How to read your row

Find your sector in the table on "The sector table". Each row is the share of companies in that sector, or the share of answers about them, that got a given result.

n is the number of companies in the sector. The smallest sector has seven. In a row that small, one company moves a figure by 14 points, so read it as a direction and not a precise number.

Named first means the company was the first one named in the answer to its own category question. We look at that, and at whether the company's own website was among the sources, rather than at whether it was mentioned at all, because in a small category almost everyone gets mentioned.

No company is named on these pages. The report shows your sector. Your own row is a separate check.

All 100 companies

12 October 22 September 4 September
Named by at least one engine 96% 93% 96%
Named by all four engines 68% 65% 62%
Named first by at least one engine 73% 71% 71%
Named first by all four engines 31% 32% 33%
Own website among the sources, on at least one engine 61% 62% 67%
Share of all sources that are the company's own website 6% 7% 8%
No schema markup found 31% 32%
No llms.txt file 84% 85%
Turns away at least one AI engine's bot 6 companies 6 companies

The last three rows come from the website checks, run on 4 September (with eight corrected addresses re-run on 5 October) and again on 12 October. The llms.txt row counts a file as there even when it is badly formed.

Most cited sources across the 100 when asked in English, by the number of companies they were cited for on 12 October: a German supplier directory, ensun.io (71 companies; 78 and 76 in September), English Wikipedia (54; 56 and 47), an American B2B data vendor's directory pages (41; 18 and 2), then a long tail of lead generation databases and ranking sites, none of which any of these companies chose.

The sector table

12 October, four engines. 22 September in brackets where it differs by more than five points.

Sector n Named by all four Named first by at least one Named first by all four Own website among sources Share of sources that are own websites
Energy, chemicals, pharma and materials 11 91% (82%) 100% (82%) 45% (36%) 73% (64%) 8%
Industrial machinery and equipment 21 90% 95% 62% (71%) 95% 9%
Electronics, telecom and measurement technology 16 81% 81% (75%) 38% (44%) 69% 7%
Forest, paper, packaging and wood 8 88% 62% (88%) 25% 38% (50%) 7%
Professional, trade and business services 20 50% 60% (50%) 20% 50% 5%
Software and IT services 17 18% (29%) 53% (59%) 6% 35% 2.4%
Construction and built environment 7 86% (43%) 43% 0% 43% (57%) 5%
All 100 100 68% 73% 31% 61% 6%

Sorted by how often the company is named first by at least one engine, highest first. In a sector of seven or eight, one company is 13 to 14 points.

On 12 October every sector's own-website share of sources sits between 2.4% and 9%. Nowhere do the companies themselves supply more than about one source in ten.

The technical setup by sector (schema markup and llms.txt) is on "The technical picture", with Finding 3. In short, the sectors with the most complete setup are not the ones AI cites most.

The sectors, one by one

Industrial machinery and equipment (21 companies). This is where Finnish companies hold their own categories. The company is named first in 81 to 83% of the answers about it, the highest of any sector, on every date. Part of that is the question: when you ask for the leading maker of a machine type "in Finland", the leading Finnish maker is often a global leader too, so this sector benefits from the wording more than any other. Even so, only about one source in ten is the company's own website.

Electronics, telecom and measurement technology (16). Named almost everywhere. Named first by all four engines for 19% of companies on 4 September, 44% on 22 September and 38% on 12 October. With 16 companies that is a handful of companies changing, so treat it as a sign of how much the answers move. It isn't a trend.

Energy, chemicals, pharma and materials (11). The most complete technical setup in the study: only one company in eleven had no schema markup. The company is first in about seven answers out of ten, and in October every one of the eleven was first on at least one engine. Its own website was a source for 82% of companies on the first date, 64% on the second and 73% on the third. When the company isn't first, the name in its place is sometimes the foreign owner of a Finnish operation or a global pharmaceutical company.

Forest, paper, packaging and wood (8). All eight are named by at least one engine and seven by all four. When a company isn't first, another Finnish forest company usually is (41%, 31% and 34% of answers on the three dates). The large names take each other's places.

