Off-site visibility

When an engine picks products to name, it weighs what many sources say about them. A brand that only praises itself on its own site gives the engine one voice. A brand named across lists, reviews and articles gives it several that agree.

Depth beats spread. Kevin Indig found that engines rebuild their trusted sources topic by topic, so a few placements on the top sources for your topic outweigh many scattered ones (Growth Memo, June 2026). Eli Schwartz adds a warning. Placed mentions that describe a brand better than reality get outweighed by everything else written about it (Product-Led SEO, July 2026).

Your tracking tool shows which third-party pages engines cite for the prompt set from chapter 5. Use this export as your initial outreach list. These pages drive results because engines already pull information from them. The goal is a brand mention on those pages. A mention counts even without a link, so track mentions apart from SEO links.

Best-of lists answer the "best X for Y" questions where engines name products. That makes a spot on a cited list the strongest placement you can win. Brands reach those spots three ways, each with its own trade-offs.

Our own list study shows why lists earn this effort. In Q3 2026, listicles on independent sites rose from 11.6% to 12.4% of citations across our clients' categories. When a vendor's list was cited, the vendor was named in 45.4% of answers, against 16.3% on the same questions without it. Somewhat known vendors gained the most, from 12.5% to 54.8%, and dominant brands gained the least.

Review platforms such as G2, Capterra and Trustpilot shape sentiment, because they carry buyers' own experience. They also get cited by several engines at once, which is rare. Omnia data published by Kevin Indig showed that only 2.37% of cited URLs appeared in all three of ChatGPT, Perplexity and AI Overviews, and review platforms were among the few sources that did (Growth Memo, May 2026). A strong review profile can help you in every engine, while most other pages help in one. In our list study, review sites and directories rose from 5.8% to 7.4% of citations between Q2 and Q3 2026.

Seer reported that brands without a Trustpilot profile had a 1% median AI citation rate, against 53.5% for brands with 1 to 13 reviews (2026, 1,926 brands). Brands with a review profile also tend to be better known, which helps them get cited too. Even if just correlational, the pattern still suggests a small base of honest reviews is worth having.

We tell clients to focus on LinkedIn and YouTube before Reddit. Brands control what they publish on the first two. Reddit belongs to its communities. How often engines cite Reddit also changes sharply. Otterly found Reddit's share of ChatGPT citations fell from 4.42% to 0.39% after August 14, 2026.

Press coverage and podcasts help when they are inside your category, where engines look for answers about your product. Coverage in general business press does less for a specific software question. Eli Schwartz argues that PR now shapes how models see a brand. Being named, he says, needs a credible third party (Product-Led SEO, September 2026).

People count as well as publications. Indig recommends bylined articles from your subject experts and data others can embed (Growth Memo, June 2026). Original research gives journalists and list writers a reason to name you.

Press releases do very little, if anything, to drive AI visibility. I have never seen a press release in a list of AI citations, unless the question is specifically about the announcement.

Wikipedia ranks among the sources engines cite heavily. Similarweb found it was ChatGPT's top cited domain at 13.15% of citations and second in Google AI Mode at 5.21% (U.S. prompts, January to February 2026). Profound reported it at 7.8% of ChatGPT citations over August 2024 to June 2025.

Wikipedia's own rules limit what a company can do there. Its conflict of interest guideline says paid editors must disclose who pays them. It also strongly discourages them from editing articles about their employer directly. Companies should propose changes on the article's talk page, where volunteer editors decide. The guideline also asks editors with a conflict of interest to send new articles through the Articles for Creation review process.

Correcting the record across the web is a long-term project. Start on your own site, as chapter 6 covers, since third parties often copy from it. Then run the branded prompts from chapter 5 and open the sources behind each wrong answer. Those pages are the ones to ask for an update, starting with review sites and the lists engines cite often.

Eli Schwartz suggests starting from an audit of what the internet already says about you, then closing the distance between that and what is true (July 2026).