AI Visibility for Asset Managers: The European Measurement Layer

AI visibility for asset managers

AI visibility for asset managers is the share of answers where an AI engine names your firm when a buyer asks it to recommend one. It is measured, not assumed. On 10 September 2026 the AI Asset-Manager Observatory logged 2,243 answers across five languages, and 104 firms were named at least once. In the English cut, Amundi appears in 71.8% of answers and BlackRock in 64.8%. The tenth placed firm, Nordea Asset Management, appears in 30.2%. Every other European manager on the roster is a rounding error. That gap is the whole problem, and it is measurable.

Why European asset managers lost control of the shortlist

For twenty years, distribution ran through intermediaries who could be met, courted and persuaded. A fund buyer heard a name from a salesperson, a consultant, a platform, a conference panel. Each of those channels was relationship work, and relationship work rewards effort.

An AI answer removes the relationship. The buyer types a question, receives three to ten names, and starts diligence on those names. There is no meeting, no follow up, no second chance to be remembered. Either the model produces your firm or it does not.

This is why the European position is uncomfortable. Asked in English for a European asset manager, engines reach for global giants: BlackRock, Vanguard, Fidelity, State Street, J.P. Morgan. In the September English cut, 17.9% of all firm mentions went to these non European firms, and the number rises sharply if you include their European subsidiaries. Recognisable global brands have more text about them, and language models recommend what they have read.

The measurement layer

You cannot manage a shortlist you never look at. Four public studies now cover the terrain, all built on saved raw answers rather than opinion.

Study	What it measures	Sample
AI share of voice in asset management	Share of voice, first place rate, grounded versus recall mode	2,243 answers, 5 languages
Financial Homepage Language Study	Language coverage on financial homepages	1,385 homepages analysed, 55 languages
AI crawler rules at European financial firms	Whether AI crawlers are allowed in at all	105 firms
European finance jobs, skills and pay	Whether the skills to act on it are being hired	794 job descriptions

Read together, they describe one system. Firms publish in one language, block or ignore the crawlers that would read them, and then wonder why the model does not know their name.

For a scorecard that separates the board numbers from the team numbers, see the six AI visibility KPIs for asset managers.

The three levers that move AI visibility

1. Be readable before you are recommendable

An engine cannot cite a page it was never allowed to fetch, and cannot summarise a claim the page never made in plain text. The AI crawler audit found European financial firms split sharply on this: some allow the major crawlers, many block them by default through a template they have never revisited. Check robots.txt, check that key pages render without JavaScript, and check that each page states in one sentence what the firm does and for whom.

2. Publish in the language the question is asked in

The Language Gap study shows that AI shortlists change with the language of the question. Ask in Dutch and the mix of named firms shifts. Ask in English and the global giants gain. A single English homepage is not a European strategy, it is a bet that every buyer asks the English question.

3. Put a number in public

Prose is forgettable and contestable. A published figure with a stated method is quotable, and quotable content is what gets cited. Our own experience is blunt: the studies on this site earn links and citations at a rate the opinion posts do not. That is a measurable difference, and it is the cheapest lever available.

What this is not

It is not a ranking of asset managers. A firm can be excellent and absent from AI answers, which is precisely the problem worth fixing. It is not a claim that AI answers decide allocations today. They decide the first shortlist, and first shortlists shape the funnel.

It is also not stable. Models update, retrieval sources shift, and a single quarter can move a name several points. That is why the Observatory is now re-read monthly rather than annually: a single reading is a photograph, a series is a trend.

Where to start

Run the check on yourself first. Ask five engines the questions your buyers actually ask, in the languages they ask them, and write down who gets named. Most European managers have never done this once. The live dashboard shows what the answers look like across 2,243 logged completions, and the study behind it explains the method. If your firm is not in the answer, you now have a baseline, a benchmark and a set of names to beat. Having that baseline run for your own firm, against the competitor set you actually pitch against, is what the AI visibility audit does.

What is AI visibility for asset managers?

It is the share of AI answers that name your firm when a buyer asks an engine to recommend a manager. We measure it as reach, the percentage of logged completions that mention the firm at least once, plus a first place rate for whether the firm is named earliest in the answer.

How many European asset managers do AI engines recommend?

Roughly ten carry the category. In the September 2026 English cut, the top ten firms took 42.1% of all firm mentions across 600 answers, and the tenth placed firm appeared in 30.2%. The long tail of European managers is effectively invisible to the engines.

Do AI engines recommend different managers in different languages?

Yes. The Language Gap study found the shortlist changes with the language of the question, and the Observatory records separate cuts for English, German, Spanish, Dutch and French. European Union firms take 82.1% of mentions in the English cut, but the specific names and their order move between languages: Amundi reaches 79.2% in French against 63.9% in Dutch.

How often should AI visibility be measured?

Monthly for a reading, quarterly for a frozen panel. Month to month movement is mostly noise, so we publish a monthly reading and hold the full multi engine panel to a quarterly cadence. A year of readings is what turns a single result into a trend.