Wikipedia's Impact on AI Visibility and Recommendation Share: 2026 Study
Published October 20, 2026
A small-business owner in Ohio looking for payroll software, or a shopper in Texas comparing mattresses, can now ask an AI assistant such as ChatGPT or Perplexity for the best options and get back a list of brands. Some of those brands have a Wikipedia article and some do not. Because larger, better-known companies are more likely to have an article, we looked at whether brands with one are recommended more often and placed higher than companies of a similar size without one.
Our analysts collected the answers OpenAI's GPT-5.5 and Perplexity Sonar gave to 92 "best X" and "top companies for Y" questions in October 2026, recorded the 1,743 brand recommendations in them, and checked each of the 1,319 brands against English Wikipedia.
In the following sections, we compare recommendation share and position for brands with and without Wikipedia articles, compare positions within the same AI answer, test whether the gap holds for brands with websites of similar size, look at employee counts among brands with an article, and split the results by B2B and B2C.
Recommendation Share for Brands With and Without Wikipedia
Two numbers describe how much of the AI answer space each group holds: its share of all recommendations, and how often its recommendations land near the top of the list. Position can matter because readers of a long AI answer may not read all the way to the end of the list.
| Platform | AI Answers | Recommendations | Brands With Wikipedia | Recommendation Share With Wikipedia | Top-3 Rate With Wikipedia | Top-3 Rate Without Wikipedia | Average Position With Wikipedia | Average Position Without Wikipedia |
|---|---|---|---|---|---|---|---|---|
| OpenAI GPT-5.5 (via Perplexity Agent API, B2B prompts only) | 45 | 897 | 62.8% | 64.2% | 19.4% | 7.2% | 10.8 | 13.8 |
| Perplexity Sonar | 92 | 846 | 74.9% | 76.1% | 37.7% | 16.3% | 5.1 | 6.7 |
| Both platforms | 137 | 1,743 | 66.9% | 70.0% | 29.1% | 10.7% | 7.8 | 11.1 |
- Across both platforms, recommendations of brands with a Wikipedia article reached the top three positions 29.1% of the time, against 10.7% for brands without one.
- Brands with a Wikipedia article made up 66.9% of the distinct brands recommended and took 70.0% of all recommendations.
- Their recommendations were first on the list 9.8% of the time, compared with 3.3% for brands without an article.

Wikipedia Brands Ranked Higher Within the Same AI Answer
Comparing brands inside a single answer removes differences between categories, such as hotel chains, where almost every option has an article, against SEO agencies, where few do. In 101 answers that recommended brands from both groups, we compared the average position of each group.
| Platform | Answers With Both Groups | Answers Where Wikipedia Brands Ranked Higher on Average | Average Positions Ahead |
|---|---|---|---|
| OpenAI GPT-5.5 (via Perplexity Agent API, B2B prompts only) | 43 | 40 | 4.3 |
| Perplexity Sonar | 58 | 50 | 2.1 |
| Both platforms | 101 | 90 | 3.1 |
- In 90 of the 101 answers, or 89.1%, the brands with a Wikipedia article sat higher on average than the brands without one.
AI Visibility After Controlling for Company Size
Larger, better-known companies are more likely to have a Wikipedia article and more likely to be recommended, so the raw gap mixes the two. To separate them, brands were grouped by the popularity rank of their website on the Tranco list, a research ranking of the most visited sites, and the two groups were compared within each size band.
| Website Popularity Rank | Recommendations With Wikipedia | Top-3 Rate With Wikipedia | Recommendations Without Wikipedia | Top-3 Rate Without Wikipedia | Average Position With Wikipedia | Average Position Without Wikipedia |
|---|---|---|---|---|---|---|
| Top 10,000 sites | 552 | 36.4% | 28 | 7.1% | 6.2 | 10.2 |
| 10,001 to 100,000 | 405 | 27.7% | 100 | 14.0% | 8.3 | 8.1 |
| 100,001 to 1,000,000 | 231 | 17.7% | 217 | 9.7% | 10.1 | 11.5 |
| Outside top 1,000,000 | 32 | 3.1% | 178 | 10.7% | 12.7 | 12.4 |
- Weighting each band by its number of recommendations, brands with a Wikipedia article had a top-3 rate 14.9 percentage points higher than brands of similar website size without one. The gap was 21.9 points on Perplexity Sonar and 7.2 points on GPT-5.5.
- The pattern reversed for the smallest websites. Outside the top million sites, the 32 recommendations of brands with an article reached the top three 3.1% of the time, against 10.7% for the 178 without one.

