A customer wants a new POS provider for their Lagos retail chain. Two years ago they’d have Googled “best POS system Nigeria,” scrolled a page of blue links, and clicked three or four. Today they open ChatGPT and type the same thing — and get back a short paragraph naming two or three providers, with a confident recommendation. They never see a results page. They never visit a homepage. The decision narrows before a single click happens.

That shift is not coming. It has happened. Around 93% of AI search sessions now end without a website click, and when Google shows an AI Overview, users click a traditional result only about 8% of the time. The click you used to compete for is increasingly the click that never happens.

For most Nigerian businesses this is still an abstract worry — something for “big tech companies” to think about. That’s exactly the mistake. The window to build visibility inside AI answers is open right now, it’s organic, and it strongly resembles SEO in 2004: a discovery channel with no ad inventory yet, where the brands that move early own the answer before competitors can buy their way in. That window will close. Let’s talk about how to use it while it’s open.

The thing nobody tells you: “AI search” is not one thing

The single most expensive misunderstanding we see is treating AI search as a monolith — as though “getting into ChatGPT” and “getting into Google’s AI” and “getting into Perplexity” are the same job. They are not. They are closer to three different platforms with three different rulebooks.

The evidence here is overwhelming. An analysis of 680 million citations across the major AI engines found that only 11% of domains are cited by both ChatGPT and Perplexity. Google’s own AI Overviews and AI Mode cite the same URLs just 13.7% of the time — despite reaching essentially the same conclusions. A brand that dominates one engine can be completely invisible in another.

Why the divergence? Because each engine sources information differently:

Ranking #1 on Google no longer means you’re in the answer

Here’s the finding that unsettles most business owners we show it to. In mid-2025, roughly 76% of Google AI Overview citations came from the top-10 organic results — so ranking well more or less carried you into the AI answer. By early 2026, that figure had collapsed to somewhere between 17% and 38%, depending on whose data you read.

Translated: a page can rank in position 3 for a keyword and never once be cited in the AI Overview above it — because it lacks the structure the AI needs to extract an answer. Meanwhile, research from Ahrefs found that around 80% of URLs cited across ChatGPT, Perplexity, and AI Mode don’t rank in Google’s top 100 at all for the original query. The old map and the new territory have come apart.

This is not a reason to abandon SEO. Strong Google and Bing rankings remain the foundation that AI engines’ “grounding” step needs to even find you. It’s a reason to understand that ranking is now necessary but no longer sufficient. You keep the SEO foundation — technical health, crawlability, Core Web Vitals — and you layer AI-citation readiness on top.

What actually earns a citation

The good news is that the tactics are known, tested, and — unlike a lot of marketing advice — backed by controlled research rather than gut feeling. The Princeton team that originated the “Generative Engine Optimization” concept demonstrated that adding statistics and citations to content can improve AI citation rates by up to 40%. Independent 2026 studies put content with statistics, named sources, and quotations at 30–40% higher visibility in AI responses. Here is what that looks like in practice.

1. Answer the question first, in the first 50 words

AI engines extract answers, they don’t read essays. Lead every important page with a direct, declarative answer block — roughly 50 words that state the answer plainly — before you elaborate. A clever headline that buries the answer gets skipped. “The best time to send marketing emails in Nigeria is mid-morning on weekdays” gets extracted; “Timing: the eternal question” does not.

2. Structure for extraction

Use question-based H2 and H3 headings that mirror how buyers actually ask. Break content into self-contained sections a machine can lift whole. Add FAQ blocks. Use tables and ordered lists — Google’s AI eagerly pulls structured formats into answers. Every section should stand alone well enough to be quoted without the rest of the page.

3. Put verifiable facts and named sources on the page

Statistics with sources, proprietary data, named experts with credentials. This is the single highest-leverage change the Princeton research identified. AI engines are building what researchers call “a web of mutual verification” — they trust content that itself cites trustworthy sources.

4. Keep it fresh, and date it

Especially for Perplexity, which cited content published within the last 30 days at an 82% rate in one 2026 analysis. Pages updated within 60 days are meaningfully more likely to be cited across engines. Visible year signals — a real “2026” in a title, not a decorative one — help. A genuinely updated post beats a six-month-old one on the same topic almost every time.

5. Build presence where the engines already look

This is the one most businesses miss. Brands are currently 6.5 times more likely to be cited through third-party sources — review sites, forums, news, communities — than through their own domain. Reddit, YouTube, and industry publications are not “nice to have”; they are the raw material the engines assemble answers from. For a Nigerian business, that means credible presence in the places your category is actually discussed, plus the review and directory profiles that AI systems treat as corroboration.

You cannot improve what you refuse to look at

Here’s a two-minute exercise worth doing before you finish reading. Open ChatGPT and Perplexity, and ask each the question your best customer would ask: “What’s the best [your category] in [your city]?” Then ask, “Compare [your brand] vs [your main competitor].”

Note what comes back. Are you named? Is the description accurate? Is your competitor there and you’re not? Most business owners have never once done this — which means they’ve never seen the shelf their buyers are now shopping from.

What you can do is build the trajectory deliberately: fix the measurement so you can see where you stand, restructure your highest-value pages for extraction, build the third-party presence the engines cite, and track how it moves — for you and your competitors — month over month. That’s a programme, not a one-off. But it’s a programme with a rare property in 2026 marketing: most of your competitors aren’t running it yet.

Figures in this article are drawn from published 2026 analyses of AI search behaviour, including large-scale citation studies and platform research current as of mid-2026. AI search is a fast-moving field; specific percentages shift as engines update. The direction of travel — discovery moving into AI answers — is consistent across every credible source.