A single credible press mention no longer spikes and decays. When large language models ingest earned media, it keeps shaping what AI says about your brand for years. That gives good coverage an almost infinite half-life and makes promotional noise more worthless than ever.
Earned media used to fade. Now it feeds the models that answer your customers' questions for years.
Here is a question that used to have a clean answer. When do you stop measuring the return on a piece of PR? The rise of AI answers has blown the answer up.
Writing in Mumbrella, the argument is that earned media now has an almost infinite half-life. Traditional ROI assumes a cutoff. A campaign runs, the coverage lands, the traffic spikes and fades, and at some point you close the books. Large language models break that model. When a press mention, a review or an expert quote gets ingested by the systems that now answer buyer questions, that coverage keeps working. It shapes what the AI says about you, long after the news cycle moved on.
A single credible mention is no longer a spike that decays. It becomes training data that feeds the machine every time someone asks it about your category.
Why it matters
For years, earned media was hard to value because its effect was diffuse and short-lived. The AI shift flips that. The coverage you earn today influences the answers customers get tomorrow, next quarter and beyond, because the models keep drawing on it. That makes credible, factual, widely-cited coverage more valuable than it has ever been, and it makes thin promotional noise more worthless. The machine repeats what the credible sources said, not what your ad claimed.
Share of Google searches that end without a click, making the AI's answer the first thing many customers read (TechCrunch)
What to do about it
The news cycle still moves on. The models do not forget. Earn the coverage that will still be working for you when the headline is old news.