Do AI Citations Come From Mentions, Not Your Site?

The data points to a different playbook for AI visibility, where the sources that mention you may matter more than the content you control. I ran two structured experiments back to back, tracked the results manually, and documented what worked, what failed, and what changed between the two tests.

What Happened

Search Engine Land published two back-to-back generative engine optimization experiments by Zeeshan Yaseen on September 14, 2026. The tests tracked 15 commercial-intent keywords across ChatGPT, Claude, Gemini, and Perplexity, logging 775 citation events. Listicles drove 72.4% of citations and PR another 24.1%, while owned content, guest posts, and LinkedIn combined for less than 4%. Roughly half of cited sources stopped being cited within 30 days.

The Numbers in Context

In traditional SEO benchmarks, owned content typically drives 40 to 60% of organic visibility. In these GEO experiments, owned site content contributed a low single-digit share of AI citations. A single Indeed SEO listicle generated 190 of the brand's citations, more than every other source combined. That is a near inversion of the owned-versus-earned ratio you have built your content programs around.

Why This Matters for B2B Marketing Leaders in HR Tech and FinTech

Your content team has spent a decade optimizing pillar pages, product content, and thought pieces on your own domain. These experiments suggest that playbook underperforms in AI answer engines. When a buyer asks ChatGPT for the best applicant tracking system or the top embedded finance platforms, the model pulls from third-party listicles and PR wires, not your solution page. If you are a mid-market HR Tech or FinTech brand fighting for consideration against category leaders, this changes your budget allocation. PR, analyst relations, and placement on the specific listicles LLMs already cite likely deserve a larger share of demand generation spend than your current content calendar reflects.

The Starr Conspiracy's Take

This matches what we have seen advising HR Tech and FinTech clients through the shift from search to answer engines. Owning the answer used to mean owning the page. Now it means being named on the pages the model trusts. The experiments also expose a decay problem you should plan around: half of cited sources drop off within 30 days, so a one-and-done PR hit will not hold. You need a sustained placement program targeting the specific publications each model favors for your category, refreshed quarterly. Treat AI visibility as an earned media discipline with an SEO measurement layer, not a content marketing tactic.

What to Watch Next

Expect analyst firms and category-review sites to price placements based on LLM citation frequency within the next 12 months. Watch whether Gartner, G2, and vertical publications like HR Executive begin publishing citation-share data. If they do, media buying in B2B will restructure around it.

Related Questions

Should HR Tech brands cut owned content investment?

No, but rebalance it. Owned content still supports sales enablement, product education, and direct search. Redirect a portion of net-new content spend toward earning placements in the listicles and analyst pieces LLMs cite for your category keywords.

How do you find which sources LLMs cite for your category?

Run your top 15 commercial-intent queries manually across ChatGPT, Claude, Gemini, and Perplexity. Log every cited URL. Patterns emerge within 50 to 75 queries. Our AI visibility measurement framework walks through the tracking method in detail.

How fast does AI citation visibility decay?

The experiments showed roughly half of sources stopped being cited within 30 days. Plan for quarterly refresh cycles on PR placements and listicle updates rather than treating any single win as durable.