How to Get ChatGPT to Cite Your Small Business

Getting ChatGPT, Perplexity, Gemini or Google's AI Overviews to cite your small business comes down to a small number of things done well: open every section with a direct answer, include a comparison table, phrase your H2s as questions, cite authoritative sources yourself, add a FAQ with FAQPage schema, keep the page fresh, and get discussed on Reddit and LinkedIn. This is the practical recipe.

If GEO itself is new to you, start with our plain-English primer: What is GEO (Generative Engine Optimisation)? A UK Guide. This post is the hands-on how-to that follows it.

How do ChatGPT, Perplexity, Gemini and AI Overviews decide what to cite?

Each engine crawls the open web with its own bot, but the citation logic differs. Knowing how each one works tells you what to prioritise.

ChatGPT. When you ask a question, ChatGPT's search layer (using OpenAI's OAI-SearchBot crawler) pulls a handful of live web pages, extracts passages, and stitches an answer with 3–5 source citations. It leans heavily on domains it already trusts on a topic — 2026 research from Kime.ai found that roughly 30 domains account for around 67% of ChatGPT citations within any given topic. Page speed matters too: pages with First Contentful Paint under 0.4 seconds average 6.7 citations versus 2.1 for slower pages.

Perplexity. Perplexity runs the most transparent citation system — every source is listed inline and at the bottom of the answer. It uses a multi-stage retrieval, rerank, and authority-check pipeline, and has the strongest recency bias of any major AI search engine. Perplexity picks up new well-structured content within 1–2 weeks of publication — faster than any of the others.

Google AI Overviews and Gemini. Google's AI answer engine (AI Overviews at the top of search, plus Gemini in a separate app) starts with Google's usual ranking signals, then reranks candidates on passage-level extractability and "entity density" — how many recognised names, tools, prices and places appear per thousand words. Notably, 47% of AI Overview citations come from pages ranking below position 5. You do not need to rank first; you need to have the clearest passage.

The common thread across all four: structured, well-labelled content wins. Wall-of-text pages lose.

The 7-step playbook to get cited by AI search engines

Do these seven things and you have covered roughly 90% of what actually moves the needle in 2026. None of them require a big budget.

1. Open every section with a direct answer. AI engines extract the first sentence or two of a section. If your opener is preamble ("In this article we will explore…"), the AI skips to a competitor. Write each H2's opener as a self-contained 40–60 word answer to a specific question, then expand underneath. Recent 2026 analysis of Google AI Overviews puts the sweet spot for an extractable passage at 134–167 words — enough to answer fully, short enough to lift verbatim.

2. Add a comparison table. Comparison tables are still one of the most-extracted formats on Google AI Overviews, Gemini and Perplexity, and they convert human readers. But this is the tactic that changed most in 2026: on 6 August ChatGPT roughly halved its listicle citations and Google demoted the same kind of thin roundup page twelve days later. Genuine comparisons with real numbers and first-hand testing still get cited; "10 best X" pages written to game citations do not. If you write about anything with options (products, prices, plans, tools, providers), put the comparison in a proper HTML table with a header row, then add a worked example and something only you know.

3. Write H2 headings as questions. Users phrase questions in ChatGPT the same way they say them out loud — "how much does a small business website cost?", "which card reader is best for a market stall?", "what's the difference between Xero and QuickBooks?". H2s that mirror those exact phrasings match the incoming query and dramatically raise the chance your section is the one cited. Rewriting "Our Pricing" as "How much does X cost?" costs nothing.

4. Cite authoritative sources yourself — they cite you back. This is the biggest counterintuitive finding in GEO. The Princeton and IIT Delhi GEO paper (KDD 2024) tested nine optimisation techniques across 10,000 queries and found "Cite Sources", "Statistics Addition" and "Quotation Addition" produced the biggest gains — each around +30% to +40% visibility. When your page links out to credible sources (government data, research papers, established publications), AI engines treat your page as more citable itself. The citation flows both ways.

