In October 2026, Google shows an AI Overview on roughly 68% of local searches. The classic three-business map pack still appears on about 39% of them. For most searches the customer reads a machine-written summary before they ever scroll to a map pin or an organic listing. If your brand is not inside that summary, you are invisible to the majority of searches you used to rank for.
AI Overviews, ChatGPT Search and Perplexity: one problem, three surfaces
The surfaces look different but the mechanics are the same. Every answer engine retrieves candidate passages from the web, ranks them for authority and relevance, synthesises an answer, and cites a handful of sources. Your job is to be in that candidate set and to be the clearest source once retrieved.
- Google AI Overviews pull mostly from Google's own index. Google ranking still matters.
- ChatGPT Search routes queries through Bing's live search tool. Bing ranking matters more than most teams realise.
- Perplexity blends Bing, Google and its own crawl. Entity consistency across sites decides who gets cited.
- Claude and Gemini increasingly cite primary sources with publication dates. Fresh, dated content wins.
The Bing gap most teams still ignore
If a page ranks top three on Bing for a sub-query, ChatGPT retrieves it with high probability. If the same page is page two on Bing but page one on Google, Google Overviews may still cite it but ChatGPT will not. Three things to do this week:
- Verify Bing Webmaster Tools and submit your sitemap. Yes, it still matters in 2026.
- Register for IndexNow with a key file served from your site root. One push notifies Bing, Yandex and the ChatGPT retrieval pipeline that depends on them.
- Audit which of your top Google pages are not even indexed on Bing. Request indexing manually; the UI is slow but it works, and the lift inside two weeks is usually visible.
Content structure that gets pulled into an AI answer
Answer engines prefer passages that stand alone. Think of every H2 section as a candidate answer to a candidate question. Four structural moves that measurably raise citation rate:
- Lead each section with the direct answer in one sentence. Save the rationale for sentences 2 to 4.
- Use declarative sub-headings that read like questions a user would actually type ('How many Google reviews do I need to rank?' not 'Review strategy').
- Put concrete numbers, dates and brand names inline. Models prefer specific sources over vague ones.
- Include a short 'what, who, where, when' block for topical pages. Entity disambiguation is half the battle.
Entity signals: being the thing, not just a page
LLMs reason about entities, not URLs. Your business is an entity; so is your category, your geography and your service. The signals that fuse your brand to the right entities:
- Consistent NAP (name, address, phone) across your website, Google Business Profile, Bing Places, Apple Business Connect and 15 to 20 industry citations.
- Organization schema with sameAs links to Wikidata (claim your entity), LinkedIn, Crunchbase and your main social profiles.
- Author and reviewer bylines on long-form content, with real humans and real credentials.
- Press and partner mentions on domains that are themselves cited by LLMs (industry publications, local newspapers, chamber directories).
The technical layer: robots, schema, llms.txt
Three files are doing more work than most teams know.
- robots.txt. Explicitly allow GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, PerplexityBot, Google-Extended and Applebot-Extended. Blocking any of them in the name of 'content protection' is the fastest way to vanish from AI answers in 2027.
- Schema. Article with datePublished and dateModified, LocalBusiness or ProfessionalService with full address and openingHours, Product or Service nodes for every offering, and BreadcrumbList on every nested page.
- llms.txt. A plain-text map of your site aimed at LLMs. Lists services, pricing, regions and key articles with a one-line description each. Serve it from the root. Takes an afternoon; pays off for months.
Measurement: how to tell if ChatGPT is actually citing you
There is no official 'LLM Search Console' yet. You can still track citations with four free touchpoints:
- Weekly prompt pack. A static list of 20 to 30 target queries that you run through ChatGPT, Perplexity, Google AI Mode and Claude. Note whether your brand appears and in what context. Automate it in a spreadsheet with manual QA.
- Server logs. Filter for the known AI bot user-agents (GPTBot, OAI-SearchBot, PerplexityBot, ClaudeBot). Rising hits mean rising retrieval.
- Referral traffic from chat.openai.com, perplexity.ai, bing.com/chat and google.com with Search Generative Experience parameters. Analytics catches more than most teams think.
- Branded search volume lift. AI citations drive 'learn about brand X' searches days later. Google Trends and Search Console catch the follow-through.
The quiet truth of AI search in 2026: it rewards the same things great SEO always rewarded, plus clarity. Clear answers, cited data, consistent entities, fast fresh content. Teams that chased hacks lost. Teams that wrote clearly, won.