Google ranks pages. An answer engine writes one answer and names a handful of sources inside it. Some of the work overlaps, like schema and content depth, but the target is being cited in the answer rather than ranking tenth on a results page. The measurement is completely different, which is why standard SEO reporting tells you nothing about it.

Influence. Models pull from content that exists, so publishing against the questions your buyers actually ask, fixing what makes you illegible to a crawler, and building the third-party citations models lean on all move the answer. What nobody can do is guarantee a specific phrasing on a specific date. Anyone promising that is selling something else.

Comparison and head-to-head prompts move fastest, usually inside two to three months, because competition on them is thin. Broad prompts take longer and never reach 100%. On our reference account the first real movement showed up around month three.

Visibility first, because nothing else can happen until the models name you. But visibility is the input, not the point. The goal is customers arriving where you actually want them, and that endpoint is yours to choose: a direct booking, a DTC checkout, a retail partner, a phone call, a filled-in form. We agree the destination before we publish anything, then measure back to it.

Yes. Paid media on Meta, Google, TikTok, Bing and Shopping is the other half of Onward, along with the data infrastructure underneath both. Plenty of clients buy only AEO and that is fine. But if you are already running paid media, the same prompt research and landing pages feed it, and one team reporting on both beats two agencies blaming each other.

Partly instrumentation, partly asking. AI referral traffic is trackable and we separate it out, but plenty of buyers read an answer and then search your name directly, so referrer data undercounts. We tie what we can, add a source question to your enquiry form, and are honest about the gap rather than claiming perfect attribution.