If you want ChatGPT, Google’s AI mode, or Perplexity to recommend your business, you need to be mentioned on websites other than your own. This practice is called “mention building,” and it is the single most effective tactic for increasing your visibility in AI search results right now.
But most businesses get it wrong. They scatter their brand name across as many websites as possible, hoping something sticks. At TJ Digital, we track AI citations for about 30 clients across different industries, and the data is clear: only a small handful of third-party sources actually influence what these models recommend.
This article breaks down which mentions matter, how to find them, and how to build a targeted mention strategy. This is the 8th AI search visibility factor identified by Kevin Indig in his Growth Memo newsletter, and it’s one we spend more time on than almost anything else.
Table of Contents
ToggleHow Do AI Models Decide What to Cite?
@tjrobertson52 LLMs only look at about 12 pages when recommending brands. Are you mentioned on the right ones? #AIsearch #ChatGPT #SEO #MentionBuilding #MarketingTips
♬ original sound – TJ Robertson – TJ Robertson
Large language models do not scour the entire internet looking for your brand. When someone asks ChatGPT or Google’s AI mode for a recommendation, the model performs a real-time search using a process called Retrieval-Augmented Generation (RAG). It might retrieve around a hundred web pages, but it only examines about a dozen of them before generating a response.
Being mentioned on 500 websites does not matter if none of those websites are in that dozen. The only mentions that count are the ones on the pages the model actually reads.
A SEMrush study of over 150,000 AI citations found that product recommendation sites were the most-cited source type for buying-intent queries. Consumer review platforms came in second. Traditional media outlets ranked third. Random blog posts and low-authority directories barely registered.
Here’s a breakdown of the most-cited source types:
| Source Type | Examples | Relative Citation Volume |
| Product recommendation listicles | Tom’s Guide, Wirecutter, niche “best of” lists | Highest |
| Consumer review platforms | Trustpilot, BBB, Google Reviews | High |
| Traditional media | Forbes, NYT, Wired | Moderate |
| Digital-native media | TechCrunch, The Verge | Moderate |
| YouTube/Video reviews | Product comparison channels | Moderate |
| Discussion forums | Reddit, Facebook groups | Varies by industry |
AI does not treat all mentions equally. It pulls from a small set of trusted sources, and those sources vary significantly by industry.
Does Every Mention Help You Rank?
No. A large language model will not recommend your business simply because your brand name appears somewhere online. The mention has to function as a recommendation, either explicitly or implicitly.
Being listed at position 8 on a directory page with two reviews sends a weak signal. Being ranked at position 1 or 2 on that same page, with strong reviews behind you, sends a much stronger one. The difference between those two positions on a frequently cited page can be the difference between getting recommended and being invisible.
This is the main reason third-party reviews matter so much for AI visibility. Reviews push you higher on the directories and listicles that AI models cite. A higher position on a cited page translates directly into recommendations.
How Do You Find the Right Sources for Your Industry?
The websites AI models cite are completely different depending on your industry. For a B2B SaaS company, the top sources might be G2, Capterra, and Crunchbase. For a local service business, it could be Yelp and a handful of Reddit threads. For an ecommerce brand, it might be niche review blogs and YouTube channels.
You cannot guess which sources matter. You need to test it.
Here’s the process we use at TJ Digital:
- Build a list of prompts. Write out every question a potential customer might type into ChatGPT or Google’s AI mode. Include broad queries (“best window tinting in Houston”) and specific ones (“affordable ceramic window tint for SUVs”).
- Run those prompts through the major models. Use ChatGPT and Google’s AI mode at minimum. Look at the sources each model cites alongside its recommendations. If sources are not shown, ask: “Why did you recommend that company?”
- Track the sources in a spreadsheet. After running dozens of prompts, patterns will emerge. Three to five websites will make up the majority of citations. This becomes your “mention hit list.”
- Prioritize ruthlessly. Those frequently cited sources are your targets. Everything else is secondary.
Software tools like Peec AI and Semrush’s Brand Performance feature can help automate parts of this process. Most tools in this space are still maturing, but the important thing is that you are tracking real prompts and real citations, not guessing from traditional SEO data.
Which Third-Party Sources Should You Focus On?
Once you have identified your top cited sources, the mentions you need to build will typically fall into three categories: directories, forums, and articles.
