Strong brands, invisible in AI answers
Deep Ocean Research tracks whether brands that already win on Google and in their market also get named when AI engines answer the same buyer questions. US brands lead the series, with genuinely strong Swiss brands tracked second, never by default, only when the strength is real. Below that is the earlier methodology archive: anonymous, aggregate scans of independent Swiss businesses that proved the audit engine before the series moved to named strong brands.
The citation index
The current spearhead of this series: a public, cumulative record of which named brands ChatGPT names or cites for their category's buyer questions. US brands lead the panel; a Swiss brand is added only when its strength is real, never by default. Runs on a weekly browser-sampling protocol; every entry is dated, and missed runs stay visible in the record. Failure included.
The citation gap on real brands
Point-in-time samples of strong brands measured against their own category's AI answers. Names withheld and metrics banded; the per-answer record, the mechanism, and the limits are published in full.
Earlier reports
Before the series moved to named strong brands, these two reports proved the audit engine on anonymous, aggregate scans of independent Swiss businesses.
Archived 2026-07-11: reports 001 and 003 are the readability series from an earlier version of the method, kept online as the record. Report 002 (Milan) has been removed.
Can AI engines read Switzerland's law firms and fiduciaries?
30 independent Swiss law firms, notaries, and fiduciaries scanned. Several block an AI crawler outright, and most publish nothing structured for the machines to read.
Can AI engines read Switzerland's private clinics?
24 Swiss private clinics scanned. Nearly 1 in 3 block an AI crawler, and most publish nothing structured for the machines to read.
About this series
AI visibility here means whether an AI engine names a brand, by name, in the answer to a real buyer question. That is what Deep Ocean Research measures, category by category. Where a brand ranks in Google feeds which sources an engine can retrieve, but ranking alone does not decide the citation, so we measure the citation itself. The public panel is browser-sampled from ChatGPT, the same meter we run on our own site; crawler access checks use our audit engine, Hadal. Client Sprints sample both ChatGPT and Perplexity, manually and dated.
The earlier archive of anonymous Swiss reports withheld business names by design, publishing only aggregate counts. The current citation index does the opposite on purpose: named brands, named sources, a published protocol anyone can audit. That is the difference between proving a method and proving an outcome.
Want to know how your brand compares?
We test one of your buyers' questions in a live AI engine, free, and email you which brands get named and whether yours does.
Free · one question, one answer, in writing · delivered within a week · no sales call required