Machine-Readiness Audit: Africa, July 2026

In Africa, the first job is still unfinished

The machine-readable web has always had two jobs: have a page at all, and make that page legible to a machine. In the United States the first job is essentially done, and our US edition could take the website as given and measure only the second. In Africa that assumption breaks. Here the first job, having a website at all, is unfinished for most businesses, and the story has to start there.

We measured it directly. Of 11,582 independent African businesses drawn from their Google Business Profile map listings, only 41.4 percent had a real website of their own. The rest were not missing from the internet; they were present on it in forms a machine cannot read as a business page: a messaging chat, a social profile, a listing on an aggregator. That is not a failure of effort. It is a different way of being online, and it is the defining fact of the African small-business web.

In Africa, being findable and being machine-readable have come apart.

How we measured this

We assembled the sample from the map listings for the same twelve trades as every edition of this benchmark, dentistry, auto repair, med-spa, small law, plumbing, HVAC, chiropractic, veterinary, salon, accounting, roofing, and physical therapy, across 17 African cities, keeping independent businesses and filtering out chains, directories, and aggregators. That produced 11,582 unique businesses. For each we recorded whether it had a real website of its own, as opposed to only a social or aggregator page, which gives the web-presence rate.

For the 4,793 businesses that had a website, we fetched the homepage and read, directly from its HTML, whether it carried JSON-LD structured data and which types, whether any was in the LocalBusiness family, and a set of on-page signals: title, meta description, H1, mobile viewport, HTTPS, Open Graph tags, and a phone number. 3,696 of those sites were reachable. We summarized each with a Machine-Readiness Score from 0 to 100. Holding the same twelve trades and the same read across every region is what makes the editions comparable. Every figure is a value a tool returned, with Wilson confidence intervals on the headline rates.

The first gap: only 41.4% of businesses have a website

The web-presence rate is the headline, because it is the first filter and the widest. Just 41.4 percent of map-listed African businesses had a website of their own. Africa is the least-connected region measured, and its internet is sharply mobile-first and urban. The ITU puts internet use across Africa at about 38 percent in 2024, the lowest of any world region against a 68 percent global average, and use is far higher in cities than in the countryside. In Sub-Saharan Africa, mobile-internet penetration was around 27 percent at the end of 2023, with a large share of people who live under network coverage still not using mobile internet, mainly because of device cost.

Against that backdrop, a business that has any web-listed presence at all is already among the more connected. Many reach customers through mobile money, WhatsApp, and social pages rather than a website, which is a rational response to the region's connectivity and device economics. Until the website exists, most African businesses are invisible to the machines that increasingly decide who gets named in an answer.

Among the sites, the machine layer is thin

Having a website is necessary but not sufficient. Among the 3,696 reachable African sites, 52.7 percent carried no structured data at all (95 percent confidence interval 51.0 to 54.3 percent), and 85.4 percent had no LocalBusiness-family type. Sorted into tiers, 52.7 percent were machine-invisible and only 10.8 percent fully readable, with a median Machine-Readiness Score of 57.0 out of 100.

The one consistent bright spot is the human-facing basics: 94.8 percent mobile-ready, 99.6 percent secure, 58.1 percent with a meta description. It is the machine-facing finish, the cheap part, that is missing most, and the gaps cluster: among sites with no structured data, only 36.4 percent carried a meta description, against 82.3 percent of sites that did.

The compound gap: about 4.5% are fully machine-readable

The two gaps multiply. Start with 100 map-listed African businesses. About 41 have a real website. Of those, 10.8 percent are fully readable. Put together, only about 4.5 of the original 100, roughly one in 22, are fully machine-readable: present, crawlable, and marked up as a local entity. In our US edition the comparable figure is about 32.1 percent. That gap is the single most important number in this study.

It is worth being careful about what this means. Being machine-readable is a precondition, not a promise. Google is explicit that structured data enables richer search features and helps an engine understand a page, but does not by itself lift ranking. A business a machine cannot find or parse cannot be named by one, and in Africa that describes about 96 of every 100 businesses a customer sees on the map.

