What an AI Says When You Search for an Accounting Firm in Geneva

What an AI says when you search for an accounting firm in Geneva, in one sentence

Asked about an accounting firm in Geneva, a generative AI such as ChatGPT builds its answer from recent reviews, a complete Google Business Profile, mentions found on third-party sites in French and, increasingly, well-populated LocalBusiness Schema.org markup. It does not read the same signals as a Google Ads ranking or a classic organic ranking. This is a selection based on a firm’s aggregated digital reputation, not on its advertising budget.

That difference in mechanics explains a pattern many Swiss business owners discover the hard way: a firm can sit on page one of Google and stay completely absent from a ChatGPT answer on the same topic, or the reverse. This mechanism is not fixed: models get retrained and periodically re-evaluate their sources, which means a position observed today is not guaranteed to hold next year.

Why this question matters now for a Geneva firm

ChatGPT passed one billion weekly active users in 2026, up from roughly 400 million about a year and a half earlier. A growing share of those queries concern local and professional questions, the kind a business owner asks before choosing an accounting firm. A related signal confirms the trend: the share of consumers who always read reviews before picking a local provider has climbed to roughly four in ten.

For a Geneva firm, this changes the nature of the problem. Ranking well on a classic Google query is no longer enough. It also matters what an AI answers when a prospective client asks it directly, and why that answer is not the same list of names.

There is also a difference in how the question gets asked. On Google, a business owner types three keywords and compares blue links. Faced with an AI, they phrase an actual sentence, with context: the size of their company, a language constraint, sometimes a budget. That richer phrasing changes which criteria matter: a page structured to answer that kind of sentence directly carries more weight than a simple stack of keywords in a title.

What comparable-market studies show

This has already been studied seriously, even though no public study focuses specifically on the Geneva or Swiss romande market. A study run across three engines with web search enabled (ChatGPT, Gemini, Claude), six cities of different sizes, several professional service categories and four query types per category measured a clear pattern for accountants: in London, all three engines cited only the four largest firms in the sector, while classic Google search surfaced independent practices far more often. No match between the two rankings was found in that city.

In a smaller city, the same protocol measured close to the opposite: independent practices scored the maximum opportunity rating, simply because no dominant global brand had enough presence in the underlying data for that market. Market size does not just change the intensity of the effect, it changes the rule itself.

The particular divide affecting accounting and fiduciary firms

Among the three professional categories tested in these studies (lawyers, dentists, accountants), accounting showed the widest gap between what an AI recommends and what a classic search ranking surfaces. Property law showed the opposite: the strongest alignment, where firms ranking well on Google were also the ones AI engines cited.

The most plausible explanation is press coverage and external mentions. Large international accounting networks carry decades of articles, sector rankings and citations in business media, exactly the kind of material language models have absorbed at scale. An independent fiduciary firm, even a strong local one, rarely has a comparable volume of French-language mentions on the open web. This is not a competence problem. It is a raw-material problem for the AI.

That gap is not neutral for the end client. An accounting or tax decision carries real liability, which pushes an AI toward identifiable trust signals rather than plain brand recognition. This is exactly where an independent firm’s room for manoeuvre sits: verifiable reviews, a clearly stated specialisation and consistent third-party sources carry proportionally more weight than in a less regulated sector.

Geneva is not London: why market size changes the rules

Geneva has neither the number of firms nor the volume of financial press coverage of a hub like London. On a smaller market, the pattern observed in mid-sized cities in the studies above likely applies more directly: fewer dominant global brands already baked into the data, and therefore more room for a firm that works its signals seriously.

That last point is worth stressing: a separate study measured that local-language queries produced richer results than their English equivalents, and that recommended firms almost systematically had a bilingual site or one written in the language of the query. For a Geneva firm, an English-only site or thin French content is a direct disadvantage, not a cosmetic detail.

In practice, this means a Geneva fiduciary firm that publishes detailed pages on its specialities (VAT, freelancer taxation, association accounting) in French, with concrete examples and questions specific to its activity, holds material that large international networks simply have not produced for this exact market. That is a structural advantage, not a passing trend.

A six-step method for testing your own visibility

Rather than trusting an impression from a single question asked to ChatGPT, which gives a different snapshot depending on phrasing and timing, here is the method for a reliable, month-over-month measurement:

1. Build a set of eight to twelve realistic queries. Not only "accounting firm Geneva", but also "accountant for freelancer Geneva reviews" or "VAT specialist firm SME Geneva", phrased the way a real client would type them.
2. Query at least three engines with web search enabled. ChatGPT, Google AI Overviews and Perplexity do not draw on exactly the same sources.
3. Log the names cited, their order, and whether your firm appears, for every answer. Mere presence is not enough, position in the list matters too.
4. Compare against a classic Google ranking for the same queries. The gap between the two lists is the single most useful piece of information from the whole exercise.
5. Repeat the test a month later. A position change after adding recent reviews or a new service page confirms which signal actually mattered.
6. Log every pass in a tracking table rather than isolated screenshots. It is the only way to tell a trend apart from a one-off result, with the method detailed in our article on measuring AI search visibility.

Table: the signals that carry the most weight in an AI answer

When this test is not the right investment

This method does not make sense in every case. Three situations where it is better to skip it for now:

• A firm that lives entirely on word of mouth and referrals from lawyers or notaries. If it never seeks new clients through the web, its visibility in an AI answer has no effect on the business.
• A firm without an up-to-date Google Business Profile or a recent website. Measuring AI visibility before fixing these basics is like timing a race before putting on running shoes.
• An activity under strict solicitation rules. In some regulated cases, compliance with professional conduct rules should take priority over general visibility optimisation.

In every other case, an owner discovering for the first time what an AI says about their sector generally prefers to know it directly rather than hear it from a client who noticed it first. That is the starting point of our overview of GEO and the six levers of visibility in AI answer engines.

None of this replaces professional judgement. A prospective client who picks a fiduciary firm from an AI answer still has to verify credentials, meet the team and check references, exactly as they would from a personal recommendation. What changes is the first filter: for a growing share of searches, that filter is now a language model rather than a list of blue links, and it rewards a different set of signals, several of which a small firm can directly influence.