Search

What is AEO, and why it matters for Kenyan businesses

A growing share of buying decisions now starts with a question typed into ChatGPT, Perplexity or Gemini rather than a search box. The assistant reads a handful of pages and composes one answer. If your business is not in that answer, you were not in the running — and you will never see it in your analytics.

17 September 2026 · 6 min read · TwinFusion

The difference in one sentence

SEO is competing for a position in a list of links. AEO — answer engine optimisation — is competing to be quoted inside a single generated answer. The first gives the customer ten options. The second usually gives them two or three, and no scroll.

That compression is the whole point. Ranking eighth on Google still earns some traffic. Being the eighth-best source for an AI assistant earns nothing, because the answer stopped at three.

What assistants actually do when they answer

An assistant answering "who builds e-commerce sites in Nairobi" does roughly this: it interprets the question, fetches candidate pages, extracts facts it can stand behind, and writes an answer citing a few sources. Each of those steps has a failure mode you control.

  • It fetches your page without running JavaScript. If your content only appears after a script executes, the assistant sees an almost empty page.
  • It looks for unambiguous facts. "We are a leading provider of solutions" tells it nothing. "An independent software agency in Nairobi, Kenya, founded in 2019, working across Kenya, Uganda and Tanzania" tells it everything.
  • It prefers content shaped like an answer. A page that poses a question and answers it directly is far easier to quote than a brochure paragraph.
  • It checks whether the claim is corroborated. A business that appears consistently across its own site, its structured data and third-party listings is safer to cite than one that appears once.

Why this is unusually winnable in Kenya right now

Most Kenyan business sites fail the first test outright. They are built on page builders that render client-side, carry no structured data, and describe themselves in marketing language that contains no extractable facts. An assistant asked a local question often has thin material to work with.

That is an opening. The work required is unglamorous and mostly technical, which is precisely why few competitors have done it.

What to actually do

  1. Make sure your content is in the HTML the server sends, not assembled by JavaScript afterwards. Test by viewing source, not by looking at the page.
  2. Add structured data: Organization, LocalBusiness, and Service markup that states who you are, where you are and what you sell in a format a machine reads without interpretation.
  3. Write pages that answer real questions in the first two sentences. Put the answer first and the context after — the opposite of how most marketing copy is written.
  4. State concrete, checkable facts: cities served, languages spoken, response times, what you do not do.
  5. Give each service its own page and its own address. One page describing five services can only ever be cited as one generic source.
  6. Publish an llms.txt describing your business and your key pages, and allow the AI crawlers in robots.txt deliberately rather than by accident.

How you will know it is working

You will not see it in Google Analytics, because an assistant reading your page is not a session. Check it directly: ask the assistants the questions your customers ask, in the words they would use, and see who gets named. Repeat monthly. That is the scoreboard.

It is a slower signal than a rankings report, and a more honest one.

Want this checked on your own site?

We will look at it and write back within one business day — speed, search visibility, and what the AI assistants currently say about you.

Start a conversation →
← All articles