ChatGPT favours its paid publisher partners, Ahrefs finds

Axel Springer signed a licensing deal with OpenAI in December 2023. Since then, Bild and Business Insider have shown up in ChatGPT citations at rates most publishers would kill for. Per Ahrefs, which tracked citation frequency across every known OpenAI publisher partnership, the pattern holds across the roughly two dozen deals OpenAI has struck since 2023: licensed publishers get cited substantially more often than their traffic or domain authority would predict. The mechanism is not mysterious. OpenAI pays these outlets for the right to surface their content directly in ChatGPT answers, complete with attribution and a link. The licensing fee buys placement. It does not buy quality, and Ahrefs' data makes no claim that it does.

This matters more than it sounds, because it quietly overturns the working assumption that has governed most brand strategy for LLM visibility over the past two years: that citation is a meritocracy of content quality, structured data, and topical authority. Ahrefs' findings say otherwise. A chunk of what ChatGPT cites is not earned. It is bought.

The commercial layer nobody priced in

SEO practitioners have spent two years reverse-engineering what makes a page citable: clear answers near the top, schema markup, strong backlink profiles, Reddit mentions, Wikipedia presence. All of that remains true and worth doing. But Ahrefs' data on publisher partnerships introduces a variable that no amount of on-page optimisation can touch: whether your organisation, or your industry's trade press, has a commercial licensing relationship with OpenAI. News Corp, the Financial Times, Axel Springer, Condé Nast, and a growing list of others occupy a privileged citation tier that has nothing to do with how well their content answers the query at hand.

For financial services firms and multilateral institutions, the implication runs in two directions. First, the obvious one: if the Financial Times has a licensing deal and your central bank's research department does not, the FT's framing of monetary policy is structurally more likely to surface in ChatGPT's answer than your own, regardless of which is more rigorous. That is a visibility gap money closes and expertise alone does not. Second, and less obvious: the publishers your organisation has spent years cultivating for earned media now carry a second, OpenAI-specific value. A placement in a licensed outlet is not just reputational anymore. It is an LLM-citation asset with a measurable, if opaque, multiplier attached.

Why this is not a passing anomaly

Skeptics will note that licensing deals cover a small fraction of the open web, and that ChatGPT still draws most of its citations from unlicensed sources via retrieval. Fair. But the direction of travel matters more than the current share. OpenAI has been adding publisher partners steadily since the Associated Press deal in July 2023, and competitors are not standing still: Perplexity has its own publisher program, and Google's AI Overviews sit inside a search index that already has commercial relationships baked into News Showcase. The trend across every major AI answer engine is toward formalised, paid content relationships, not away from them. Expecting this to stay a rounding error is wishful thinking.