Construction and built environment (7). The sector where a foreign-owned company most often takes the first place: in 54%, 39% and 50% of answers on the three dates, mostly a foreign-owned infrastructure builder and a Swedish engineering consultancy. The company itself is first in about a third of answers, among the lowest in the study. None of the seven has an llms.txt file. Seven companies and 28 answers per date, so read this as a direction.

Professional, trade and business services (20). The company is first in its own category in only about a third of answers on every date. A foreign or foreign-owned firm is first in 21%, then 31%, then 21% of answers: logistics, engineering, facility services, maintenance and pharmaceutical distribution, each with a different Nordic or German owner. Own websites supply about one source in twenty. The gap is filled by business data and lead generation sites, including an American B2B data vendor whose Finnish pages appeared in the answers within weeks and is now the second most cited source in this sector.

Software and IT services (17). The second most complete technical setup and the weakest result. Only 35% of companies had their own website among the sources, on each of the three dates, and own websites supply 2.4 to 4% of sources. When the company isn't first, another Finnish company usually is, about half of all answers, often a consultancy that isn't in the sample. The sources are almost entirely ranking, directory and data sites: Goodfirms, Clutch, the American data vendor, a Finnish sales software startup's ranking pages and an Indonesian higher education fair's list of "top software development companies in Finland". Named by all four engines: 18%, then 29%, then 18%.

Size and engine

By revenue quarter, 12 October (22 September, 4 September)

Revenue rank Named first by all four engines Own website among sources
1 to 25 56% (56%, 56%) 76% (68%, 84%)
26 to 50 20% (20%, 24%) 64% (60%, 64%)
51 to 75 28% (28%, 32%) 48% (60%, 64%)
76 to 100 20% (24%, 20%) 56% (60%, 56%)

The largest 25 are named first by every engine more than twice as often as everyone else, at exactly the same 56% on all three dates. Own-website citation moves around much more: the top quarter's lead was 20 points, then 8, then 12. Being big gets you named first. It doesn't reliably get your own website read.

By engine, all 100 companies, 12 October

Named Named first Own website among sources Different sites cited per answer, on average
Claude 81 49 24 7.7
ChatGPT 78 51 32 4.2
Gemini 93 55 31 5.0
Google AI Mode 77 52 46 4.4
Perplexity 89 51 44 10.7

Google AI Mode and Perplexity read the companies' own websites most often. Claude, least, on every date. Gemini names the most companies because its answers are long lists, around 13 companies an answer, which is why we compare engines on who they name first. About half of Gemini's answers came with no sources at all. Perplexity cites the most sites per answer and leans on LinkedIn company pages and financial data sites where the other engines lean on ranking lists.

4 Sept22 Sept12 OctGoogle AI Mode*504946Perplexitynot asked4444ChatGPT353432Gemini373531Claude292624
Companies whose own website was among the sources, per engine. Perplexity was added on 22 September. *Google AI Mode answered 95 of 100 questions on 12 October.

Where each sector's answers come from

Five most cited sources per sector, by the number of companies they were cited for (12 October, four engines).

Sector Most cited sources
Industrial machinery and equipment ensun.io, English Wikipedia, an environmental technology directory, a market research publisher, LinkedIn
Electronics, telecom and measurement technology ensun.io, English Wikipedia, then three of the sector's own company websites
Energy, chemicals, pharma and materials ensun.io, English Wikipedia, two lead generation databases, a contact data service
Forest, paper, packaging and wood an American B2B data vendor, English Wikipedia, two of the forest companies' own sites, ensun.io
Construction and built environment a Finnish sales software startup's ranking pages, a contact data service, English Wikipedia, the Finnish construction trade magazine, an American B2B data vendor
Professional, trade and business services ensun.io, an American B2B data vendor, a Finnish sales software startup's ranking pages, a contact data service, Goodfirms
Software and IT services ensun.io, an American B2B data vendor, Goodfirms, Clutch, a Finnish sales software startup's ranking pages

Asked in English, the German directory is the most cited source in five of the seven sectors, with Wikipedia close behind, and under them each sector has its own layer of lists and databases. The two newcomers from Finding 4 are now in the top five of four sectors. In construction, ranking pages a sales software startup published at the end of August had the top spot on 22 September and shared it on 12 October. Electronics and forest are the only sectors where the companies' own sites make the top five.

How to get your own company's row is on "Your own row".