Editorial links and AI search are becoming important parts of modern SEO. As search engines and AI-powered platforms change how people discover information, businesses need to focus on trustworthy content and credible sources. Earning links from reputable publications can strengthen website authority, improve search visibility, and help readers discover your expertise.

However, successful link building requires more than collecting backlinks. You need relevant content, genuine editorial relationships, and a clear digital marketing strategy. By combining editorial link building with AI search optimization, you can improve your brand’s online presence and create useful content that deserves attention. These efforts can support brand credibility and long-term organic growth.

First, what do I mean by editorial and sponsored?

An editorial link is one a writer or editor added because the thing was worth mentioning. A journalist quotes your data. A reviewer includes your tool in a roundup. Someone on Reddit says “just use this one.”

A sponsored link is one where money or a deal changed hands: advertorials, paid guest posts, “partner content,” and affiliate-heavy articles.

Google already draws this line. Its guidance says paid and sponsored links are fine as long as they carry a rel=”nofollow” or rel=”sponsored” attribute, according to a PR/SEO write-up summarizing the rules. AI engines have no tidy rulebook like that, so you have to look at the pattern of what they cite.

The number that made me stop and rethink

The study everyone in this space is quoting is from Muck Rack. In May 2026, it looked at more than 25 million links cited by ChatGPT, Claude and Gemini. Earned media made up 84% of AI citations, and paid or advertorial content made up just 0.3%.

That’s not a typo. Paid placements were a rounding error.

A few honest caveats before you run off and cancel everything:

  • This comes from a PR-industry company, and PR companies like earned media.
  • The result depends on how “paid” was classified. Advertorials are easy to spot, but quiet sponsored deals often aren’t labeled.
  • Other studies report different percentages, though they point the same direction.

One more number I find useful: branded web mentions correlated with AI visibility at roughly 0.66, while raw backlink acquisition like guest posts and generic directory links came in at only 0.218. Correlation isn’t proof. But it suggests that being talked about matters more than being linked to.

What each tool seems to cite

The engines don’t agree with each other. Across 680 million citations analyzed by Averi, only 11% of domains were cited by both ChatGPT and Perplexity. So “what AI cites” is really three or four different questions.

ChatGPT leans on encyclopedic and established sources. Wikipedia alone is roughly 16% of ChatGPT citations in Ahrefs’ study of 78.6 million queries. Directories and review sites also show up a lot. In a Yext study, ChatGPT sent 44.7% of its citations to directories and review platforms for local-business style queries. One more detail: ChatGPT mentions things more than it links to them. An AirOps study found it mentions brands 3.2 times more than it cites them. So you can be “named” without getting a clickable link.

Perplexity is the most visible about its sources, with numbered citations right in the answer. It also loves community content and fresh pages. The Reddit numbers vary wildly by study. Some say 46.7% of its citations point to Reddit, while another analysis puts Reddit closer to 20–24% of an answer’s citations. I wouldn’t trust any single figure, but “Reddit matters a lot here” holds up across all of them.

Gemini behaves differently again. Yext found 52.15% of Gemini chat citations came from brand-owned websites, the highest of any major engine. If that’s right, your own site’s quality matters more with Gemini than with the others.

Last point on this: don’t assume Google rankings carry over. One roundup notes that Ahrefs found only 12% of URLs cited by assistants like ChatGPT, Gemini and Copilot rank in Google’s top 10. Perplexity is the exception at around 29%.

How I check this myself (step by step)

You don’t need a paid tracking tool to start. I’d do this:

  1. Write 10 real questions. Use the questions your customers or readers actually ask, not marketing-speak. “Best budget mechanical keyboard for coding” works. “Top innovative keyboard solutions” doesn’t.
  2. Run each one in ChatGPT, Perplexity and Gemini. Use a fresh chat each time, with search or browsing on.
  3. Log every cited domain in a spreadsheet. Columns: question, engine, domain, and whether it looks editorial, sponsored, community, brand-owned or directory.
  4. Check your own name. Were you cited, only mentioned, or absent?
  5. Click through the sources. Is the cited page a real article, or an obvious “sponsored post”? Actually reading them is where you’ll learn the most.
  6. Repeat in two weeks. Results move around. One study found only 30% of brands keep consistent AI visibility from one answer to the next. A single snapshot tells you little.

If you want tooling later, trackers like Profound and LLM Pulse exist for this, but a spreadsheet is enough to see the pattern.

What I’d do with this information

Treat paid placements as a different budget line. They may help with direct traffic, brand awareness and referral clicks. Just don’t count them as AI visibility work. That’s the one clean takeaway from the 0.3% figure.

Go after real mentions. That means original data, a useful free tool, a clear opinion, or a genuinely good product that reviewers want to include. One analysis of the Muck Rack data adds that journalism alone accounted for 27% of cited sources. Being useful to journalists beats paying them.

Show up where people talk. Real answers on Reddit, Stack Overflow and niche forums matter, especially for Perplexity. Help first. People spot a disguised pitch in about two seconds.

Fix your own site. Since Gemini leans on brand-owned pages, clear about, product and FAQ pages are worth the effort. Keep them accurate and crawlable. One source claims 44.2% of LLM citations come from the first 30% of a page’s content, so put the answer near the top.

Label your sponsored links properly. Use rel=”sponsored” and disclose the deal. It keeps you inside Google’s guidelines, and it’s honest with readers.

Mistakes I see (and have made)

  • Chasing one engine. Optimizing only for ChatGPT while ignoring Perplexity, or the other way around. With 11% overlap, that leaves most of the map empty.
  • Judging by one prompt. Answers vary run to run. Ten prompts, two weeks apart, tells you far more.
  • Buying links because the domain score looks good. Domain score is a Google-era metric. It doesn’t tell you whether an AI tool will ever cite that page.
  • Confusing a mention with a citation. Being named in an answer and being linked are different wins. Track both.
  • Believing every stat. Most of these studies are from agencies selling GEO or PR services. Use them as direction, not gospel, and compare more than one source.
  • Hiding sponsorship. Disguising paid content as editorial can get you in trouble with Google and, honestly, with your readers.

Where I land on this

The boring answer is the useful one. AI tools seem to reward things people actually chose to talk about, and a purchased link is the opposite of that. I’d rather spend a month making one genuinely useful page or dataset than buy five placements.

Test it yourself before taking my word for it. Ten questions, three engines, one spreadsheet. The gaps you find will point at your next move better than any study can.

Sources: