AI search is not killing SEO — it's adding a parallel game

The discourse is split between "SEO is dead" and "AI search is overhyped." Both are wrong. Traditional SEO and generative engine optimization (GEO) are two related games with a large shared toolkit and a few real divergences. Treating them as either identical or completely separate both lead to wasted effort.

What's the same

• Crawl, render, index. Both Google and the AI retrieval indexes need to be able to reach your URLs.
• Content quality. Thin or AI-spam content gets demoted in both.
• Structured data. FAQPage, HowTo, Article schema help both rankings and AI extraction.
• Internal linking. Topical authority compounds for both.
• Mobile rendering. Both treat slow / broken mobile renders as a quality penalty.
• Freshness. Both prefer recent timestamps for time-sensitive queries.

Where they diverge

1. Query → URL relationship

Search returns a ranked list of URLs. A user clicks one (or none). AI search synthesizes an answer and cites 1-8 URLs. Most users never click. So traditional SEO optimizes for the click; GEO optimizes for the citation — which is upstream of the click but doesn't necessarily produce one.

2. What gets extracted

Search engines rank the whole page. AI engines extract specific sentences. A 3,000-word definitive guide can rank well on Google and still lose AI citations to a competitor's tighter 800-word page that has the exact extractable answer in the second paragraph. Length is no longer free.

3. Authority signals

Traditional Google search heavily weights link equity and topical authority built over years. AI engines also weight authority but lean harder on per-page extractable quality. A new site can get cited in ChatGPT search results faster than it can rank on Google for the same query.

4. Access policy

Search has one ecosystem (Googlebot). AI search has 10+ bots with different ownership, different opt-out semantics, and rapidly evolving identity. The access-policy question — "which AI crawlers do I allow?" — has no equivalent in pre-AI SEO.

5. Indexing latency

Google's regular index is on the order of hours-to-days behind a freshly published URL. AI retrieval indexes lag more. ChatGPT Search and Perplexity often don't see a URL for a week or more after publish. This means indexing pushes (via Indexing API + IndexNow) compound across both channels.

Where one playbook works for both

1. Publish with FAQ and HowTo schema where applicable.
2. Front-load the answer in the first paragraph.
3. Submit to Google's Indexing API on publish and update.
4. Submit to Bing via IndexNow — feeds ChatGPT and Perplexity retrieval indexes.
5. Maintain a sitemap and confirm coverage in Search Console.
6. Run quarterly audits of robots.txt to make sure access policy hasn't drifted.

Where GEO needs its own moves

• Publish an llms.txt to give AI assistants a curated tour.
• Audit AI bot access explicitly — block training crawlers, allow answer-time fetchers.
• Write paragraphs that work as standalone citations (one claim, one paragraph).
• Add specific numbers, dates, named entities — LLMs disproportionately cite specifics.
• Get cited externally by trusted sources — LLMs aggregate citations across the web when deciding whose URL to surface.

What's overhyped

The volume of "AI citation tracking" tools and "GEO consultancies" is well ahead of any controlled data showing what actually moves citations. Most of what works is the same playbook as good SEO — clear answers, structured data, fresh content, internal links — applied with a few extra moves. If you have to choose between "hire an AI-SEO consultant" and "ship 10 more well-written articles," ship the articles.