State of AI Visibility in Finnish B2B

Executive summary

AI builds its picture of Finland's largest B2B companies mostly from sources they didn't write. About one source in fifteen was the company's own website: 6% on 12 October, 7% and 8% on the September dates. We asked the same questions in Finnish on 12 and 15 October, and the company's own site was an even smaller share both times. The rest is directories, data vendors, ranking lists, Wikipedia and the press.

Who does the describing depends on the language you ask in. In English the most cited source is ensun.io, a German directory built automatically by category and country. It was among the sources for 71 of the 100 companies on 12 October, more than the companies' own websites (61). Counted on the four engines that answered in both languages, it was a source for 77 companies in English but only 22 and 25 in Finnish, and Finnish Wikipedia, press release services, trade associations and business databases took its place. The companies' own websites didn't gain a bigger share.

Two thirds of the companies have schema markup, the code that tells search engines and AI what a page is about. It didn't decide whether AI used their website, and most of it says almost nothing about the company. Companies with it and without it had their own site cited at about the same rate. Only 18 of the 100 use their markup to state any fact about themselves, such as an address or a founding year, so what this compares is mostly empty markup against none.

The basic facts are mostly right. The errors that remain are last year's facts. We also asked each engine for five basic facts about each company and checked every answer against the company's own sources. On two dates, 4.8% and 5.4% of the answers had a wrong fact, most often a previous CEO named as the current one, and somewhat more often about the smaller companies. Schema markup as a whole made no clear difference to this either. The 18 companies whose markup states facts about themselves had fewer wrong facts, a lead we didn't plan to test (see "Where the errors are").

What this means. When a buyer, an analyst or a future employee asks AI about a Finnish B2B company, the answer is mostly put together from sources the company didn't write and doesn't watch. Some of those sources are being built right now, by companies with something to sell, and AI started citing them within days of publication. Inside most companies nobody owns this. Marketing owns the website, communications owns the press, investor relations owns the reports, and the layer AI actually reads falls between them.

We asked the same questions on three dates. Most of the overall figures held within five percentage points from one date to the next, while the answers about individual companies changed a lot. The one that moved was the German directory, cited for seven fewer companies in October than in September, though still for more of them than their own websites in English. So the picture is stable. Any single company's position isn't.

Every company in the study is listed, and listed companies have to publish a great deal about themselves. To see how much that matters, we asked the same questions about 29 private companies of similar size. For the 19 of them that fit a category we had already asked about, AI named them as often as the listed companies in the same answers, but almost never used their own websites as a source (2 of 144 answers, against 19 for the listed companies).

Find your sector on "The sector table".