What changed (and what didn't)
Two things are true at once:
1. AI search is taking real share from traditional Google. Roughly 1 in 4 information queries now happens in ChatGPT, Perplexity, Google AI Overviews, or another LLM-powered surface. 2. The fundamentals of being findable haven't changed. Authority, clarity, structured content, real citations, and a site Google can crawl — all still required, all still the foundation.
The businesses winning in AI search are mostly the ones who already won at traditional SEO. Here's how to be one of them.
How AI search actually picks sources
LLMs don't browse the way humans do. They:
- Pull from a curated index of pages they've crawled and trust
- Heavily weight pages that other authoritative sites link to
- Prefer pages with clear structure (H2/H3 hierarchy, lists, tables, FAQs)
- Cite sources that directly answer the user's question in the first paragraph
- Look for content with clear authorship and dates
The 7-part playbook
1. Write answer-first paragraphs
Stop burying the lead. The first 1–3 sentences of every section should directly answer the question that section's heading promises.
2. Use a clear H2/H3 outline
LLMs parse headings to understand what your page is about. Vague headings like "Our Approach" tell them nothing. "How much does residential roof replacement cost in 2026?" tells them everything.
3. Add FAQ sections with FAQPage schema
Q&A format is gold for AI search. It's exactly the format the model wants to ingest.
4. Include data, dates, and specifics
"Most roofs last 20–25 years" beats "roofs last a long time." LLMs prefer concrete claims they can attribute.
5. Get cited and linked by authoritative sites
The single biggest predictor of AI citation is whether other trusted sites already trust you. Local press, industry publications, and partnerships still drive this.
6. Maintain a public author / "About" presence
LLMs increasingly weight author credibility. A real About page with real names and credentials matters more than ever.
7. Make sure you're crawlable
Sounds basic — but if your robots.txt blocks GPTBot, ClaudeBot, or PerplexityBot, you've voluntarily made yourself invisible to AI search. By default, allow them.
What about Google AI Overviews specifically?
Google's AI Overviews pull primarily from sites already in the top 10 organic results. The implication: traditional SEO is the entry ticket to AI search visibility on Google. There's no shortcut around ranking well.
What about ChatGPT and Perplexity?
Both prefer authoritative, well-cited sources. Both surface specific claims with links. Both reward businesses with:
- A real Wikipedia mention (hard but high-leverage)
- Mentions in industry publications
- A consistent, factual digital footprint across reviews, press, and social
What's NOT working
- Keyword stuffing — already dead, more dead now.
- Thin AI-generated content — LLMs can sniff this out and de-rank it.
- Hiding behind walled gardens (Facebook-only, Instagram-only) — AI can't cite what it can't crawl.
The new measurement question
You can no longer measure visibility purely by rank. Add to your monthly reporting:
- Branded search volume (proxy for AI mentions driving people to look you up)
- Direct traffic growth (proxy for AI-driven discovery)
- Mentions in AI answers (manually checked monthly for your top 5 queries)
The bottom line
AI search rewards the same things old-school SEO rewarded: clarity, authority, and a real reputation. Small businesses that double down on those will be cited more, not less, as AI eats the rest of search.