Study · May 2026
May 5, 2026 · 400 probes · 4 models · 10 brands · 10 queries

GEO Citation Study — how LLMs see your brand today

Empirical analysis across Claude, ChatGPT, Perplexity and Gemini: which brands get cited, which disappear, and why? Including a reproducible probing setup so you can run the same test on your own brand.

01 · Top line

The numbers in four sentences

What 400 probes across 4 models and 10 brands consistently show.

Overall citation rate
98%
Average across all models. The high rate shows: these 10 Hidden Champions are present — they have not "disappeared".
Mention density
11.3
Average brand mentions per response. Range 4.9 to 17.6.
Visibility index
88/100
Composite score from citation rate, mention density and list position.
Avg. response length
503 w.
Average word count per response. Range: 263 (Perplexity) to 814 (Gemini).
Headline finding

The Hidden Champions in this study do not disappear from the models — they show up in over 95% of relevant queries. What does differ strongly is how models answer: how deeply they embed a brand, and which sources they pull from.

02 · Brand ranking

Who is visible — and who is less so

Visibility index per brand, averaged across all 4 models. 100 = perfectly cited, high mention density, frequently first in lists.

# Brand Type Visibility index Citation rate Density Best model
1 Knauf B2B
95
100% 13.6 ChatGPT (GPT-5)
2 Stihl Consumer
92
100% 11.8 ChatGPT (GPT-5)
3 Miele Consumer
91
100% 11.7 Gemini 2.5 Pro
4 Festo B2B
90
100% 12.9 ChatGPT (GPT-5)
5 Kärcher Mixed
88
100% 8.8 Gemini 2.5 Pro
6 Würth B2B
87
95% 9.9 Gemini 2.5 Pro
7 Sennheiser Consumer
87
98% 11.3 ChatGPT (GPT-5)
8 Trumpf B2B
86
98% 11.1 ChatGPT (GPT-5)
9 Liebherr B2B
86
95% 10.6 Gemini 2.5 Pro
10 Hilti B2B
82
90% 11.3 ChatGPT (GPT-5)
03 · Model divergence

The four models don't think alike

Same brands, same queries — four very different answer styles.

ChatGPT (GPT-5)

Visibility index 96

Citation rate: 98.0%  ·  Mention density: 17.6  ·  Avg. words: 577

High-volume profiler. GPT-5 writes long-form (577 words on avg.), repeats brand names often (density 17.6) and tends to produce structured top-N lists.

Gemini 2.5 Pro

Visibility index 95

Citation rate: 99.0%  ·  Mention density: 16.2  ·  Avg. words: 814

Longest output. Gemini 2.5 Pro produces the most verbose answers (814 words, density 16.2). Reasoning tokens must be budgeted generously.

Claude Sonnet 4.5

Visibility index 82

Citation rate: 100.0%  ·  Mention density: 6.5  ·  Avg. words: 359

Structured and concise. Claude Sonnet 4.5 answers more compactly (359 words, density 6.5) — high hit-rate, less wordy noise.

Perplexity Sonar Pro

Visibility index 80

Citation rate: 93.0%  ·  Mention density: 4.9  ·  Avg. words: 263

Source-anchored. Perplexity Sonar Pro stays brief (263 words, density 4.9) and often produces inline citation markers — closest to a classic search experience.

Take-away

The models do not differ primarily in whether they know a brand, but in how they embed it. GPT-5 and Gemini generate verbose, dense profiles (>500 words, high mention density). Claude and Perplexity answer more concisely with a clear focus on sources. If you want to "rank" a brand, you should not just measure "am I mentioned", but "in what style, with what depth, with which sources".

04 · B2B vs. consumer

The expected bias — does not exist in this sample

Hypothesis going in: B2B Hidden Champions disappear more than consumer brands. The data says otherwise.

Consumer brands (n=12)

89.6 visibility

Citation rate99.2%
Mention density11.6
ExamplesSennheiser, Stihl, Miele
B2B brands (n=24)

87.5 visibility

Citation rate96.3%
Mention density11.6
ExamplesTrumpf, Hilti, Würth, Festo, Knauf, Liebherr
Difference

2.1 points

Visibility gap+2.1 consumer
Citation gap2.9 pts
AssessmentWithin noise — no systematic bias against B2B in this sample.

