Lately, the hot issue in marketing has been “optimize for AI,” but there isn’t one overarching AI engine that you should optimize for.ChatGPT, Perplexity, Claude, and Gemini all have different engines, algorithms, and models. They don’t pull information from the same places. Instead of just trying to optimize for AI, you should ask, “Which AI engines actually matter to my buyers, and what do those engines tend to cite?”
Those were among the questions that we at Greenfire Strategy were looking to get answered, and we couldn’t find solid existing research to answer them. So we did our own study and found that AI engines cite different sources, and that ChatGPT draws about 84% of its citations from vendors’ own content, while Perplexity pulls roughly four times as many from third-party platforms like Reddit and YouTube.
|
TL;DR Takeaways
|
For the Greenfire Strategy AI Citation Study, we analyzed 35,755 citations generated from 485 B2B marketing software queries across ChatGPT, Claude, Gemini, and Perplexity. The study classified all 5,091 cited domains to understand where citations were coming from. What we discovered was quite surprising.
What we discovered is that no single AEO tactic works equally well across all engines. The citations you need to become visible depend partly on which AI engines your buyers actually use. The extremes between ChatGPT and Perplexity are a perfect example.
ChatGPT is the most-used engine, but it’s also unusually difficult for non-vendor brands to break into. It cites the fewest sources, and 84% of the citations in the Greenfire study were to first-party vendor content. Meanwhile, Perplexity is much more reliant on earned media and third-party platforms, citing third parties about 4 times as often as any other engine. Claude and Gemini have a more balanced source mix, utilizing both first-party and third-party sources.
What one engine trusts may be completely invisible to others, and “optimizing for AI search” means you could be optimizing AEO for 4+ distinct answer retrieval systems at once, with very different ways of operating. Each engine has its own citation ecosystem, which matters when deciding where and how to optimize.
Greenfire’s research found an interesting piece of data: despite ChatGPT being the most popular AI engine currently used, it represented less than 4% of the citations in the study. Gemini, meanwhile, accounted for more than half. In other words, the AI engine with the largest audience isn’t automatically the one where brands will have the easiest path to visibility.
It may be tempting to make ChatGPT the center of your AEO strategy, but if your buyers don’t use it, it can leave a large portion of the AI-search audience untouched. Our research suggests that optimizing for additional engines can expand your potential audience coverage. Focusing only on ChatGPT reaches less than half of the audience represented in the study’s directional usage data. Adding Gemini brings that coverage to roughly 75%, while adding Claude pushes it into the mid-80% range.
Those figures should be treated as directional since different market-share sources use different methodologies. The broader lesson still stands, however. Instead of asking which engine is best to optimize for, ask which engine your audience actually uses and what citation environment you’re trying to enter.
According to Greenfire’s study, about 84% of ChatGPT’s citations came from vendor sources. For B2B marketers, that creates an important distinction compared to other AI engines. For example, if you’re a software vendor, your own content may already make up a significant portion of the citations ChatGPT draws from when providing answers. If you’re a non-vendor software company, however, you may never break through ChatGPT’s source ecosystem because you don’t necessarily have the same built-in citation opportunities as the vendors.
“I think the first thing you really need to answer is why you care about showing up in ChatGPT specifically. A lot of companies are looking to answer engines to recoup their lost organic traffic, and I think that’s a big mistake.” –Tracy Graziani
That doesn’t mean you should ignore ChatGPT. Its large user base makes it an important part of AEO and the AI-search landscape. It does mean that creating valuable content and optimizing it for AI may not be enough. You need to understand what kinds of information ChatGPT is actually drawing into its answers and where your brand fits into that system. ChatGPT isn’t impossible to enter; however, it should be treated as its own citation environment rather than a representative example of “AI search.”
Greenfire’s research found that approximately 21% of Perplexity’s citations came from third-party platforms such as Reddit, YouTube, and LinkedIn, compared to just 5% in the study’s ChatGPT breakdown. It casts the widest citation net of the four AI engines tested. That creates a very different opportunity for marketers.
Information published or discussed on relevant third-party platforms may be more likely to become part of Perplexity's citation ecosystem. On Perplexity, being represented outside of your own content can matter more than your own website. There’s an important distinction regarding first-party and third-party content, though: you don’t control third-party websites.
