PLATFORMS · Chapter 3

Optimizing for ChatGPT, Perplexity, Claude, and Copilot.

Four engines, four citation models. ChatGPT leans on its own crawl plus a Bing-derived index. Perplexity blends a live fetch with its own crawl. Claude reads deeply into a single source. Copilot grounds on the Bing index without a separate crawler. The on-page work overlaps; the per-platform tweaks are where citation share is won.

KEY TAKEAWAYS

Perplexity is usually the fastest engine to win citations on — passage discipline alone moves it.
ChatGPT rewards entity-graph investment more than the other engines — Wikidata and LinkedIn presence matters.
Claude rewards depth — fewer, longer, structured pages outperform many short ones.
Copilot is Bing SEO with extra steps — and Bing index coverage is the most common silent gap for SMBs.

Five engines side by side

Platform	How it retrieves	What lifts a citation	SMB action
ChatGPT (Search)	OAI-SearchBot indexes; ChatGPT-User fetches live when needed. Grounded on a Bing-derived index plus OpenAI's own crawl.	Clean canonical URL, fast HTML response, Article schema, and presence on third-party sources the model already trusts (Wikipedia, Wikidata, LinkedIn, industry directories).	Allow OAI-SearchBot in robots.txt. Ensure key pages render real HTML on first response (no JS-only content). Build out the entity graph (sameAs, Wikidata, LinkedIn).
Perplexity	PerplexityBot crawls and indexes. Perplexity-User fetches in real time on user queries. Both behave like a hybrid of search engine and live agent.	Inline citations that match the user's query intent, dense fact-per-paragraph content, recent dateModified, and being the page that answers the question most directly (not most comprehensively).	Allow PerplexityBot and Perplexity-User. Re-write key pages so the first paragraph directly answers the page's title query. Keep dateModified honest — Perplexity de-ranks stale content visibly.
Claude (Search and Computer Use)	Claude-SearchBot indexes; Claude-User fetches live when an agent or user runs a query. ClaudeBot is the training bot — separate from the retrieval stack.	Long-form structured pages with clear hierarchy, table-driven comparisons, and explicit entity disambiguation. Claude reads more deeply into a single source than ChatGPT often does.	Allow Claude-SearchBot and Claude-User. Invest in pillar pages with internal navigation — Claude will follow links inside a single domain to compose an answer.
Microsoft Copilot	Bingbot — Copilot grounds on the Bing index, no separate crawler. IndexNow speeds up Bing-side discovery.	Bing Webmaster Tools verification, clean indexability in Bing (not Google — they diverge), IndexNow submission for fresh content, and schema Bing's parser reads cleanly.	Verify the site in Bing Webmaster Tools. Submit a sitemap to Bing separately from Google. Enable IndexNow if the CMS supports it. Audit Bing index coverage — it is usually 60-80 percent of Google's.
Google AI Overviews + AI Mode	Googlebot. Gemini grounds on the regular Google index. Google-Extended is the opt-out token for Gemini training (separate from grounding).	All standard Google ranking factors plus query-fan-out coverage, passage-shaped content, and entity-rich author/publisher schema.	Treat AIO and AI Mode as outcomes of strong classical SEO plus passage discipline. Run query fan-out audits on priority pages to find sub-queries the page does not cover.

Sources: OpenAI bots documentation, Anthropic crawler reference, Perplexity bot guide, Microsoft Bing crawler reference. Tactic notes are from José's implementation experience.

The optimization sequence that actually works

Trying to optimise all five engines at once is how SMBs end up with thin progress everywhere. The sequence below puts the highest-leverage work first.

Google AI Overviews and AI Mode first. They ground on Google's index, so every existing SEO investment compounds into them. Run query fan-out audits on the top 10 pages.
Perplexity second. Re-write the opening paragraph of every priority page to answer the title query directly. Keep dateModified honest. Single biggest lever for Perplexity citation share.
ChatGPT third. Ship Person + Organization schema with sameAs links. Create LinkedIn company page if it does not exist. Build a Wikidata entry. ChatGPT favours entities with a presence outside the brand's own domain.
Copilot fourth — and it is mostly Bing hygiene. Verify in Bing Webmaster Tools, submit a sitemap, enable IndexNow, and check Bing index coverage. Most SMB Bing coverage is 60 to 80 percent of Google's — the gap is the opportunity.
Claude last. Audit pillar pages for depth, internal navigation, and table-driven structure. Claude follows links inside a single domain to compose answers, so cluster architecture rewards Claude more than the other engines.

