AI Search Isn’t SEO: how to get cited by AI Overviews 👍💡
I watched “Scrunch’s Head of Marketing Kevin White on being cited, not just found” on The Agile Brand and he’s right, painfully right. Ranking is optional; being a reliable source is not. If you want citations from AI Overviews, here’s the play I actually ship for clients and my own stuff too: write extractable claims with clear attribution, make your entities unambiguous, add schema that truly disambiguates, and measure citations and mentions across engines, not just rankings. I’ll show you the page pattern, the entity hygiene, and the measurement setup I use because opinions are cheap and instrumentation pays. No fluff, just the parts that move. (Agile Brand episode)
// My reaction
Reaction GIF: Mind Blown
- **Optimize to be cited**Short, source‑ready claims with provenance beat generic long‑form
- **Entities first**Make your org, people, products unambiguous to machines
- **Schema supports, it doesn’t save**Markup amplifies clarity; it won’t fix mushy content
- **Measure the right thing**Track citations and mentions, not just rankings
“Ranking is optional; being a reliable source is not.”
What this video gets right
Kevin’s core point tracks with what I see across AI Overviews, Copilot, Perplexity, and ChatGPT browsing and I mean every single week in the wild: be cite‑able or be invisible.
- AI answers need quotable, source‑clear statements. If a claim is specific, scoped, and attributed, you earn a shot at the link. Google’s own AIO update says AI Overviews include relevant links so people can explore further, which underscores the need for clear, supportable claims (Google blog).
- Entity disambiguation now decides inclusion. If machines can’t tell exactly who said what, they won’t risk citing you. Google’s Organization structured data docs say properties like sameAs help Google uniquely identify your organization (Google Organization SD).
- Formats that map to intents get extracted. Definitions, steps, comparisons, and tight tables travel well into answers. That’s consistent with featured snippet guidance on clear, concise answers that tools can lift (Google featured snippets).
Sanity check: visibility moves. Independent tracking showed AIOs dropped to roughly 15% of queries in early June 2024 after early churn, so don’t fixate on one surface; get cite‑able everywhere (Search Engine Land).
What changes in your playbook
Here’s what I’d change this week if you’re bleeding clicks to AI answers and yes I’d do it in this order because order matters.
- Lead with the claim, not the story. One to two crisp sentences that answer the query. Then expand. That top chunk is what engines can lift. It also forces clarity. If you can’t state it tight, you didn’t earn it.
- Attribute non‑obvious facts. Link to primary sources. If it’s your data, say so, and show method, sample, and date. No hand‑waving. If you need a deeper how‑to, I wrote my approach here: How to show up in AI search (my field playbook after Marvin Chow’s chat).
- Tighten entities. On every key page, make Organization/Person/Product details consistent: name, logo, sameAs, address, IDs. Same strings everywhere. Don’t let “ACME, Inc.” and “Acme Widgets” fight.
- Use the right schema, correctly. Article, Organization, Person, Product, FAQPage, HowTo. Fill the fields that actually disambiguate (author, datePublished, about, mentions). My write‑up shows what moved and how we measured it: Does schema markup help AI search? Practical proof and metrics.
| Type/Area | Fields that matter most |
|---|---|
| Organization | name, logo, url, sameAs, foundingDate, address |
| Person | name, affiliation, jobTitle, sameAs, url |
| Article/HowTo/FAQPage | headline, author (Person/Org), datePublished/dateModified, about, mentions, mainEntity |
| Product | name, sku/gtin/mpn, brand (Org), review/aggregateRating (if real), model |
Does ranking top 3 still matter if AI Overviews are present?
Yes, but less. If you’re cited, you can win visibility without a top slot.
Will adding FAQ or HowTo schema guarantee citations?
No. Schema helps clarify. Clarity and evidence earn citations.
How do I get cited for competitive terms?
Niche the claim. Publish original data or clear comparisons others avoid.
Should I produce less content now?
Produce fewer, sharper, source‑ready pieces. Depth beats volume.
- AI Overviews: Google’s own update and links in AIOs
- AIO visibility dropped to ~15% of queries (June 2024)
- Featured snippets favor extractable formats
- Structured data policies: helpful, not guarantees
- Organization structured data helps disambiguate
- Microsoft Copilot plugin citations
- Kevin White, Scrunch: “Get Cited, Not Just Found” (YouTube)
Why these pieces matter across engines
- Perplexity says every answer includes citations and even flags “Trusted” sources. If you publish tight, attributed claims, you raise your odds of inclusion there (Perplexity Help, Trusted labels explained here).
- ChatGPT Search will cite when it actually searches the web, and eligibility assumes you allow its search bot. OpenAI states that sites opted out of OAI‑SearchBot “will not be shown in ChatGPT search answers” (OpenAI Help, OpenAI Crawlers).
- Microsoft Copilot’s extensibility docs outline how plugin responses attach citations, which tells you the product principle: claims need sources ([Microsoft Learn](https://learn.microsoft.com/en-us/microsoft-365/copilot/ext
