Our production analysis “Why HTML-to-LLM Scraping Fails” is now permanently archived on Zenodo, the open-access repository operated at CERN.

The preprint is citable under a permanent DOI: 10.5281/zenodo.21797856, published under a Creative Commons Attribution 4.0 (CC BY 4.0) license. The archived PDF extends the web version with the full methodology, a results table, and a discussion of limitations - and the measurements remain reproducible on any public website with browser developer tools alone.

A companion analysis of the browser-LLM guidance protocol behind our concierge, “Guiding, Not Clicking”, will follow on Zenodo as well.

You can cite the analysis as: Coman, R. (2026). Why HTML-to-LLM Scraping Fails: A Production Analysis. Zenodo. https://doi.org/10.5281/zenodo.21797856