If you've read how these engines choose citations, you know the core problem: ChatGPT reads Bing, Gemini reads Google, Claude reads Brave, and Perplexity reads its own index.
This article is the work itself. Do it in order — the early steps decide whether the later ones can function at all.
Step 0: Redeploy your project with analytics and tracking installed
This translates to adding Google Analytics, Google Search Console, and Bing Webmaster Tools to your business webiste. Ask ChatGPT or any other answer engine how to install Bing Webmaster Tool or GA, or GSC. It is very easy to setup.
Step 1: Get indexed in all three search engines
Almost every business we audit is indexed in Google and nowhere else. That's one engine out of four.
- Google Search Console — you probably have this. Submit your sitemap. Powers Google + Gemini.
- Bing Webmaster Tools — free, ~10 minutes, and you can import directly from Search Console. Powers ChatGPT. This is the single highest-leverage neglected task in the field.
- Brave Search — has its own independent index. Powers Claude. Check whether you're in it by searching
site:yourdomain.comon Brave.
There's no clever trick here. It's a chore. It's also the difference between being invisible to ChatGPT and not.
Step 2: Restructure content as answers
Write the question the way a customer says it. Not "Spring Replacement Services" but "How much does a garage door spring replacement cost in Oklahoma City?"
Answer it in the first sentence, with the actual number. Compare:
❌ "At Acme, our certified technicians take pride in delivering exceptional value on every spring replacement, tailored to your unique needs."
✅ "Garage door spring replacement in Oklahoma City costs $150–$350, depending on door weight and whether it uses torsion or extension springs."
The first is unquotable — it contains no fact. The second gets lifted verbatim into an answer, because it is the answer.
Then add the depth underneath. Specifics, edge cases, comparisons, what changes the price. That's where you earn the click after the citation.
Include a limits section. Counterintuitive but consistently true: pages that state what doesn't apply — exclusions, caveats, when a cheaper option is better — get cited more than pages that only sell. Engines are tuned to prefer balanced sources, and honest limits read as balance.
Step 3: Win Perplexity with granular detail
Perplexity is the most winnable engine for a small business, because ~80% of what it cites isn't in Google's top results. Domain authority matters far less. Specificity and freshness matter far more, and it hands out 20+ citation slots per answer.
What wins here is the operational knowledge you have and have never published:
- Real local price ranges, updated when they change
- What the climate in your area does to your equipment
- Comparisons between two options with an actual recommendation
- Failure modes, warning signs, what to do before help arrives
- Anything with a number, a date, or a place in it
Publish it, date it, and keep it current. Freshness is weighted heavily — a page updated last month outranks a better page from 2023.
Step 4: Win Gemini by ranking in Google
There's no shortcut here, and anyone promising one is selling something. 76% of Google AI Overview citations come from pages already ranking in Google's top 10.
So for this engine, classic SEO is the AEO work: technical health, crawlability, internal linking, page speed, real backlinks. Get into the top 10, then make sure the page is structured to be quotable once you're in the candidate pool.
Anyone telling you SEO is dead and AEO replaces it is describing a strategy that fails on the biggest surface there is.
Step 5: Become a legible entity
This is the GEO layer, and it's what separates "a website" from "a business an engine is willing to name."
- Identical name, address, and phone everywhere — your site, Google Business Profile, Bing Places, Apple Maps, directories, socials. Inconsistency creates doubt, and doubt loses citations.
- Organization schema with a real, crawlable
logoURL. We've audited plenty of sites whose schema pointed at a 404 — invisible to the exact system it's meant to feed. - LocalBusiness schema with service areas, hours, and price ranges.
- Verifiable existence — state registration, licenses, an address that resolves.
- Real reviews with real names, surfaced on your own site, not just on Google.
Engines are answering "who should I trust with this?" Every one of these is a signal that you're a real business rather than a page.
Step 6: Measure what actually gets cited
Ask the engines directly, on a schedule. Query ChatGPT, Gemini, Perplexity, and Claude with the questions your customers ask. Record who gets named and what gets quoted.
Then check your server logs for OAI-SearchBot, Claude-SearchBot, and PerplexityBot — they tell you which engines are actually reading you, and how often. Referral traffic from chatgpt.com and perplexity.ai shows up in analytics as its own source; watch it separately from organic, because it converts very differently.
Build more of whatever gets quoted. Most of the compounding comes from this loop, not from the initial build.
What to skip . This is straight out of Generative Engine when asked what works and what does not. No human modification here.
Being straight about what doesn't work, because plenty of it is being sold right now:
llms.txt— 8.7% of top sites publish one, 39.6% of those are empty stubs, and no major engine has committed to reading it. Google's John Mueller said Search doesn't use it. Useful for developer docs, useless for a garage door company.- Keyword stuffing for AI — engines synthesize meaning; repetition just makes the page worse to quote.
- AI-generated bulk content — it has no specifics, which is precisely what gets cited. Volume without facts produces nothing.
- "Prompt injection" tricks — hidden text trying to instruct the model. It doesn't survive retrieval, and it's a good way to get classified as spam.
The honest timeline
Indexing changes show up within first weeks. Content restructuring shows up in 45 days. Entity and authority signals take 3–6 months to compound.
Anyone promising AI citations next week is lying. What you can have quickly is the diagnosis — which engines can currently see you, and what's blocking the ones that can't.
Proof this works
Open Sesame Garages, a single-technician garage door company in Oklahoma City, is now quoted verbatim by AI search on OKC garage door questions — the $150–$350 spring pricing, the metro coverage across Edmond, Norman, Moore, and Midwest City. Word for word, from FAQ content we built for that purpose.
No franchise budget. No decade-old domain. Just this playbook, executed properly.
Want us to run Step 0 for you right now? We'll check your robots.txt, your indexing across Google, Bing and Brave, and your schema — and tell you exactly which engines can currently see you. Book a strategy call. The first four booked from this site are free.
Sources
- OpenAI user-agent documentation (
GPTBot,OAI-SearchBot,ChatGPT-User) - Anthropic crawler documentation, updated February 20, 2026 (
ClaudeBot,Claude-SearchBot,Claude-User) - Profound, Claude / Brave Search citation overlap analysis
- Ahrefs, analysis of 1.9 million Google AI Overview citations
- 680-million-citation cross-platform domain overlap analysis
- Rankability, llms.txt adoption across the top 1,000 websites, June 2026