One agent client is refused at the edge
When somebody asks an assistant about you, it answers from whatever it can reach. That is not your site.
100% of your page text needs JavaScript
These clients do not run JavaScript. They are being handed an empty page and answering from somewhere else.
One of the Five AI clients can't read this page at all.
6 of your 14 findings have a file below with your own URLs in it. The rest are a punch list ordered by effort, and a prompt that applies the lot one commit at a time.
Covers kaplan.com. Another domain later is $5.
Your files open here in the browser the second you pay. We email a copy too, so check your spam folder for it: we are a new domain and filters have not met us yet.
Serve this at https://kaplan.com/llms.txt. It tells a reading agent which of your pages matter, in the order they matter.
# kaplan.com
Retrievability is costing you the most — 21 of the 85 points you're missing. Open any line to see what we measured.
We read at most 6 pages a scan, sequentially, one second apart, and obey your robots.txt on every page rather than only the first. The full catalog and the weights are published, and blocking us takes one line.
This site sends a header that forbids putting it in a frame, so we are not showing one. That is a deliberate setting and a reasonable one; the name above opens it.
This page has no meta description, so there is no sentence of its own to show here.