Blog
AI Search
An In-Depth Look at Source Distribution Patterns in AI Search
This data may flip your assumptions about where AI search engines & LLMs look for tech sector information.
Yahel Oren
Data Scientist
Aug 26, 2025
Spotlight: The Technology Sector
AI search engines and LLMs pull up all kinds of information for all kinds of prompts and purposes. And as you may already know, in the eyes of LLMs, not all information sources are equal. Certain sources stand out as their preferred choices.
What are top-choice sources across all major AI Search platforms?
Blogs, Wikipedia and Reddit. Across major platforms such as Gemini, OpenAI, and Perplexity, blogs are a primary source of content. These models frequently turn to and remix blogs when forming answers. Next comes Wikipedia. OpenAI’s models show a particular preference for Wikipedia and Reddit, drawing heavily from these sources to inform responses and generate content.
All that said, the weight given to blogs, Wikipedia, or Reddit varies greatly, and sometimes surprisingly so, depending on what’s being asked. We’ll explore this further below, focusing on the tech sector.
Does geo-location impact AI Search source distribution?
Yes. While some trends can be generalized, it’s important to note that sources are distributed differently depending on the geographic region, illustrating that no single source dominates globally, but rather, different regions prioritize different types of content. Below, we’ll examine U.S. data.
Why is it important to understand source distribution patterns in AI search?
Understanding these patterns helps us see what influences AI-generated responses and how we can prepare our brands to be found, cited, and positively presented by different AI platforms.
How the Technology Sector is Different
To pinpoint whether the patterns discussed above, along with others, apply to the tech sector (or not), let's examine the source distribution patterns of the three main AI search platforms in the U.S., specifically regarding tech prompts.
We'll explore whether they are similar or differ significantly and we’ll look at what this could mean for technology brands as they shape and reshape their content. Let’s begin with OpenAI.
Distribution of Top Sources in OpenAI's Answers
Wikipedia’s presence at 20.8% makes sense: we already know OpenAI has a bias towards citing Wikipedia. But the fact that blogs dominate at a whopping 35.3% is interesting. That's brands fully beating out Wikipedia for influence in AI search. That’s one in three versus one in five.
Reddit stands at 21.7%, while TechRadar at 4.9% is nearly tied with Arxiv.org at 4.8%. Think about that for a second: when it comes to pulling up tech information, random Reddit comments are being cited far more often than the world's most prestigious tech review publication or archive (arxiv.org boasts nearly 2.4 million scholarly articles).
Medium, where many tech leaders share thought leadership, scores just 1.4%, trailing behind LinkedIn at 2.3%. Meanwhile, news sources outrank all of them, including TechRadar and arXiv.
What we're seeing is OpenAI prioritizing readability and accessibility over authority signals. Why? Because news is easily winning over Medium articles and Reddit discussions are competing with Wikipedia. That's just strong confirmation.
Let’s compare all this with Gemini now.
Distribution of Top Sources in Gemini’s Answers
Without taking a close look, we can instantly see that one single category beats all others by a huge margin. That’s blogs. With Gemini, blogs don’t just dominate tech information sources, they almost monopolize them. That’s incredibly important for tech brands to know if they want their content to come up in Gemini’s answers.
Blogs’ presence at 79.9% is truly impressive. That’s almost 80%, or four out of five sources.
We can confidently say that Gemini has an ultra-strong bias towards citing blog posts – tech brands, take note! Especially if you were thinking the content era was over.
Next comes Medium at 4.8%, higher than for OpenAI. Then, digitalprojectmanager at 3.2% and news at 2.8%. Wikipedia is second to last, with 1.5%. Reddit does not even feature on the list. This is very different from what we observed in OpenAI’s source distribution.
Let’s take a look at Perplexity now.
Distribution of Top Sources in Perplexity's Answers
After reviewing OpenAI and Gemini, we expected a strong presence of blogs here as well. That said, the 72% source reference is twice as high as for OpenAI and slightly lower than for Gemini, though still comparable.
What’s super interesting here is Reddit’s “return” at 18.9%, comparable to 21.7% in OpenAI. We can conclude that Gemini is the only AI search engine that does not turn to Reddit for answers when it comes to the tech sector.
What does all this mean for brands? It means, if they want their brand to be cited by OpenAI and Perplexity, they should continue to focus on blogs as well as their presence on Reddit. But Reddit presence does not always mean positive exposure. At 18.9%, it’s key for brands to pay attention to user feedback on this platform, as negative sentiment could impact AI’s view of a tech brand.
LinkedIn stands at 3.8% here, compared to 2.3% in OpenAI whereas Gemini doesn’t turn to LinkedIn for answers. While this figure may seem low, it might not be a bad idea for brands to repurpose their blog posts for LinkedIn posts, if they want to increase their chances of being cited.
Like Gemini, Perplexity does not turn to Medium for answers. Cloud.google is at 1% here, compared to 1.7% for Gemini, and zero in OpenAI.
What Do Source Distribution Patterns in AI Search Answers Mean for the Tech Sector and Tech Brands?
Understanding how AI search engines (in general as well as particular platforms) prioritize sources is key for brands aiming to enhance their visibility and positive reputation in AI Search answers – an incredibly fast-growing search medium.
The U.S. data for the tech sector reveals that blogs dominate with an overwhelming presence, especially in Gemini, where nearly 80% of references come from blog content. Meanwhile, OpenAI heavily relies on a mix of blogs, Wikipedia, and Reddit, highlighting the importance of a diverse content strategy.
Tech brands that want to appear in AI’s answers, in a positive way, must optimize their blog content and maintain a strong, positive presence on Reddit. Conversely, the nearly exclusive reliance on blogs by Gemini suggests that brands focusing on content creation in blog formats could gain a competitive edge there – that is, if their blogs are written in a way that AI can easily read and understand.
Staying on top of these source distribution patterns can enable tech brands to adapt their AI search optimization strategies proactively, ensuring they are not only visible but also favored by AI-driven search platforms. The future belongs to those who understand and leverage these insights effectively.
Yahel Oren
Data Scientist
