When an assistant recommends a few brands, it can feel like a black box. The mechanics are not fully public, and they change, but the broad shape is understood well enough to act on. Three factors do most of the work.
Can it read you
Everything starts with access. If the AI crawlers cannot fetch your pages, nothing else matters, because you are not in the pool the model draws from. Clean robots rules, an llms.txt file and content that renders without heavy scripting keep you readable.
Can it quote you
Assistants prefer sources that state a claim plainly and back it with detail. A page that answers the question in its first lines, uses clear headings and marks up its facts with structured data is easier to lift into an answer than a page that makes the model hunt. Clarity is a ranking signal in its own right.
Can it trust you
Models weigh reputation. Citations, reviews and mentions on sources the assistant already trusts raise your odds of being named, especially for competitive questions. This is why authority built off your own site matters as much as the site itself. We explain the wider shift in the rise of AI agents in brand discovery.
What this means for you
You cannot script an assistant's answer, and anyone who guarantees a placement is selling. What you can do is remove the blockers, write answers worth quoting and build honest authority, then track which questions you win and lose. If your brand is missing from the answers, one of these three factors is usually the reason.
See where you stand with a free scan, then work the ranked list of fixes it returns.