Each site is scanned once, first the first site and then the second. Nothing is stored. Allow up to a minute.
What the comparison shows
Each site gets the same scan as the free scanner: a score out of 100, three pillar scores, every check and the access state of each AI crawler. The page lists only the differences. Checks and crawlers where both sites match are left out, and there is no winner and no verdict, only the figures.
The scores measure readiness: whether AI crawlers can reach a site and whether its homepage carries the signals machines read. They do not show whether any assistant names either brand in an answer.
The MCP server offers the same comparison as its compare_sites tool, for use inside Claude, Cursor and other MCP clients. See the MCP page.
Common questions
What does the comparison cover?
It runs the same scan as the free scanner on each site and shows where the two results differ: the score out of 100, the three pillar scores, each check and the access state of each AI crawler. Checks and crawlers where both sites match are left out. It reads homepages and public files only.
Is anything stored?
No. The page sends each domain to the scanner, which reads public files from that domain and returns the result. The only thing kept is in your address bar: a link with the two domains, which you can copy or share. To enforce its rate limit the scanner keeps a one-way hash of your IP address and a request counter, which expires within a day and is not linked to the domains scanned.
Why are the two sites scanned one after the other?
Each site gets its own full scan, exactly as if you had run the free scanner on it. Running them in turn keeps within the scanner’s rate limit, and it means a problem with one site, such as a timeout, does not stop the other from showing.
Next step
Where the other site is ahead on crawler access, use the robots.txt generator. Where it is ahead on structured data, use the schema generator. Where it has a file you lack, use the llms.txt checker.