A buyer in Nairobi needs a supplier. Two years ago they would have opened Google, skimmed five tabs and called two companies. Now they may ask an assistant to shortlist three. The difference matters: you might never see the visit that decided against you.

AI answers are built from sources. The model does not know your business by instinct. It reads what is publicly available, compares it with other sources and produces a summary. If the available information is thin, contradictory or out of date, the summary will be too.

Start with the facts, not the keywords

The first job is to make the basic facts unambiguous across every place your business appears: the website, Google Business Profile, LinkedIn, directories, social bios and any partner listings. The same legal name, description, services and locations should appear everywhere.

  • Write one clear sentence that says what the business does and who it serves.
  • Keep service names consistent. Do not call the same offer three different things.
  • Publish real addresses, hours and contact details in text, not only in images.
  • Make sure the About page names the people or the entity behind the brand.
  • Update old pages that still describe a service you no longer offer.

Write pages that can be quoted

Answer engines prefer content that answers a question directly. A page that begins with 400 words of brand history before explaining the service is harder to summarise than one that opens with a plain definition.

This does not mean writing for robots. It means leading with the answer, then adding the detail a human needs. A useful pattern is: what it is, who it is for, what it includes, what it costs or how pricing works, and what happens next. Add FAQs where the same questions keep coming up in sales calls.

Give the machine evidence

AI systems look for corroboration. A claim on your website is stronger when it appears alongside a named author, a date, a source or a third-party mention. That is also good for buyers. A page that says “we are the best” gives them nothing. A page that explains the method, the scope and the limitations gives them something to trust.

  • Use structured data for the organisation, services, locations and FAQs.
  • Credit the author of articles and keep bylines current.
  • Link to sources for statistics and technical claims.
  • Publish genuine case studies when clients allow it, with the scope and the numbers.
  • Keep content dated so old advice does not outlive its accuracy.

Track the answers, not just the rankings

Search Console will not tell you how often ChatGPT recommended you. You have to ask the systems the questions your buyers ask, in the same language they use, and record the answers. Do this monthly. Look for whether you appear, what is said about you, which competitors appear and which sources are being cited.

The goal is not to trick an AI into mentioning you. It is to make the business easy to understand and hard to misrepresent. That work also improves the website for humans, which is the part that still pays the bills.

If a system cannot explain what you do, a buyer will not trust it to recommend you.