What classic SEO still gets right (and wrong) for AI search
This phase ends where strategy debates usually begin: should we do SEO or "AI optimization"? The framing is wrong, and seeing why is the synthesis of everything above. The two surfaces share a foundation, diverge in specific places, and the divergences are learnable. Here's the honest ledger.
What classic SEO still gets right
The entry ticket is unchanged. Retrieval gates everything: pages that don't get crawled, indexed, and ranked don't enter AI answers. Phases 4 and 5 are not legacy work.
Content that answers real questions wins everywhere. Phase 4's content strategy - buying guides, comparisons, fit content, built from actual customer questions - is precisely the inventory fanout retrieves. Nothing to unlearn.
Authority still compounds. Being mentioned, linked, and reviewed across the web mattered for rankings; it now also shapes what models learn about your brand and which sources they cite. PR and SEO were always cousins; they've merged.
And the deepest continuity: structured, complete, honest product data was the right investment in 2020 for filters and feeds, and is the right investment now for extraction and fact-checking. The catalog was always the asset.
Where classic instincts mislead
Four habits need updating. Volume-first targeting: keyword tools can't see fanout queries, so "no volume" no longer means "no value" - specificity is the new premium. Position obsession: rank three of ten links is a win; mention three of three brands might not be - the AI game is more winner-take-most, which raises the stakes on being the definitive source rather than a competitive one. Length as quality: the 3,000-word comprehensive guide optimized for dwelling on page can extract worse than a tight, front-loaded, well-structured 1,200 words. On-site tunnel vision: classic SEO could mostly be played on your own domain; AI visibility is substantially decided on pages you don't control - the roundups, reviews, and comparisons that assistants cite. Earning presence there is now core work, not nice-to-have PR.
The both-surfaces strategy
The closing move of this phase is noticing there's only one strategy, weighted differently: make your catalog the best-structured source of truth about your products (Phases 1–2), make it technically readable everywhere - pages, schema, feeds, all agreeing (Phases 3, 5, 7), build content that answers the question-space around your products, structured for extraction (Phases 4 and 6), earn presence in the sources machines trust, and measure both surfaces - GSC for classic, the quarterly sampling audit for AI. Every item on that list pays on both surfaces simultaneously. That's not a coincidence; both systems are, in the end, trying to connect people with reliable answers - and the durable optimization has always been being one.
Next phase: Feeds & Marketplaces