The Machine-Readiness Atlas: How the World's Small Businesses Compare

One method, six economies

Over July 2026 we ran the same measurement six times, in the United States, Europe, India, Japan, Southeast Asia, and Africa. Each edition sampled independent small businesses from their Google Business Profile map listings across the same twelve trades, recorded whether each had a real website, and for those that did, read the homepage for structured data and on-page signals. Together the six editions cover 205,103 businesses. This is the atlas that sets them side by side.

The point of holding the method fixed is comparability. A single-country study tells you where that country stands; six of them, measured identically, tell you how the small-business web differs across the world, and where the machine-readable web is thickest and thinnest. Each economy also has its own detailed edition; this study is the overview.

Measured the same way, the small-business web varies about 7.6-fold in how much of it a machine can read.

The headline: a 7.6-fold range

The single number that captures each economy is the share of all map-listed businesses that are fully machine-readable: present with a real website, crawlable, and marked up as a local entity. It ranges from 32.1 percent in the US, down through 12.7 percent in Europe and 7.3 percent in Japan, to around 4.5, 4.6, and 4.2 percent in Africa, Southeast Asia, and India. The US lead is not small: it is roughly 7.6 times the level of the lowest economies.

That range is wider than most cross-country digital metrics, and it matters because machine-readability is becoming a precondition for being found. The economies at the bottom are not less commercially active; several are among the most digitally engaged on earth. They have simply built their small-business presence in forms a machine cannot read.

Two worlds: whether the website exists at all

The first divide is the starkest: whether a business has a website at all. In the mature-web economies the answer is usually yes, with web-presence around 97.0 percent in the US, 76.3 percent in Europe, and 67.5 percent in Japan. In the mobile-first economies it is usually no: about 33.3 percent in India, 30.3 percent in Southeast Asia, and 41.4 percent in Africa.

The reason is not neglect but a different way of being online. In India, Southeast Asia, and Africa, small-business commerce runs on messaging and social, WhatsApp, Instagram, LINE, TikTok Shop, mobile money, which are reachable to people but invisible to a crawler. These economies leapt to mobile and messaging without ever passing through the website era at scale. That is efficient for human commerce today, and a growing liability as discovery moves to machines that read pages, not chats.

The US outlier, and the markup gap everywhere else

Having a website is only the first filter. The second is whether that site tells a machine what it is, and here the picture is bleak almost everywhere, with one exception. Among businesses that have a site, the share using a LocalBusiness-family schema type is 44.7 percent in the US but only 22.1 percent in Europe, 19.5 percent in Southeast Asia, 17.7 percent in India, 14.6 percent in Africa, and 13.1 percent in Japan. Web-wide, the rate is under 4 percent.

The US stands alone, at roughly twice the next economy and more than three times Japan. The most likely explanation is industry, not technology: the US has a large, mature local-SEO and local-marketing sector that has spent a decade adding schema to small-business sites. Everywhere else, that industry is younger or thinner, and the markup simply has not been done. Japan is the sharpest case of the reverse pattern, a wealthy, deeply digital economy where 67.5 percent of businesses have a site but only 13.1 percent mark it up, sites built but not made readable.

Two failure modes

Put the two filters together and a clean picture emerges: the world's small-business web fails to be machine-readable in two distinct ways. In the mobile-first economies the gap is the website itself, most businesses never build one. In the mature-web economies the gap is the markup, most businesses build a site but never make it legible to a machine. Both end in the same place, a business a machine cannot fully read, but the fix is different: a first website in one case, a few lines of structured data in the other.

The table below sets every economy against both filters. Reading down the compound column is reading the machine-readable small-business web of the world, best to worst, and no economy clears even a third except the United States.

How Asian MSMEs are organized differently

There is a subtlety in these numbers that matters most for Asia, and it cuts against the audit itself. The twelve trades held fixed across every economy, dentistry, small law, accounting, med-spa, the home services, were chosen because they exist everywhere and so can be compared. But across Asia they are not the businesses that make up the market. The Asian high street is built from kirana grocers, tailors, and mobile-repair kiosks in India, warungs, sari-sari stores, and mamak stalls in Southeast Asia, and izakaya, ramen shops, greengrocers, and sento baths in Japan, and those trades behave very differently online.

