25 of the 60 scored B or above, against the 6 percent leaderboard figure. Its Ora mean of 64.1 says nothing about the web at large. The sample's ten Tranco sites, which nobody tuned for Ora, are the control; see the method.

What this guide cannot tell you

It cannot say whether a higher score makes an agent complete more tasks for you. Ora publishes the score and, separately, a journey run that sends a real agent at a site, but the score itself is a set of checks, not a task result. It cannot say how a site would score under another scanner; see why scanners disagree for that. For the product itself, read the Ora profile.

Sources

  1. Ora, llms-full.txt (layers, scoring, analysisStatus). https://ora.ai/llms-full.txt, read 2026-11-04.
  2. Ora, methodology (grade bands). https://ora.ai/methodology, read 2026-11-04.
  3. Ora, leaderboard statistics. https://ora.ai/api/stats, read 2026-11-04.
  4. Ora, check catalogue. https://ora.ai/api/checks, read 2026-11-04.

Last reviewed 2026-11-05