How AI assistants choose sources
How assistants decide what to cite
An assistant answering a buying question does not read the whole web. It selects a short list of sources it can retrieve, judge and quote. Being on that list is the work.
Finding candidate sources
Assistants that search the web draw on search indexes and other retrieval systems, so a page has to be findable and readable before it can be used. Clear HTML, fast pages and crawl access for each assistant's bots all matter. A page that is blocked or hard to parse is rarely quoted.
Judging what to trust
Assistants tend to favour sources that answer the question directly, state their claims plainly and are supported by other reputable sites. Academic work on generative engine optimisation, including a 2024 Princeton study, found that adding citations, quotations and statistics to content improved its visibility in generative answers. We treat that as a working principle rather than a guaranteed formula.
Choosing what to name
When an assistant recommends suppliers, it leans on brands it can connect to a topic: consistent company details, reviews, mentions on trusted sites and content that shows real expertise. Platforms differ and change over time, so we test the answers buyers actually see rather than assume how they work.
Why it matters commercially
A buyer who is named in an AI answer at the moment of comparison arrives with you already on the shortlist. We measure whether that shows up as branded search, direct visits and enquiries, not only as mentions.