®A growing share of searches end without a click. The user reads a synthesised answer, and the only question that matters commercially is whether your brand was one of the sources it drew from. That is winnable, but not with the tactics that won featured snippets.
Be an entity before you are a page
Answer engines reason about things — companies, products, people — not URLs. If your organisation is described three different ways across your site, your profiles and your press coverage, you are three weak entities instead of one strong one.
- One canonical name, spelled and cased identically everywhere
- Organization schema on the site, with sameAs pointing at the profiles you actually control
- A single clear description of what you do, reused rather than rewritten per platform
- Consistent address, founding details and contact points across every listing
Write things that can be lifted
A model composing an answer needs a self-contained statement it can attribute. Long preambles, copy that only makes sense after three paragraphs of context, and claims buried mid-sentence are all hard to quote.
- Answer the question in the first two sentences under the heading, then elaborate
- Use headings that match how people phrase the question, not internal jargon
- Keep definitions, specifications and prices in structured, self-contained blocks
- Attach the specifics — numbers, units, conditions — to the claim, not to a footnote
A good proxy: could a stranger copy one paragraph out of context and have it still be true and useful? If not, it will not be cited either.
Structured data still does the plumbing
Schema markup does not make a model like you, but it removes ambiguity about what a page is. Product, FAQPage, Article, BreadcrumbList and Organization are the ones that earn their keep. Mark up what is genuinely on the page — invented markup is both a policy risk and a trust problem when the summary contradicts the page.
Mentions are the new links
Links still matter, but unlinked brand mentions in credible places now carry real weight for models assembling a view of who is reputable in a category. Coverage, comparisons, directories, community threads and documentation that names you all contribute — which makes digital PR and community presence a search activity, not just a brand one.
Measure what you can, admit what you cannot
Attribution here is genuinely harder than classic search. Referral traffic from AI surfaces is partial and inconsistent, so treat citation frequency as its own metric: track how often your brand appears in answers to the questions you care about, on a fixed prompt set, checked on a schedule. It is a survey, not analytics — and it is still more honest than pretending the traffic report explains it.