Company Size Among Brands With Wikipedia Articles
Wikidata, the structured database linked to Wikipedia, lists an employee count for some companies. Among brands with a Wikipedia article and a listed employee count, the top-3 rate stayed between 31.2% and 35.1% in each band from 1,000 to 9,999 employees up to 100,000 or more. The 13 brands with fewer than 1,000 employees had a top-3 rate of 11.8%, from only 17 recommendations.
| Employees in Wikidata | Brands | Recommendations | Top-3 Rate | Average Position |
|---|---|---|---|---|
| Fewer than 1,000 | 13 | 17 | 11.8% | 9.1 |
| 1,000 to 9,999 | 64 | 97 | 35.1% | 6.9 |
| 10,000 to 99,999 | 80 | 110 | 32.7% | 7.2 |
| 100,000 or more | 80 | 125 | 31.2% | 7.9 |
| No employee count listed | 645 | 871 | 28.0% | 7.9 |
Wikipedia and AI Recommendations in B2B and B2C Categories
The gap appeared in both business and consumer questions. Consumer categories such as airlines, smartphones and running shoes are dominated by brands with articles, while B2B service categories such as SEO agencies and IT staffing include many firms without one.
| Category Type | Recommendations With Wikipedia | Top-3 Rate With Wikipedia | Recommendations Without Wikipedia | Top-3 Rate Without Wikipedia |
|---|---|---|---|---|
| B2B | 943 | 25.2% | 455 | 10.3% |
| B2C | 277 | 42.2% | 68 | 13.2% |
What Wikipedia AI Visibility Trends Mean for 2027 and Beyond
A company's Wikipedia article has long been one of the first things people find when they look the company up. As more people research purchases by asking AI assistants, the article may also affect which companies those assistants recommend and how high they place them, and that link is likely to last as long as the assistants keep drawing on Wikipedia. Wikipedia's notability guideline for companies asks for significant coverage in independent, reliable sources before a company gets an article, so the companies most likely to benefit are the ones journalists and other independent writers have already covered.
A company with a Wikipedia article can keep it accurate and well sourced, and a company without one can start by earning coverage in independent publications, which Wikipedia editors look for before they accept a new company article.
Data Collection and Methodology
The Top GEO Agencies research team wrote 103 prompts in October 2026 asking for the best companies or brands in 50 B2B and 53 B2C categories, from CRM software and payroll providers to mattresses and cruise lines, each ending with a request to list specific companies or brands. Each prompt was sent once to each platform with web search turned on: Perplexity Sonar, and OpenAI's GPT-5.5 called through Perplexity's Agent API with its web search tool. Perplexity Sonar answered 92 prompts and GPT-5.5 answered 45, all of them B2B; the remaining calls failed when the study's API credit ran out, and no prompt was run twice. In total, 103 prompts were run and 92 returned recommendations, so every figure in this study comes from answers to those 92 prompts. Gemini was also planned, but the Gemini API refused Google Search grounding on quota grounds and returned only 3 ungrounded answers, so Gemini is left out of every figure.
A separate low-cost model call read each answer and listed the brands it recommended in order of first appearance, with each brand's official website. Position is that order, counting each brand once per answer. Product names were kept as written, so "Paychex Flex" counts as a recommendation of Paychex.
Each brand was checked against English Wikipedia through the Wikipedia and Wikidata APIs. A brand counted as having an article when an article's Wikidata record listed the same official website, including a parent company's website for product names, or when the article title matched the brand name and its description identified a company, brand or product. Disambiguation pages did not count. In a manual check of 40 brands marked as having no article, 1 had one (the car rental brand Enterprise), so the without-Wikipedia group includes a small number of misclassified brands.
Recommendation share is a group's recommendations divided by all recommendations. The top-3 rate is the share of a group's recommendations that appeared in positions 1 to 3. Average position is the mean list position. The website size bands use each brand's domain rank on the Tranco list 64VKX, downloaded in October 2026; domains not in the top million form the last band. The size-adjusted gap is the difference between the groups within each band, weighted by the band's total recommendations.
The study is correlational and controls for company size through website popularity bands. It shows that brands with Wikipedia articles were ranked higher by these AI platforms, and it cannot show that adding an article would move a given brand up, since brand age, press coverage and the volume of reviews and comparison articles affect both. Each prompt was run once, and AI answers vary between runs. The GPT-5.5 results come only from B2B prompts and from Perplexity's search tool, which may differ from the ChatGPT app. Domains came from the extraction model and were not all checked by hand, and Wikidata employee counts exist mainly for companies that already have an article.
Sources
- Top GEO Agencies, Wikipedia AI Visibility Study: analysis of 1,743 brand recommendations in 137 answers from OpenAI GPT-5.5 and Perplexity Sonar to 92 "best X" prompts, matched against English Wikipedia and Wikidata, conducted October 2026.
- Wikimedia Foundation, MediaWiki Action API, English Wikipedia
- Wikimedia Foundation, Wikidata API
- Tranco, A Research-Oriented Top Sites Ranking, list 64VKX
- Perplexity, Agent API documentation
- Wikipedia, Notability (organizations and companies)