Directories and Review Sites
For many industries, one to three directories dominate AI citations. It might be Clutch for marketing agencies, G2 for software, Yelp for local businesses, or Google Business Profile for just about everyone. BrightLocal found that Yelp appears in roughly one-third of AI local search results.
The strategy is straightforward. Get listed on the directories that AI models actually cite, then focus on getting reviews there. Not reviews everywhere. Reviews on the specific one to three platforms that matter.
If you are already on a cited directory but ranked low because you only have a few reviews, that is one of the fastest wins in AI optimization. A handful of new reviews can move you from position 8 to position 1 on a page that ChatGPT reads every time someone searches for your type of service.
Discussion Forums (Reddit, Facebook Groups)
For B2C businesses especially, Reddit threads and Facebook group discussions are frequently cited by AI models. But it is typically the same threads and posts being cited over and over again.
Instead of randomly posting about your brand across Reddit, take a surgical approach:
- Identify the specific threads that show up in your AI citation tracking.
- Leave genuinely helpful comments on those threads.
- Mention your brand naturally only if relevant. Do not promote. You will get banned from the subreddit and your comment will be deleted, which defeats the entire purpose.
The goal is to be present in the exact conversations that AI models are reading when they make recommendations.
Articles and Listicles
This is the hardest category to influence, but also the most impactful. “Best of” listicles and comparison articles account for a disproportionate share of AI citations. Research suggests that nearly half of AI-cited URLs for purchase-related queries come from comparison-style content.
If a highly cited article in your industry does not mention your brand, reach out to the author. Ask what it would take to be included. Some authors are happy to add you for a small fee or a reciprocal mention, and that kind of investment can pay for itself quickly.
You can also create your own comparison content on your website. AI models will reference “best X” lists regardless of who published them, as long as the content is well-structured and directly answers the query.
Is Mention Building More Important Than Link Building?
For AI visibility specifically, yes. Brand mentions appear to carry more weight than backlinks in how AI models decide what to recommend.
BrightEdge found that ChatGPT mentions brands 3.2 times more often than it links to them. Separately, analysis from RankScience shows that brand mentions correlate roughly three times more strongly with AI visibility than backlinks.
Getting a mention is also much easier than earning a link. A backlink requires someone to edit their page and add a hyperlink. A mention just requires your brand name to appear in text on a relevant, cited page. The barrier is lower and the impact on AI visibility is higher.
At TJ Digital, we still build links because they help with traditional Google rankings. But mention building is where we see the highest return for AI search visibility.
What Does an Effective Mention Strategy Look Like?
Effective mention building is targeted, not scattered. You are not trying to get your name everywhere. You are trying to get it on the right pages, in the right context.
In practice, this means:
- Get reviewed on the 1-3 directories AI models actually cite in your industry. Skip the rest until these are covered.
- Engage in the specific forum threads that appear in your citation data. Not random threads. The exact ones your prompt tracking identified.
- Reach out to authors of highly cited articles. Offer updated data, an expert quote, or a small fee in exchange for inclusion.
- Create comparison and “best of” content on your own website. AI models cite well-structured lists regardless of who published them.
- Keep existing mentions current. 79% of AI-cited listicles were recently updated. If you are in an article that has not been touched in over a year, reach out to the author with fresh information.
A few well-placed mentions on frequently cited sources will outperform hundreds of mentions on pages that AI models never look at.
How Stable Are AI Citation Sources?
These AI models are still changing fast. We have seen sources that dominated one month drop significantly after a model update. For example, Reddit has historically been one of the most-cited sources across industries. But in September 2025, we tracked ChatGPT cutting its Reddit citations by roughly half across almost all of our 30 client accounts.
Even after that drop, Reddit remained the most-cited source overall. No single site replaced it. Reddit just lost share across the board.
The core strategy holds up despite this volatility. Monitor which sources AI models are citing, keep your brand mentioned on those sources, and expect the specific mix to shift over time. The businesses that track and adapt will stay visible. The ones that set it and forget it will not.
Start With Your Hit List
If you take one thing from this article: do not try to be everywhere. Identify the three to five sources AI models actually reference in your industry. Focus your mention building energy there.
Run the prompts. Track the citations. Build your hit list. Then get mentioned on those specific pages in a way that makes the model confident you are the best choice.
Very few businesses are doing this right now, which means the opportunity is wide open for those who start.
If you want help identifying the right sources and building a mention strategy for your business, request a free audit from TJ Digital. We will show you exactly where your brand needs to show up and how to get there.