5. Add a FAQ section. The FAQ section matters because AI answers are Q&A shaped, and a block of real questions with 40-80 word answers is the easiest thing on your page for an engine to lift. On the schema: Google retired FAQ rich results on 7 May 2026, so FAQPage markup no longer earns a search feature, and there is no reliable evidence it lifts AI citations on its own. Add it if you like (it is harmless), but spend the effort on the questions and answers, not the markup. Aim for 5-6 questions per page.

6. Keep the page fresh — `dateModified` matters more than `datePublished`. AI engines de-rank stale content quickly. Semrush's December 2025 tracking found the median uninterrupted URL citation lasts around 6 days, and a URL cited by ChatGPT today has roughly a 15% chance of still being cited three weeks later. Add a visible "Last updated: [month year]" line to every content page, revisit key posts every 3–6 months, and republish. Do not fake the original `datePublished` — Google flags that as spam. Bump `dateModified` instead.

7. Show up on Reddit, LinkedIn and Quora. A 2026 Profound study of 1.4 million AI citations found Reddit was the single most-cited source across ChatGPT, Google AI Mode, Gemini, Perplexity and AI Overviews. LinkedIn climbed from around #11 in late 2025 to #5 by February 2026 — the fastest authority shift of the year. If your business is genuinely discussed on these platforms, your name enters the citation graph. Answer real questions in your industry's subreddit, publish weekly on LinkedIn, and reply to Quora questions in your field. This is off-page GEO, and it is now roughly 40% of the win.

How do you check if you're actually being cited?

There is no dashboard yet, so most owners never check. That is a mistake — one hour a month tells you whether the work is landing.

The DIY method (free, takes 45 minutes):

1. Write down the 5–10 questions your customers most commonly ask before hiring you or buying from you.
2. Ask each question in ChatGPT, Perplexity, Gemini, and via a Google search that triggers an AI Overview.
3. Log whether your site is named or linked in the answer.
4. Repeat monthly and track whether you move up or down over time.

The paid method: tools like Otterly.AI, Profound, Semrush's AI Toolkit and Peec.ai automate that same query set weekly and turn it into a report. Prices start around £25–£40 a month at the entry level. Not necessary for a solo business, but useful if you are managing this for clients or want a proper time-series view.

Also check:

• Google Search Console → Performance → filter by "AI Overview" appearance (Google added this filter in 2026).
• GA4 → filter referral traffic by source containing `chat.openai.com`, `chatgpt.com`, `perplexity.ai` or `gemini.google.com`.

Zero cost to run, and it tells you whether all this work is actually turning into visits.

First-person case study: our card-reader post is cited by all four engines

We published Square vs SumUp vs Zettle: Best Card Reader for a UK Small Business in early 2026. Nothing clever: direct-answer opener, HTML comparison table, question-format H2s, per-option "What you get / Best for / Limitations" blocks, a worked example with real numbers, and a 6-question FAQ.

Within a few months, referral traffic started appearing from `chat.openai.com` and `perplexity.ai` in Google Analytics. Google Search Console started showing AI Overview appearances for the query "best card reader for a UK small business". When we ran that query through each engine ourselves, the SME Shack post was cited by name in ChatGPT, Perplexity, Gemini and Google AI Overviews — all four.

The strongest bit of proof came from outside our own analytics. In September 2026, Dale from Jar Pay (a UK payments start-up we know) messaged to say his team had run a fresh UK card-reader query through ChatGPT and the SME Shack article was one of the sources ChatGPT quoted. Independent confirmation from someone who has no reason to inflate our numbers, running the query cold.

One post, not a controlled study. But every ingredient in the 7-step playbook above is in that one article. If you copy the structure honestly, you have a genuine shot at the same outcome.

What doesn't work — the AI-visibility traps

Keyword stuffing. AI engines are trained specifically to detect and demote keyword-stuffed pages. Repeating "best card reader UK" 30 times lowers your citation odds, not raises them. Write for a person reading out loud.

