When was the last time you typed your business name into ChatGPT? Open a new chat and ask it to recommend the best business in your category in your suburb. If your name does not come up, you are not alone. Most South African businesses are completely invisible to AI, and that quiet phone you are blaming on the economy might actually be a citation problem.

ChatGPT local business optimisation is the practice of structuring your business information so AI tools like ChatGPT, Perplexity, and Google AI Overviews confidently recommend you. It works by making your name, reviews, and content easy for AI engines to verify across trusted sources. The benefit is being chosen inside AI answers, where customers now decide who to call.

The Short Version

  • ChatGPT and most AI tools do not pull live local data. They rely on training data and a handful of trusted real-time sources.
  • Your Google Business Profile is still the single biggest signal, even for AI engines that do not officially use Google.
  • Reviews on Google, Yelp, and Hello Peter shape how AI describes you, not just how it ranks you.
  • Bing Places quietly feeds ChatGPT directly. Most SA businesses have never claimed theirs.
  • Third-party citations from directories, articles, and Reddit drive over 65 per cent of AI recommendations.
  • The fix is not a trick. It is consistent, verifiable presence across the five or six places AI engines actually read.

Why ChatGPT Does Not Know You Exist (and Why Most Agencies Get This Wrong)

Here is the thing. Most agencies handling local SEO in 2026 are still optimising for one platform: Google. That made sense five years ago. It does not make sense now.

Roughly 2 billion people use Google AI Overviews every month, and tools like ChatGPT, Perplexity, Claude, and Gemini are quietly fielding “best plumber near me” and “good accountant in Durban” queries that used to live inside Google search. The customer flow is the same. The route to your front door is not. We covered the underlying shift in detail in our piece on why Adobe just paid $1.9 billion for Semrush, but the local-business angle deserves its own treatment.

What most agencies miss is that AI engines do not work like Google. They do not crawl your site, rank it, and serve it back in a list of ten blue links. They pull from a layered cake of training data, real-time web fetches, third-party APIs, and trusted directories. If your business name only appears on your own website, you are essentially a rumour. The AI cannot verify you, so it confidently picks someone it can.

“We have stopped chasing rankings and started chasing citations. Our clients are seeing three times the qualified leads from AI traffic alone.”

Karl Worner, Founder of ohKarl

This is the gap. Closing it is not technical wizardry. It is hygiene applied across five or six sources you probably already have accounts on, plus one or two you did not know mattered.

The 7 Steps to Make AI Engines Recommend Your Local Business

Work through these in order. Most can be done in an afternoon. The compounding effect kicks in within four to eight weeks.

Step 1: Audit your Google Business Profile

Log in at business.google.com. Confirm the category is specific (Italian restaurant, not just restaurant). Add at least 10 photos taken in the last 90 days. Fill every service field. Add the full physical address with a Plus Code. Set accurate opening hours including public holidays. This single profile feeds Google AI Overviews and gets quoted, often word for word, by ChatGPT when it runs a live search.

Step 2: Claim and complete Bing Places

Go to bingplaces.com and either import your Google profile or set up from scratch. This step takes 15 minutes and almost no SA business does it. ChatGPT uses Bing under the hood for real-time search. If you are not on Bing Places, you are not in the live answer set.

Step 3: Fix your NAP across every directory

NAP stands for Name, Address, Phone number. Pick the exact format you want (for example “ohKarl, 4 Sandalwood Crescent, Ballito, 4420, 064 555 1234”). Then update Yelp, Hello Peter, Brabys, Yellow Pages South Africa, Apple Maps, and any industry directory you appear on. Inconsistencies confuse AI engines and they default to whichever business has cleaner, matching data.

Step 4: Build a review engine

AI tools read review text, not just star averages. A 4.9 rating with five vague reviews (“great service”) loses to a 4.6 with fifty detailed reviews (“Karl helped us redesign our brand and our leads doubled in two months”). Set up a simple post-job text or email that asks the customer for a Google review with one specific question: “What problem did we solve for you?” Aim for one new review per week, minimum. Read that again. One per week, every week.

