NALMERT RESEARCH · 2026
Which GEO & AI Search Providers Does AI Recommend? — 2026 Study
80 measured AI responses across ChatGPT, Perplexity and Yandex Alice reveal which providers appeared most often — and which sources AI systems cited.
Last updated 3 September 2026
Executive summary
Key findings from a bounded measurement wave.
iPullRank
Most frequently mentioned provider · 10%iPullRank
Most explicit recommendations · 5%No provider dominated
The highest mention rate remained one in ten responses.Search & GEO baseline
No observed mentions, recommendations or provider citations.Provider-owned citations
56.6% of 578 validated visible citation rows.This study measures frequency of appearance within a fixed prompt set. It does not measure service quality and is not a ranking of the best agencies.
Measurement coverage
Three engines measured. Four markets observed.
Gemini is not counted as measured. Alice covers Russia only, and each market is reported as its own bounded sample.
One UK prompt: technical failure after two attempts.
Russian market only.
Unavailable in the research environment.
ChatGPT + Perplexity
ChatGPT + Perplexity
ChatGPT + Perplexity
ChatGPT + Perplexity + Alice
Trust layer
Data validation and entity normalization.
Provider names, recommendations and citation sources were separated before any aggregate was calculated.
Recommendation flags rejected
Recommendation required explicit recommendation language and a real identifiable entity.
Manual spot-check
Twenty observations sampled across engines and markets.
Detected errors after validation
The post-cleaning spot-check found no remaining classification errors.
Provider frequency
Most frequently mentioned providers in the measured prompt set.
| Provider | Mentions | Rate | Prompt coverage |
|---|---|---|---|
| iPullRank | 8 | 10.0% | 6 |
| Head Promo | 6 | 7.5% | 4 |
| Found | 6 | 7.5% | 5 |
| Pixel Plus | 6 | 7.5% | 5 |
| Go Fish Digital | 5 | 6.3% | 5 |
| First Page Sage | 5 | 6.3% | 5 |
| NeuroReach | 4 | 5.0% | 4 |
| White GEO | 4 | 5.0% | 4 |
| Digital Geeks | 4 | 5.0% | 4 |
| Directive Consulting | 4 | 5.0% | 3 |
Explicit recommendations
Most frequently explicitly recommended providers.
Recommendation is calculated separately from mention. Citation sources do not inherit a recommendation.
| Provider | Explicit recommendations | Rate |
|---|---|---|
| iPullRank | 4 | 5.0% |
| Head Promo | 3 | 3.8% |
| Found | 3 | 3.8% |
| Ashmanov & Partners | 3 | 3.8% |
| Passion Digital | 3 | 3.8% |
| Pixel Plus | 2 | 2.5% |
| NeuroReach | 2 | 2.5% |
| White GEO | 2 | 2.5% |
| Omnius | 2 | 2.5% |
| Varn | 2 | 2.5% |
Signal definitions
Mention ≠ Recommendation ≠ Citation.
Mention
An identifiable provider appears in the answer. Appearance alone does not imply endorsement.
Explicit Recommendation
The answer specifically recommends or proposes the provider for consideration.
Visible Citation
The interface exposes a visible source URL. The cited publisher is not automatically the mentioned provider.
Engine and market overlap
Where providers appeared across engines and markets.
No Gemini inference is included. Alice is explicitly limited to the Russian dataset.
Cross-engine providers
- iPullRankChatGPT + Perplexity
- FoundChatGPT + Perplexity
- Head PromoAlice + Perplexity
- Pixel PlusAlice + Perplexity
- Go Fish DigitalChatGPT + Perplexity
- Directive ConsultingChatGPT + Perplexity
Market leaders by observed mentions
- International EnglishiPullRank · 4
- United StatesiPullRank · 4
- United KingdomFound · 6
- RussiaHead Promo / Pixel Plus · 6
Market samples differ by language and engine coverage; counts are not treated as directly interchangeable market scores.
Citation ecosystem
What kinds of sources AI interfaces exposed.
Research sources and AI citation sources are reported separately. Other/Unclear remains visible rather than being forced into a stronger category. For a context-level view of these relationships, see our secondary citation–provider association audit.
Most visible citation domains
- rank4ai.co.uk17
- ipullrank.com12
- found.co.uk11
- tilio.co.uk11
- passion.digital / vc.ru10 each
Transparent baseline
Search & GEO recorded no observed presence in this wave.
Search & GEO recorded 0 observed mentions, 0 explicit recommendations and 0 visible provider citations in this measurement wave.
At the time of measurement, the publisher operated under the working brand Search & GEO. The project was subsequently renamed Nalmert before publication. The historical result is retained without adjustment.
Methodology
A frozen, evidence-led measurement process.
Frozen prompts
Thirty-six prompts were fixed before measurement and were not adapted after viewing results.
Sequential collection
Requests were executed one at a time. Clean sessions reduced—but cannot eliminate—personalization and platform variance.
Markets
International English, United States, United Kingdom and Russia were retained as separate samples.
Engines
ChatGPT, Perplexity and Alice were measured. Alice covers Russia only. Gemini was unavailable.
Raw evidence
Completed responses were preserved with Prompt ID, timestamp, visible URLs and screenshots.
Entity validation
Identifiable entities were separated from headings, categories, examples and citation publishers.
Calculations
Rates use 80 completed observations. Coverage counts distinct prompts, engines and markets. No composite score is used. Future measurement waves will measure Nalmert as the current brand.
Research sources
The frozen matrix, raw observations, screenshots, validation rules and calculation outputs produced by this study.
AI citation sources
External URLs visibly surfaced by an AI interface. These domains are the object of citation analysis—not methodological authorities for the study.
Limitations
What this dataset cannot prove.
Time and session dependence
AI answers may change across dates, sessions, accounts and interface versions.
Small bounded sample
Eighty responses support descriptive findings, not universal market claims.
Prompt sensitivity
Different wording can produce different provider and source selections.
Uneven engine coverage
Alice is Russia-only, Gemini is unavailable, and one Perplexity prompt failed.
Visible citations vary
Interfaces expose and format source links differently.
Applying this approach to a particular company requires a separately scoped baseline; the provider frequencies above cannot substitute for one. The AI Visibility Audit outlines how that diagnostic scope can be defined.
Raw data & reproducibility
Download the sanitized research package.
Includes the frozen Prompt Matrix CSV, aggregated provider metrics CSV, citation-domain classification CSV and public methodology. Full raw AI responses are withheld pending separate privacy and copyright review.
Research author
Evidence, interpretation and accountability.
Continue exploring
Related services, methodology and expertise.
Companion analysis
ChatGPT and Perplexity did not surface identical provider sets.
Read the evidence-led comparison: ChatGPT vs Perplexity: Do They Recommend the Same GEO Providers?
Cite this research
Suggested attribution: Nalmert Research 001 (2026), “Which GEO & AI Search Providers Does AI Recommend?” — nalmert.com/research/which-geo-ai-search-providers-does-ai-recommend-2026-study/
A backlink is not required. When referencing the findings, please preserve the measured-sample context and stated limitations.