AI has split search into two surfaces: the ranked results everyone optimised for twenty years, and the answers assistants now compose on top of them. This page explains how visibility works on the second surface, using controlled research rather than vendor decks, from a provider whose own pages earn AI citations.
$199, $950, and $2,500 monthly, month to month. Every figure lives on the pricing page.
Controlled research supports a short list, and it is not the list most agencies sell. In order of confirmed weight:
Meaning one is workflow: using models for research, drafts, and analysis. Done with editing and judgment it compresses production time. Done raw it produces the generic text search systems have been actively demoting, so the tool is an accelerant in either direction.
Meaning two is visibility inside AI answers, sometimes labelled GEO or AEO. Strip the acronyms and it is structural: say the answer early, format it liftably, be resolvable as an entity, and be mentioned where models look.
What connects them is that neither replaces the foundation. AI systems draw from indexes built by crawlers, so a page a crawler struggles with is a page no model will ever quote.
Schema as a citation lever: 1,885 treated pages against 4,000 controls, no meaningful uplift on any platform. Keep schema for rich results and entity clarity; stop paying for it as AI magic.
Platform specific hacks: the systems share sources and behaviour shifts monthly, so tactics tuned to one assistant decay on contact. The durable work is the five factors above, which hold across all of them.
And volume as a strategy: publishing more generic pages faster is the one approach the last two years have punished most consistently.
Two things at once, and the muddle between them is where bad advice lives. It is using AI tools to do SEO work faster, and it is optimising your visibility inside AI answers themselves: Google AI Overviews, ChatGPT, Claude, and Perplexity. The second meaning is the one reshaping where buyers actually see businesses.
Traditional SEO wins a ranked position a person clicks. AI SEO wins being the source an answer engine quotes, which requires liftable structure: direct answers placed early, ranked lists, and an entity the system can resolve confidently. Good traditional SEO is the foundation; AI SEO is the extraction layer on top.
Not by itself, on the best controlled evidence. A study tracking 1,885 pages that added structured data against 4,000 matched controls found no meaningful citation uplift on any AI platform. Schema still earns rich results and entity clarity, which are worth keeping. It is not the citation lever it is commonly sold as.
Yes when it is edited into something with real experience and specificity, and increasingly no when it is published raw. Search systems have been demoting generic machine text and rewarding first hand evidence. The tool is fine; unedited output is the risk.
Place a direct answer in the first third of the page under the question it answers, use ranked list formats on commercial topics, keep your business facts identical everywhere you appear, and cover the sub questions systems generate around the main one. Ranking helps and is not sufficient alone.
Usually not as a separate line item. The honest structure is AI answer optimisation built into every campaign, because the same page should win both the ranked result and the citation. At Portland Peak SEO it is included in every tier rather than sold as an upgrade.
Google AI Overviews first, because it sits on top of existing search volume. ChatGPT, Claude, and Perplexity next, growing fast and drawing from overlapping sources. Optimising the underlying signals covers all of them at once, which is the practical reason platform by platform tactics are mostly noise.
Track which of your pages get cited for your money questions across the major assistants, alongside classic positions. Movement is lumpy and platform behaviour shifts, which is exactly why the recorded baseline matters more here than anywhere.