We measured it directly. Re-running the audit across the trades that actually dominate the Indian economy, twenty-two of them, web-presence falls from 33.3 percent on the standard trades to 18.4 percent, and the fully-machine-readable share falls from 4.2 percent to just 1.2 percent. Southeast Asia shows the same move, from 30.3 to 11.4 percent web-presence and 4.6 to 0.9 percent fully readable. Most striking, wealthy Japan shows it too: across its traditional high street, izakaya, ramen shops, greengrocers, sento baths, own-website presence falls from 67.5 to 43.4 percent and the machine-readable share from 7.3 to 2.3 percent. Across all three the standard trade mix overstates machine-readiness by roughly 3.9 times.

The reason is structural. The trades a standard audit samples, the professional services a customer researches carefully before booking, are exactly the ones most likely to keep a website. The trades that make up the bulk of an Asian economy, everyday retail and informal services bought on the spot or through a neighbour's recommendation, are the ones least likely to. A tailor or a kirana grocer runs on a phone number and word of mouth; a Tokyo izakaya or ramen counter runs on Tabelog, the review aggregator, rather than a page it controls. Different door, same outcome: not a crawlable site the business owns. So the reading of the atlas is doubled: on the comparable professional trades India, Southeast Asia, and even Japan already sit well below the United States, and measured across their real trade mix the machine-readable share is a few percent at most. The India, Southeast Asia, and Japan editions of this benchmark lead with that truer, native picture; the atlas keeps the comparable trades so the economies can still be lined up.

What stays the same across the world

For all the variation, three patterns held in every economy we measured. Machine-readiness rose with how established a business was: the more reviews, the more likely a site carried local-entity markup, everywhere. It varied by trade in the same direction: the credence-service trades, accounting, law, therapy, whose buyers most need to verify before committing, tended to invest least in machine-readable signals, in every region. And the human-facing basics, mobile-friendliness and security, were nearly universal everywhere, while the machine-facing finish was the part left undone.

Those constants matter as much as the differences. They suggest the gap is not cultural but structural: a cheap, low-visibility task that businesses skip until something forces the issue, in rich economies and poor ones alike. What differs is only how far along each economy is, and the United States is further along mostly because an industry there made it its business to be.

What it means

The atlas points to a single, actionable reading. Machine-readability is becoming a precondition for being found, it is cheap to achieve, and almost everywhere it is mostly absent. That combination, high and rising importance, low cost, low prevalence, is exactly the profile of a large and durable advantage for the businesses and markets that move first.

It is worth stating the limit of the claim, because it is easy to oversell. Structured data does not buy ranking; Google is explicit that it enables features and understanding, not position. What it buys is eligibility, the ability to be read, placed, and cited by a machine at all. This atlas measures how many businesses in each economy have that eligibility today. In most of the world, most do not.

And the comparison itself must be read with care. Differences across economies reflect machine-readiness, web-versus-social norms, language, and how completely Google's map covers each market. The atlas is a first-order map, not a final verdict, and every figure in it is a floor, because every sample is drawn from the businesses already visible enough to keep a map profile.

A living atlas

This is the first edition. We repeat the measurement each quarter, in every economy, holding the method fixed, so the atlas becomes a time-series showing where the small-business web stands and which economies are closing the gap fastest as AI-driven discovery spreads. Each region has its own detailed edition, linked below, and future waves will add regions and quarters.

The prediction the atlas will test is simple. If discovery keeps shifting toward engines that assemble answers, the economies and businesses that become machine-readable first should pull ahead, and the gap this first edition found, wide, cheap to close, and mostly unaddressed, should start to move.

Limits, stated plainly

Every figure here is measured, and every figure is a floor. All six samples are drawn from businesses with an active Google Business Profile on the map, the more-digital tier in every economy; firms with no map profile are absent and almost certainly less machine-readable, and map coverage itself varies by country, most in the mobile-first economies, which means the true gaps are wider than shown and the cross-region comparison is directional. The US web-presence figure is measured differently from the others, from near-universal website adoption in the probe and external data rather than the same map draw, so it is an approximation; the US markup and readiness figures are directly measured.

We read the homepage only, so structured data on inner pages is undercounted, and detection parses JSON-LD, the dominant format. Some trades are sparse in some economies. All measurements are a single snapshot from late July 2026, and every relationship reported is an association, not a proven cause. None of these limits changes the central finding: measured the same way, the small-business web varies enormously in how much of it a machine can read, and in every economy the machine-readable share is a minority.