AI agent readiness benchmark: how 1,287 online stores score

No store earned an A

RivalSweeper scores each store from 0 to 100 on how machine-readable and agent-ready it is: crawlability, structured data, llms.txt and entity signals. Of 1,287 stores scored in the 60 days to 30 September 2026, not one reached an A (70 and up). The average was 25.4 and the best store scored 68.8, so the market as a whole sits in the D band.

llms.txt, schema.org and AI crawlers

48% of the scored stores publish an llms.txt file and 42% expose schema.org structured data. Only 3% block AI crawlers.

Checkout rails are increasingly handed out by the platform. What stays in the brand's hands is being found, read and trusted by an AI assistant.

Edition trend

September 2026 is the first Online Retail Radar edition to publish the readiness benchmark, so there is no earlier month to compare yet. This page adds each new edition's figures when they are published.

Method and sample

• Stores scored 0-100 on how machine-readable and agent-ready they are: crawlability, structured data, llms.txt and entity signals.
• 1,287 stores scored in the 60 days to 30 September 2026. Grade A means a score of 70 or more.
• Public pages only; results are aggregated and no store or brand is named.

Full edition, with every section and sample size: The Online Retail Radar, September 2026. Methodology for the whole series: Online Retail Radar.

Questions