Step 5: Publish one location-specific answer page per month

Pick a real question a customer in your area asks. Write the answer as a page on your site, with a clear H1, a 40 to 60 word summary at the top, and a structure of questions and answers below. Title example: “How much does a logo design cost in Durban in 2026?” These pages match the way customers actually ask AI for help, which is why AI engines reward them. Our guide on growing your local business online goes deeper on the page-by-page template we use.

Step 6: Add LocalBusiness schema

Schema markup is invisible code that tells search engines and AI exactly what your business is, where, what you do, and your hours. Use Google’s Structured Data Markup Helper or a WordPress plugin like RankMath to add LocalBusiness schema to your home page. This is the single highest-impact technical task in the list. It takes 30 minutes and most SA sites have never done it.

Step 7: Get cited on third-party sources

Over 65 per cent of AI recommendations come from sources other than your own site. Get listed in industry roundups, get quoted in local press, answer questions on Reddit and Quora using your real name and business. Pitch one journalist a month. Comment thoughtfully on three relevant Reddit threads per week. This compounds. Within six months you will be the name AI engines pull when someone in your suburb asks.

If you want a hand auditing your site against all seven steps, our SEO, AEO and GEO services include a 14-point audit that maps directly to the criteria above.

Free download

AI Citation Audit Checklist

The 14-point checklist we use to audit every client’s AI citation readiness, with the exact tools, settings, and queries we run. No email gate.

Download the PDF

How to Test Whether It Is Working

Once your seven steps are in place, you need a simple way to know if AI engines have caught up. Here is the test we run with every client, once a month.

Ahrefs walks through exactly how AI search engines decide which businesses to recommend. Worth 12 minutes of your time.

The principle is straightforward: ask the same five questions to four different AI tools and log who gets mentioned. If your name is not in the answer, that is your gap. If your name is in the answer but the description is wrong, your reviews and content need work.

AI tool What to test What it tells you
ChatGPT “Recommend the best [your service] in [your town]” Whether you are in the considered set at all
Perplexity Same query, then check the cited sources panel Which directories and articles AI trusts about you
Google AI Overviews Type the query into Google as a signed-in user Whether Google considers you authoritative enough to summarise
Gemini Same query, then ask “why did you pick that one?” The exact signals Google’s AI is reading about you

A Durban-based accountant we worked with ran this test on their firm in February. ChatGPT did not mention them once across 12 prompts. We worked through the seven steps above with their team. By mid-April, ChatGPT cited them above two of the big four firms when asked for “best small business accountant Durban”. Their lead form started receiving four to six AI-attributed enquiries a week, where it had been zero.

If your business is not showing up yet, that is your roadmap. Get in touch with ohKarl and we will walk you through the same audit, or read our breakdown of how SEO, AEO and GEO actually fit together for the bigger picture.

Frequently Asked Questions

Can ChatGPT recommend local businesses?

Yes, but with a catch. ChatGPT recommends businesses it has training data on, plus businesses that appear prominently in its live web search results. If your business is only listed on its own website, ChatGPT will not confidently recommend you. You need to be cited across multiple trusted sources for AI to surface you reliably.

Why does ChatGPT not know about my business?

Most likely because your business is invisible to the sources ChatGPT was trained on, or pulls from in real time. That usually means a thin Google Business Profile, no Bing Places listing, no schema markup, and very few third-party mentions. ChatGPT cannot recommend what it cannot verify, so it defaults to better-cited competitors.

Does ChatGPT use Google reviews?

Indirectly, yes. ChatGPT pulls from web sources that aggregate Google review data and from articles that reference review content. It also reads review text itself for sentiment and context. A common misconception is that AI tools only look at star ratings. They actually read the words inside reviews, which is why detailed reviews outperform vague five-star ones.

How long does it take to show up in ChatGPT recommendations?

For most local businesses, four to twelve weeks. Google Business Profile and Bing Places updates take effect within days. Building the third-party citations and reviews that AI engines trust takes longer. Plan for a three-month investment before you see consistent AI recommendations, with results compounding from month four onwards.

Do I need to pay for AI search visibility?

No. Every step in this guide can be done for free. ChatGPT, Perplexity, and Google AI Overviews do not currently sell paid placements inside their recommendations. You earn visibility through cleaner data and better third-party presence, not ad spend.

Sources: Ahrefs AEO course on how AI search engines work, Neil Patel on local SEO for LLMs, Semrush AI Overviews user data.