Kenya SEO Benchmark 2026: we checked the 515 most-visited .ke websites
Most SEO advice written for Kenya quotes global numbers. We wanted Kenyan ones. So we took every .ke domain in the Tranco top-1 million list of the world’s most-visited websites (515 domains) and checked each homepage for the SEO basics: titles, descriptions, headings, sitemaps, structured data, and how each site treats AI crawlers.
- 1 in 3 homepages has no meta description. On government sites it’s 78%.
- Only 44% have exactly one H1 heading.
- 18% send almost no text in their HTML. They depend on JavaScript, which most AI crawlers don’t run.
- 12% block AI bots such as GPTBot and ClaudeBot. 35 of those 49 sites do it through Cloudflare’s managed robots.txt, not a rule they wrote themselves.
- 40% have no XML sitemap we could find. On government sites it’s 72%.
- HTTPS is nearly universal (99.5%). Kenya has got the security basics right.
1. Meta descriptions: missing on a third of homepages
A third of top Kenyan homepages leave Google to guess their search snippet.
Among homepages that send real text in their HTML, 33.5% have no meta description. Google writes its own snippet when one is missing, and that snippet is often a cookie notice or a menu.
| Group | No meta description |
|---|---|
| Business (.co.ke) | 16.9% |
| Universities and colleges (.ac.ke) | 50.0% |
| Government (.go.ke) | 78.0% |
2. Headings: fewer than half get the H1 right
Fewer than half use one clear H1. Universities and government sites do worst.
44.2% of homepages have exactly one H1, 39.6% have none and 16.2% have several. Universities and government sites do worst: about a quarter have exactly one.
3. 18% of top Kenyan homepages are nearly empty HTML
73 homepages look almost blank to AI crawlers, because JavaScript draws the content.
73 of the 401 homepages sent fewer than 100 words in their HTML. The content is drawn by JavaScript after the page loads. Google can usually render it, though later and less reliably. Most AI crawlers (ChatGPT, Claude, Perplexity) read the raw HTML only, so to them these sites look blank.
4. The AI-bot surprise: Cloudflare is choosing for you
Most AI-bot blocks come from a Cloudflare setting, not from the site owner.
49 sites (12.2%) block at least one AI crawler in robots.txt. GPTBot, ClaudeBot, CCBot, Google-Extended and Bytespider are blocked most often. But 35 of those 49 blocks come from Cloudflare’s managed robots.txt, a setting Cloudflare can switch on for you. Many site owners probably don’t know their site is opted out of AI training.
Blocking AI training bots doesn’t remove you from Google Search. It’s a business choice. Our point is that it should be your choice. Only a handful of sites block the AI search bots (ChatGPT-User 8, PerplexityBot 4, OAI-SearchBot 2). Those are the bots that fetch pages to answer people’s questions.
We looked closer at which bots Kenyan sites block, and why, in a follow-up study: Are Kenyan websites blocking ChatGPT?
yourdomain.co.ke/robots.txt. If you see “Content-Signal” or a Cloudflare-managed block, review it in Cloudflare under Security → Settings → AI bot policies.5. llms.txt has arrived in Kenya
One in seven top sites now has an llms.txt file. No government site does.
56 sites (14.0%) publish an llms.txt file, a short guide to the site written for AI tools. That’s 20.2% of business sites and none of the government sites. 23 of the 56 were generated by an SEO plugin: Yoast (10), Rank Math (9) and All in One SEO (4). Nobody has shown yet that llms.txt changes AI answers, but it costs nothing to switch on.
6. Sitemaps, robots.txt and broken “page not found” handling
Government sites trail on sitemaps, robots.txt and structured data.
| Check | All | Business | Universities | Government |
|---|---|---|---|---|
| XML sitemap found | 59.9% | 64.1% | 66.3% | 28.3% |
| robots.txt present | 80.5% | 84.3% | 83.7% | 54.3% |
| Missing pages return “200 OK” (soft 404) | 16.5% | 15.7% | 11.6% | 17.4% |
| Structured data (any) | 49.6% | 60.1% | 40.7% | 19.6% |
| Images without alt text | 35.2% | 29.9% | 49.5% | 48.5% |
A soft 404 means a page that doesn’t exist still answers “OK”. Google then spends crawl time on junk URLs and may index empty pages.
7. What Kenya’s top sites are built with
WordPress runs 37% of top .ke homepages, and about half of those show no SEO plugin.
37% of the homepages run on WordPress, rising to 55% of university and college sites. Of the 150 WordPress sites, 77 show no sign of an SEO plugin. Yoast leads (46), followed by Rank Math (18) and All in One SEO (8). Local touches are less common than you might expect: 24% of homepages link to WhatsApp and 19% mention M-Pesa.
Methodology
Methodology
Sample: all 515 domains ending in .ke in the Tranco top-1M list (list ID PY2NJ, downloaded 12 Nov 2026). Tranco ranks domains by popularity across several traffic sources, so this is a sample of the most-visited Kenyan domains, not of all Kenyan businesses. It includes media, betting, government, universities and companies.
What we checked: each homepage as sent by the server (no JavaScript run), plus /robots.txt, /sitemap.xml (and common WordPress variants), /llms.txt and a made-up URL to test 404 handling. Checked on 12 November 2026 with a standard desktop browser user agent.
Excluded: 76 sites could not be checked: 27 refused our request, 42 showed a bot-protection page (“One moment, please…”) and 7 showed a bare server placeholder. Another 38 were unreachable or returned errors. That leaves 401 homepages. Heading, description and alt-text figures use the 328 homepages that sent at least 100 words of HTML. JavaScript-only pages would otherwise count as “missing” everything.
Limits: homepage only, one snapshot in time. We publish percentages for businesses, not a list of their weaknesses. If you run a .ke site and want your own results, ask us.
Corrections and right of reply: if you think a figure is wrong, or you have fixed something since 12 November 2026, tell us. We check every report and correct this page with a dated note.
