BlueWing Local · Field Notes

What ChatGPT actually says when someone asks it to recommend a business like yours.

Two conversations, side by side. The difference between them is the whole story.

Most local business owners have never opened ChatGPT and asked it to recommend a business in their own category and area. If you haven't, I'd encourage you to stop reading this right now, open ChatGPT, type in the question your customers would type, and read what it tells them. What you'll see may surprise you.

For the rest of this post, I want to show you what those conversations actually look like — not in theory, but in real text. The two conversations below were both with ChatGPT, both asking similar questions, both about small businesses in Sandton. The only difference is that one of the businesses had done the structural work to make itself readable by AI assistants, and the other hadn't.

Both businesses below are illustrative — I've invented the names to make the point clearly without identifying anyone real. But the conversations themselves are exactly the pattern you'll see if you do this test yourself.

Conversation One

A homeowner in Morningside opens ChatGPT and types:

Customer asks ChatGPT
"Can you recommend a good electrician for a domestic renovation in Sandton? I'm in Morningside and need someone reliable who does DB board work and small projects."
ChatGPT replies

"I don't have access to live business directories or real-time reviews, so I can't make a specific recommendation for an electrician in Morningside, Sandton with full confidence."

"A few approaches you could try: search Google Maps for 'electricians near Morningside Sandton' and filter by rating, ask in local community groups on Facebook, or check the Electrical Contractors Association of South Africa directory for accredited members in your area."

"For a small domestic project, personal recommendations from neighbours or your estate agent often give the best results."

Notice what's happened. The customer asked a specific question and received a generic, deflecting answer. ChatGPT essentially said "I can't help you — go look somewhere else."

That deflection isn't because ChatGPT was being cautious or careful. It's because no specific electrician serving the Morningside area had given ChatGPT enough information to recommend them with confidence. So ChatGPT did what it does in those situations — it told the customer to search Google instead.

The customer in this conversation now goes to Google. They might find an electrician there. They might not. But one thing is certain: every electrician in Sandton lost this customer the moment ChatGPT shrugged.

Conversation Two

A few days later, a different customer opens ChatGPT and types:

Customer asks ChatGPT
"I need a tax consultant who can help a small business owner in Sandton with VAT registration and ongoing bookkeeping. Can you suggest one?"
ChatGPT replies

"For a small business in Sandton needing VAT registration and ongoing bookkeeping, you might consider Greenfield Advisory in Hyde Park, run by Karen Whitfield."

"They specialise specifically in small owner-operated businesses, offer fixed monthly fees rather than hourly billing, and handle both SARS registration and ongoing compliance. Their initial consultation is free. They tend to be a good fit for businesses turning over between R1 million and R10 million annually."

"You can reach them through their website. For something larger or more complex, you might also want to consider a chartered accountant rather than a tax consultant."

Greenfield Advisory is fictional — invented to illustrate the contrast.

Look at what just happened. ChatGPT gave the customer a specific business name, a location, the owner's name, a clear description of who the business is right for, the pricing model, what's included in the first contact, and an honest qualifier about when not to use them.

The customer doesn't go back to Google. The customer calls Greenfield Advisory.

This is what AI recommendation looks like when it works. The customer gets a clear answer. The business gets a qualified lead who has already been told why they're a good fit. The competition gets nothing.

Why the difference exists

You might assume that the second business was simply bigger, more established, or better known. None of that is necessarily true. AI assistants don't recommend based on size or popularity. They recommend based on what they can read and trust about your business.

The fictional Greenfield Advisory was AI-visible because someone — its owner, a consultant, an organised developer — had done the structural work that allowed ChatGPT to form a confident picture of it. The business had:

  • A complete, verified, accurate Google Business Profile.
  • A website where the identity, services, pricing model and ideal client were stated in plain language.
  • Consistent contact details everywhere the business appeared online.
  • Specific information about what the business does and who it's best suited to.
  • Owner credentials and credibility signals — the things AI looks for to decide whether to recommend with confidence.

The electricians in the first conversation had none of these things in place — or if they did, the information was scattered, inconsistent, or buried in places AI couldn't read. So ChatGPT couldn't recommend any of them. It defaulted to "go search Google" instead.

The difference between the two conversations isn't about marketing. It isn't about advertising spend. It isn't about size, age, or reputation. It is about whether the basic structural work of being AI-readable has been done.

What this means for you

Open ChatGPT now. Type in the question your most valuable customer would type — "best [your service] in [your area] for [their situation]" — and read what ChatGPT says.

If your name comes up with confidence and accuracy, you are in the second conversation. You already have AI visibility, and your job is to maintain it as competitors notice and try to catch up.

If your name doesn't come up — or comes up vaguely, or is misdescribed, or appears alongside competitors who aren't really in your category — you are in the first conversation. ChatGPT is sending your customers somewhere else, and you don't even know it's happening.

This is not a future problem. It is happening right now, in every category, every day, in Sandton and Randburg and across Johannesburg. The question is not whether AI is recommending businesses in your category. The question is whether one of those businesses is you.

What to do about it

There are two paths.

The first is the path I'd recommend if you have the time and inclination — work through your own website and Google Business Profile, make sure every detail is accurate, complete, consistent across every platform, and described in plain language. Add the information AI is looking for: who you are, who you serve, what makes you different, what credentials you carry. Test it monthly by asking ChatGPT the same question and seeing if your name appears.

The second path is the one most owners take: have someone do the structural work for them. That's what BlueWing Local exists to do. We don't write your content or build your website, but we install the AI-visibility layer underneath what you already have, so that when ChatGPT is asked about a business like yours, you're the one it recommends.

Either way, the cost of doing nothing is the most expensive option. Every conversation like the first one above is a customer you didn't know you lost.

A free check, no obligation

Want to see what ChatGPT says about your business?

Send your website address to peter@bluewinglocal.co.za. I'll do the test personally — ask ChatGPT, Gemini, Claude and Perplexity to recommend a business in your category, and send you back exactly what they say. No sales pitch, just an honest answer.

Email peter@bluewinglocal.co.za →

— Dr Peter Thompson, PhD
Founder, BlueWing Local Marketing
Lone Hill, Sandton