“Eveyone is hyping 3rd party platforms as the way to win AEO, and there is some value there, but there is also risk. Never build your empire on rented land.” –Tracy Graziani
You can participate in relevant communities, contribute useful information, and build a legitimate presence where your audience already spends time, but you don’t control Reddit, YouTube, LinkedIn, or any other third-party platforms. What this means is that third-party visibility can be part of your AEO strategy without becoming the entire strategy.
Your goal with third-party platforms isn’t to figure out what or how you can manipulate sites into mentioning your brand, but instead, to become a useful, credible source wherever your buyers and the AI engines serving them are looking for information. Community participation can support your visibility, but it shouldn’t replace building your own authoritative content and brand presence.
Greenfire’s research found relatively balanced citation mixes for both engines, with vendor and agency content accounting for significant portions of their citations alongside earned media, third-party content, and other sources. Gemini’s citation profile also stands out in the study’s overall numbers, accounting for more than half the citations captured. What this means is that the middle ground matters, especially since Gemini is keeping up with ChatGPT with having 1 billion monthly users.
The data collected from Claude and Gemini suggests that marketers shouldn’t abandon their existing content strategies and build four different AEO programs. Instead, marketers should understand where the differences matter. Some engine-specific considerations are worth layering onto a strong foundation.
For Gemini, that includes making your organization and subject matter easy to understand through clear entities, structured information, and strong connections to relevant knowledge sources. YouTube may also be a potentially important channel for Gemini visibility. Claude’s more balanced source mix similarly reinforces the value of building broad authority rather than relying solely on first-party or third-party content.
Marketing teams don’t have unlimited time and resources to create separate strategies for every AI platform. They shouldn’t need to, either. Instead, think about AEO in three layers.
Before deciding which AI engine to prioritize, first figure out which ones your buyers actually use. If your audience primarily uses ChatGPT, it belongs in your strategy despite its citation environment being a challenge. If your audience relies heavily on Perplexity, third-party visibility deserves more attention. If your audience is distributed across several platforms, a cross-engine strategy makes more sense than putting all of your resources into one.
Fundamentals don’t change from engine to engine. Make it clear who your company is, what you do, and what you know. Publish accurate, current, and specific content. Establish genuine subject matter expertise. Most importantly, make your information easy to understand and easy to reference. Having a strong foundation gives your brand something worth citing in the first place.
Once your foundation is established, you can adapt your strategy to each relevant engine. The goal isn’t to chase every possible optimization tactic but to understand why an engine behaves differently and make strategic decisions accordingly.
For ChatGPT: Focus on clear entities, encyclopedic and authoritative information, freshness, and content that communicates exactly what your company or experts are known for. Be quotable and unambiguous about what you are. The model extracts heavily from what it cites, so depth per page counts.
For Perplexity: Pay attention to freshness, well-supported facts, and legitimate participation in relevant communities and third-party platforms. A strong community presence is important. Breadth and frequency of mentions across platforms matter more than depth.
For Gemini and Claude: Prioritize entity clarity, structured information, and connections to relevant knowledge sources. Gemini leans on Google’s other platforms, so YouTube may deserve consideration, and Google rankings may be more important to Gemini than to other engines.
There’s no universal answer because there’s no universal AI citation ecosystem. “AI” isn’t one distribution channel. It’s 4+ different answer engines with different preferences, rules, and source collections.
ChatGPT may have the largest audience, but Greenfire’s research suggests its citation environment skews toward vendor content. Perplexity is more reliant on third-party platforms. Claude and Gemini occupy the middle ground between the two. That means “optimize for AI” is ultimately too broad to be useful advice. A better approach is to ask “where are my buyers looking for answers, and what sources do those AI platforms trust?” Once you know that, you can build a strategy around the engines that matter most without treating AI as one giant, interchangeable channel.
Download Greenfire Strategy’s full AI Citation Study to explore the methodology and research behind the per-engine citation patterns. Explore 35,755 citations across all four engines, with the per-engine breakdowns, methodology, and actionable tactics.
You can also run a free Signal Fire Snapshot to see where your brand currently appears across AI engines. This free snapshot will let you see where you actually stand, by engine, before you decide where to focus.