Setup effort and time to first citation

Not all platforms require the same investment. This table maps the realistic effort and timeline for each engine so you can sequence the work by payback speed.

Platform	Setup effort	Time to first citation	Primary investment
Google AI Overviews + AI Mode	Low — builds on existing SEO	Weeks (if already ranking)	Query fan-out audit, passage shaping
Perplexity	Low — passage discipline only	Hours to days (live-fetch)	Rewrite opening paragraphs, honest dateModified
ChatGPT (Search)	Medium — entity graph required	Weeks to months	Wikidata entry, LinkedIn company page, sameAs schema
Microsoft Copilot	Medium — Bing hygiene gap	Days after Bing indexing	Bing Webmaster Tools, sitemap, IndexNow
Claude	Low-Medium — pillar depth	Days (live-fetch)	Pillar page depth, internal navigation, table structure

Which engine to prioritise — and when

Optimise for Google AI Overviews first if…

You already rank on page 1 for any target keyword.
Your ICP starts research on Google before moving to ChatGPT.
Your content program publishes regularly and has accumulated links.

Prioritise Perplexity if…

You need the fastest measurable citation wins (hours, not months).
Your content is factually dense and updated frequently.
Your ICP is technical — developers, analysts, ops leaders.

Invest in ChatGPT entity graph if…

Your buyers are senior decision-makers who use ChatGPT for vendor research.
Your brand has zero Wikipedia or Wikidata presence.
You are building long-term domain authority, not just page-level wins.

Fix Copilot/Bing first if…

Your Bing index coverage is below 70% of Google's (check Bing Webmaster Tools).
Your ICP is enterprise IT or Microsoft ecosystem buyers.
You have already done Google SEO work and want fast downstream wins.

Best for

Sites that already have classical SEO foundations and want to extend visibility into AI answers.
B2B and professional services SMBs whose buyers research across multiple AI surfaces before contacting sales.
Brands building thought leadership where being the cited expert in answers is more valuable than the click.

Fails when

You try to optimise all five engines simultaneously in the first quarter.
You skip the entity-graph work because Wikidata and LinkedIn feel like a different team's problem.
You expect Copilot citation share without Bing-side hygiene.

Verify before

Robots.txt allows OAI-SearchBot, Claude-SearchBot, PerplexityBot, Perplexity-User, and Bingbot.
Person and Organization schema are live with sameAs links to LinkedIn and any directory profiles.
Bing Webmaster Tools is verified with a current sitemap.

Frequently asked

How do I get cited by ChatGPT?

ChatGPT Search retrieves citations through OAI-SearchBot and live-fetches with ChatGPT-User. The retrieval ranks pages that return clean HTML on first response, carry Article schema with author and publisher identity, and live on a domain with strong third-party signals (Wikipedia, Wikidata, LinkedIn, industry directories). Allow both bots in robots.txt, ship the schema stack, and invest in the entity graph.

Why is Perplexity easier to win than ChatGPT for SMBs?

Perplexity rewards directness and recency. A page that answers the user's question in the first paragraph, carries an honest dateModified, and has dense fact-per-paragraph content can be cited even from a low-authority domain. Perplexity-User fetches in real time, so a freshly published page can be cited within hours — unlike ChatGPT, which depends more on indexed content.

Does Claude actually drive measurable traffic?

Claude's referral volume is small compared to ChatGPT and Perplexity in 2026, but the qualified-buyer signal is high — the people using Claude for research are disproportionately technical decision-makers and senior operators. Optimising for Claude is a high-margin, low-volume play. Claude-SearchBot indexes and Claude-User fetches live, so the technical setup is identical to OpenAI's bots.

Is Copilot worth optimising for separately, or is it just Bing SEO?

Copilot grounds on Bing, so the foundation is Bing SEO. That said, the work pays off twice — every fix you ship for Copilot also lifts Bing organic rankings. Submit a sitemap to Bing Webmaster Tools, verify the property, enable IndexNow if your CMS supports it, and audit Bing index coverage. Bing-indexed pages are usually a smaller subset than Google's, and the gap is where Copilot citation share leaks.

Which platform should an SMB optimise for first?

Start with Google AI Overviews and AI Mode — they share Google's index, so existing SEO work compounds. Then layer in Perplexity (fastest wins from passage discipline), then ChatGPT (heavier dependency on entity graph), then Claude and Copilot. Doing all five at once is not the right sequence for most SMBs — the foundations work pays back across all of them.