What this means: The common assumption "B2B has a visibility problem in LLMs" does not hold for established Hidden Champions with strong Wikipedia/press footprint. The brands probed here are global market leaders with decades-long PR trails — and that is exactly the trail the models read.

The flip side: brands without Wikipedia entries, without German business-press coverage, without strong industry-media backlinks — those could in fact disappear from the answers. This study cannot test that, because it deliberately targets established brands. The next iteration should explicitly include "mid-tier" Mittelstand companies.

05 · Disappearance points

Where a brand drops out — and why

Concrete brand × model combinations with citation rate below 50%. With few hits, more the exception than the rule.

This sample contains no real disappearers — no brand falls below 50% citation rate on any model. That is itself a finding: established Hidden Champions are reliably visible across all 4 major LLMs.

The interesting pattern is not "disappearance" but list position: a brand that lands at position 6 or 7 in a "top providers" list is essentially invisible to the end user. This study captures position rank for list queries — see the CSV for details.

06 · Source clusters

Where models draw their answers from

Distribution of cited source clusters across all 400 probes.

Company website
179
Other
152
Social (LinkedIn / X)
25
Wikipedia
10
Government / research
6
German business press
3
Industry / trade press
2
International business media
2

Top sources across all brands

kununu.com 22× cited 10 brands
liebherr.com 21× cited 2 brands
festo.com 17× cited 1 brands
sennheiser.com 17× cited 1 brands
trumpf.com 17× cited 1 brands
glassdoor.de 16× cited 9 brands
hilti.group 16× cited 2 brands
stihl.de 14× cited 1 brands
wuerth.com 13× cited 2 brands
knauf.de 13× cited 1 brands
Observation

Across all models, Wikipedia and the respective company domains dominate as primary sources. German business and industry press is visibly present — international outlets like Reuters/Bloomberg appear noticeably less often than they would for a US-only probe.

07 · Query types

Which questions surface a brand best

Citation rate per query type. Helps you understand which user searches your brand actually competes in.

1·brand_direct Direct brand profile
100%
2·brand_leadership Brand leadership / market position
100%
4·comparison Brand-vs-brand comparison
100%
5·innovation Innovation track record
100%
6·reputation Reputation / customer perception
100%
8·recommendation Personal recommendation
100%
10·future_outlook 5-year outlook
100%
9·news_recency Recent news / changes
98%
3·category_leader Top providers in category
95%
7·hidden_champion Hidden Champions of German Mittelstand
83%
08 · Reproduce

Measure your own brand

The full probing setup is open. Clone the repo, add your brand, run it — and compare your brand to the 10 Hidden Champions here.

What you need

  • Node.js 20+
  • API keys: Anthropic, OpenAI, Perplexity, Google AI (Gemini)
  • An .env file with the four keys
  • ~$5 of API credits per 100 probes

Setup in 3 steps

$ git clone https://github.com/craid/geo-citation-study
$ cd geo-citation-study && npm install
$ cp .env.example .env <- add your keys
$ node probe.js --brand=your_brand --smoke <- 1 brand × 4 models × 10 queries
$ node analyze.js <- aggregation + CSV export

Data files for this study

  • probes.csv — all 400 individual probes with sentiment, mentions, source clusters
  • aggregated.csv — brand × model rollups
  • sources.csv — top sources with frequency per cluster

Methodology in two minutes

Models
Claude Sonnet 4.5 · ChatGPT GPT-5 · Perplexity Sonar Pro · Gemini 2.5 Pro
Brands
Trumpf, Sennheiser, Hilti, Würth, Stihl, Miele, Festo, Knauf, Liebherr, Kärcher
Queries
10 templates (brand-direct, category leader, Hidden Champion, comparison, recommendation, recency, outlook ...)
Probes
10 brands × 10 queries × 4 models = 400 requested probes; 400 successfully evaluated (0 errors).
Parameters
Temperature 0.3 (reasoning models: default), max_tokens 1,500 (16k for reasoning models), German system prompt, factual research instruction.
Sampling
1 run per brand × query × model. Limitation: no multi-run for stability — next iteration with n=3.
Visibility index
Composite (0-100): 60% citation rate, 30% mention density (cap 10), 10% inverse list position. Higher is better.
CRAiD

Want a deep probe for your brand?

We build GEO measurement setups for brand teams that want to know how LLMs describe them today — and how that shifts when new sources, new models or new competitors enter the picture.

← Back to Insights