The gap by trade

Machine-readiness differed by trade. Ranking the twelve by LocalBusiness-type use, Accounting was lowest at 9.9 percent and Chiropractic highest at 20.1 percent. A chi-square test of trade against local-entity presence returns 38.2 on 11 degrees of freedom, past the 0.001 threshold. The credence-service trades, whose buyers most need to verify before committing, again tended to sit low, echoing the pattern from the US and India editions.

Holding the same twelve trades across every region is what keeps the comparison consistent, even where a trade is sparse in a given market.

The investment effect

As in every edition, the more established a business, the more machine-readable it tended to be. LocalBusiness-type use rose monotonically with review volume, from 13.5 percent for businesses with ten or fewer reviews to 23.3 percent for those with more than five hundred, though the correlation with the readiness score was weak (0.09). Star rating tracked the same way, from 8.1 percent for sites rated under 4.0 to 17.8 percent for the highest rated.

Reachability among sited businesses was 22.9 percent unreachable, and rose with how established a business was: 27.6 percent of the quietest businesses' sites responded, against 66.6 percent of the busiest. Dead and parked domains, more common among newer businesses, widen the effective web-presence gap: a site that does not load is, to a machine, the same as no site.

How this compares

Set against the published record, the Africa numbers are coherent. Web-wide, HTTP Archive's 2024 Web Almanac finds no JSON-LD on about 59 percent of pages and the LocalBusiness type on under 4 percent; the African sites we read, at 52.7 percent with no structured data, sit near the global page-level baseline rather than above it, unlike US small businesses, which were well ahead of it. And the web-presence figure lines up with the African connectivity data, which is the strongest sign a single study can give that its numbers are real.

The sharpest lens is the comparison to the other editions of this benchmark. African businesses come out at about 4.5 percent fully machine-readable, against roughly 32.1 percent in the US. This is the first of several regional editions; the cross-region comparison, once every region is measured, sets these differences side by side with the care they need.

What it means

The Africa picture points to one specific, fixable thing: whether the machines now mediating discovery can read these businesses. Today, for most, the answer is no, and the reason has a particular shape in Africa.

In Africa today, most businesses forgo machine-readability not out of neglect but because a mobile-first, messaging-first way of operating works well enough. The bet this study makes, and will test each quarter, is that as more discovery runs through engines that assemble answers, the businesses a machine can read will pull ahead of the ones it cannot.

None of this is a verdict on African businesses, which are digitally active in their own way and growing more so. It is a measurement of eligibility: the ability to be read, placed, featured, and cited at all. Structured data does not buy ranking. It buys the chance to be found by a machine, which in Africa most businesses do not yet have.

A benchmark we intend to keep

This is the first Africa edition of a measurement we repeat every quarter, part of a multi-region program running the same method across the United States, India, Africa, and other economies. A single snapshot says where Africa stands; a series says which way it is moving.

Every wave holds the method fixed: the same twelve trades, a fresh draw of independent businesses from map listings across the same kind of city spread, the same homepage read, the same Machine-Readiness Score, the same tiering rules. Each wave is a fresh sample, not the same sites re-checked, so the series measures the population moving. Once several regions and several quarters exist, a cross-region evolution study will set them side by side.

Limits, stated plainly

Every number here is measured, and every number has bounds. The sample is drawn from businesses with an active Google Business Profile on the map, which is the more digital tier of African small business; firms with no map profile are absent and almost certainly less machine-readable, so these figures set a floor on the gap, not a ceiling. Web-presence is measured as a listed own-domain versus a social or aggregator link; a business reachable only through messaging is counted as having no website, which is correct for machine-readability but understates its commercial presence.

We read the homepage only, so structured data on inner pages is undercounted, and detection parses JSON-LD, the dominant format, so the few sites using microdata are missed. Reachability among sited businesses was 22.9 percent. All measurements are a single snapshot from late July 2026, and relationships reported are associations, not proven causes. None of these limits changes the central finding, which is large and consistent: in Africa, most independent small businesses are not yet readable by the machines that increasingly decide who gets found.