Thin content. 400-word "what is X?" pages do not get cited. AI engines cite the most comprehensive, extractable page on a topic — not the shortest. 1,500–2,500 words is the honest range for a comparison or how-to post.

PR spam and mass guest posts. The old link-buying game does not translate. AI engines look at the quality and context of a mention, not the raw count. Fifty mediocre backlinks do almost nothing; one genuine Reddit thread discussing your product does more.

"Guaranteed AI visibility" services. There is no paid slot inside a ChatGPT or Perplexity answer. Anyone charging you for "guaranteed citations" is selling snake oil.

Ignoring the off-page half. A perfectly structured page still underperforms if nobody discusses your business elsewhere on the web. On-page structure is roughly 60% of the win; off-page presence (Reddit, LinkedIn, Quora, industry blogs) is the other 40%.

How long does it take to get cited?

Realistic timeframes, based on public data from Semrush and the tools above:

• Perplexity: 1–2 weeks for a well-structured page on a topic people actively search. Perplexity is fastest by a clear margin.
• ChatGPT: 4–8 weeks. Semrush found around 8% of new pages get cited within 24 hours, 17% within a week, and 42% within 30 days.
• Google AI Overviews: 4–12 weeks. Moves fastest of the four initially but also loses citations most erratically as its ranking mix shifts.
• Gemini: Similar cadence to AI Overviews since they share the underlying model.

The honest expectation for a UK small business: publish one well-structured post today, expect early Perplexity citations within a month, ChatGPT citations within two months, and consistent multi-engine visibility within three months. Citations are not permanent — the median citation life is short — so refreshing every 3–6 months is part of the ongoing job, not a one-off launch task.

This landscape changes fast. The specific percentages will look different in a year; the framework — clear answers, tables, FAQ schema, freshness, off-site presence — is what stays useful.

Frequently asked questions

Q: Do I have to submit my site to ChatGPT or Perplexity?

A: No. Both engines crawl the open web using their own bots (OAI-SearchBot for ChatGPT, PerplexityBot for Perplexity, GoogleBot for AI Overviews and Gemini). Check your `robots.txt` doesn't block them and they will find your content on their own. There is no submission form — anyone selling you one is misleading you.

Q: Can I do this without a developer?

A: Mostly, yes. Writing direct answers, question-format headings, comparison tables and FAQ content is copywriting — any business owner can do it. The only bit that needs a developer is FAQPage and Article JSON-LD schema markup, which is an hour of work. Most modern platforms (WordPress, Webflow, Shopify) have plugins that handle it automatically.

Q: Which AI engine should I focus on first?

A: Google AI Overviews if you have to pick one — they now appear on a large share of UK informational search results, so the volume is highest there. In practice, a well-structured page tends to be cited across all four engines simultaneously, so you optimise once and benefit everywhere.

Q: What if my industry is very niche?

A: Niche is an advantage. AI engines cite the best-structured source they can find, and niche topics usually have fewer well-structured pages competing for the citation slot. A UK dog groomer with one honest "which nail clipper for a small dog" comparison post will out-cite big pet chains that only publish thin product-page copy.

Q: Will AI engines eventually stop showing citations?

A: All four major engines currently show source links, and there is regulatory and legal pressure keeping it that way — publisher lawsuits in the US and EU largely depend on citations-with-links being visible. Even in a worst case where a specific engine dropped them, the underlying skill (clear, honest, well-structured content) is the same skill that wins on Google, in email and on social. GEO is not a fad tactic — it is just good publishing.

Q: Should I write for AI or for humans?

A: Both, and they are the same thing. AI engines are trained to reward content a human reader would find clear, honest and useful. If your page reads well to a customer, it reads well to an AI. If you find yourself writing awkward keyword-stuffed sentences to "trick" the AI, you are doing it wrong — that content will be demoted, not cited.