AgenticWeb CensusCheck your site
EN

Technology · September 2026

Is Treasure Data ready for AI agents?

Treasure Data scores 28.1%, 6.2% below the web overall (34.3%).

Official website

Weak28.1%

AWCS

What “AWCS” measures and how

The figures the score is made of, each with its own series: a technology whose score is flat can still have moved on one dimension.

Measured on 106 pages, 97.3% of which carried the agentic category, with a confidence margin of ±5.15.

Read

What Read measures

Lighthouse builds the accessibility tree an agent would traverse and scores the page pass or fail. The denominator is only the pages the audit could evaluate: a not-applicable result or an error is excluded rather than counted as a failure, and the fraction that could be evaluated is published beside it as the coverage.

n=106 · coverage 97.3% · margin ±5.15 pp.

40% of the score
Read on its full scale

Poor12.3%

—
Stable

What Stable measures

Lighthouse's Cumulative Layout Shift audit, counted per page: a page passes when Lighthouse rates its layout shift as good. The denominator is the pages the audit scored. It matters to an agent that works the page the way a person does, by taking a screenshot and clicking where it saw a button: if the page has moved in between, it clicks something else.

n=106 · coverage 97.3% · margin ±5.15 pp.

25% of the score
Stable on its full scale

Good78.3%

—
Guide

What Guide measures

The pages that pass the llms-txt audit: they serve a file and Lighthouse considers it valid. The denominator is every sampled page, so this is a fraction of the whole web and not of the sites that have a file. This is the figure the score uses, because a broken file does not guide an agent.

n=109 · coverage 100.0% · margin ±5.08 pp.

20% of the score
Guide on its full scale

Poor19.8%

Guide: 2 months, up▲ +15.7 pp
Act

What Act measures

A count, not a percentage: how many sampled sites register WebMCP tools that an agent can call. It is the audit that defines the Act dimension; the other WebMCP audits measure how good those tools are, not whether they exist.

n=106 · coverage 97.3% · margin ±5.15 pp.

15% of the score
Act on its full scale

Poor14.2%

—
Reachable

What Reachable measures

Read from the robots.txt of the sampled sites: of the AI agents we track, how many are not disallowed. It is gradual rather than a yes or no, because turning away one agent is not the same as turning away all of them. A site with no robots.txt blocks nobody, so it counts as fully reachable, and that is what was measured. This is the only figure that multiplies the score instead of adding to it, because it is a precondition: if an agent cannot fetch the page, nothing else about the page matters. One caveat: a blanket Disallow under User-agent: * is not counted, because the corpus-wide measurement cannot see it either, so both are measured the same way.

n=109 · coverage 100.0% · margin ±5.08 pp.

multiplies the rest
Reachable on its full scale

Excellent92.1%

Reachable: 2 months, no significant change= −2.1 pp

Measured but not scored

These signals are kept apart from the dimensions because, next to them, a low figure would read as a bad grade for something the score never counts.

22.9%
Sites with an AI crawler policy

Sites with an AI crawler policy

Sample
n=109
Coverage
100.0%

Not a Lighthouse audit: it is read from the robots.txt of the sampled sites, and counts the ones whose file names any of the AI crawlers we track. It measures having an explicit policy, whichever way it goes, and it does not count towards the score.

Sites with an AI crawler policy: 2 months, no significant change= +3.1 pp
43.4%
Sites serving an llms.txt

Sites serving an llms.txt

Sample
n=109
Coverage
100.0%

The same llms-txt audit as the figure below, read differently: this one counts the pages that serve something at /llms.txt, valid or not. The distance between the two figures is large, and it is the interesting part: of the pages that serve something, 40.2% serve a file that parses. We believe a good share of the rest are soft 404s, that is, a server answering 200 with an HTML error page, although we cannot check it because we do not fetch those files. So the figure to cite is the valid one below, and this one is best read as an upper bound. Pages whose audit errored are excluded from both, and the coverage says how many that was.

Sites serving an llms.txt: 2 months, no significant change= +1.6 pp

Compared with the rest of the Customer data platform category

Every technology in this category with a published score (5), and the web overall for scale. Treasure Data is the highlighted bar.

Tealium
Tealium: 40.3%
40.3%
Rudderstack
Rudderstack: 37.2%
37.2%
All sites
All sites: 34.3%
34.3%
Segment
Segment: 33.7%
33.7%
Treasure Data
Treasure Data: 28.1%
28.1%
Adobe Experience Platform Identity Service
Adobe Experience Platform Identity Service: 25.7%
25.7%
0%25%50%75%100%
Every technology in the Customer data platform category that the census can publish, against the web overall (34.3%), on the full scale.

Every other figure this month

49.7%
Lighthouse's category score

Lighthouse's category score

Sample
n=106
Coverage
97.3%
22.9%
Sites with an AI crawler policy

Sites with an AI crawler policy

Sample
n=109
Coverage
100.0%
92.1%
Agents that can reach the site

Agents that can reach the site

Sample
n=109
Coverage
100.0%
12.3%
Pages an agent can parse

Pages an agent can parse

Sample
n=106
Coverage
97.3%
43.4%
Sites serving an llms.txt

Sites serving an llms.txt

Sample
n=109
Coverage
100.0%
78.3%
Pages that hold still

Pages that hold still

Sample
n=106
Coverage
97.3%
19.8%
Sites with a valid llms.txt

Sites with a valid llms.txt

Sample
n=109
Coverage
100.0%
15
Sites exposing tools

Sites exposing tools

Sample
n=106
Coverage
97.3%

How this was measured