How Long Does Earned Media Remain Useful in AI-Generated Answers?

What does the evidence show about recency?

• Most cited journalism is recent. Muck Rack's May 2026 analysis found 57% of journalism citations with known dates were less than a year old. Its December 2025 report found half of all citations came from the previous 11 months, so the bias towards recent coverage had slightly increased.
• The curve is steep, then flat. Citations peaked in the first month after publication, fell quickly through month six, and then flattened into a long tail.
• Old coverage still counts. Articles more than three years old still made up a meaningful share of citations.
• Question type matters. Muck Rack found questions about industry trends cited journalism in 46% of responses, more than twice the rate of how-to and comparative questions.

Freshness shows up beyond journalism too. Seer Interactive saw the same tilt in its 2026 study of pages cited by ChatGPT, Gemini and Perplexity: roughly 75% had been updated in the preceding year, and 88% within two years.

Why do some pieces of coverage last longer than others?

The pattern follows the question people ask. Coverage that answers a question people keep asking stays useful. Coverage tied to a single moment fades with it.

How does coverage persist in model knowledge?

Search-based answers favour recent pages, but models also answer from what they learned in training. Coverage that was widely published before a model's training data was collected can shape how that model describes you. That applies especially when the model answers without searching. That influence lasts until the next model version, and sometimes beyond, if the same facts are repeated in newer sources. It cuts both ways. Positive coverage from two years ago may still colour descriptions of you, and so can an old negative story. How to tell which source an answer used is covered in how to tell whether an AI answer used retrieval or training data.

How should PR plans account for this?

• Aim for a steady flow, such as coverage every month or quarter, rather than one big launch spike that fades within six months.
• Target evergreen formats: explainers, buying guides and "how to choose" pieces where your expert can be quoted, and annual rankings you can re-enter.
• Create original research on a regular schedule, such as an annual survey. Each edition refreshes the citations, and the series builds a reputation.
• Keep your own pages current, so when coverage links to you, the page it points to still says the right things.
• Refresh relationships with publishers whose guides you appear in, so updates include you with current facts.

How long it takes to shift negative sentiment in answers is covered in how long it takes to reverse negative brand sentiment in AI answers.

Does the outlet affect how long coverage lasts?

Somewhat. Muck Rack found journalism citations spread across thousands of outlets rather than concentrated in a few. Publications that maintain evergreen sections, such as buying guides, explainers, glossaries and annual lists, give coverage a longer life, because they update pages instead of replacing them. Trade publications in narrow fields often stay relevant longer than general news sites, since fewer sources cover the same specialist questions. Crawler access also matters. Some large news sites block AI crawlers such as OpenAI's OAI-SearchBot or Google-Extended, which makes their coverage of you less likely to appear in ChatGPT, Claude, Perplexity or the Gemini app, though a block does not always mean no citations. Google's AI Overviews and AI Mode can still use those pages, because they rely on Googlebot.

What should you do when coverage starts to fade?

Treat the fade as a prompt, not a loss. Turn the substance of the coverage into a lasting page on your own site, such as the data, the explanation or the customer story, and link to the original article. Offer the journalist a follow-up when you have new results. Update any figures the coverage quoted if they have changed. And plan the next piece, so the flow continues before the previous one disappears from answers.

How do you measure the lifespan of your own coverage?

1. Log each piece of coverage with its date, outlet, format and the topics it covers.
2. Map it to questions in your tracking set, the ones it could plausibly answer.
3. Record citations monthly: which pieces are cited, for which questions, in which assistants.
4. Chart citations by age of the coverage to see your own curve.
5. Compare formats: do guides and research last longer than news mentions for you?

After a few quarters, you will know which kinds of coverage are worth pursuing for AI visibility in your category, and when to expect them to fade.

What might an illustrative case look like?

Consider an illustrative cybersecurity company that earned wide coverage for a product launch in January. AI answers cited the launch articles for about two months, then stopped. Its March quote in a trade publication's explainer on ransomware insurance, by contrast, was still cited a year later. The explainer kept answering the same question and the publisher updated it twice. The company shifted its PR plan towards quarterly contributions to explainers and an annual threat report. Two years in, the report's figures were being quoted in answers through dozens of articles that cited them.

What do people ask about earned media and AI answers?

Should we still pursue news coverage if it fades quickly?

Yes, for reach and reputation. Just do not count on it for long-term AI visibility.

Does updating an old article extend its life?

When publishers update evergreen pieces, citations often continue. Offer fresh data or comment to help that happen.

Is old negative coverage permanent?

It fades from search-based answers as newer sources appear, but can linger in model knowledge. Newer, accurate coverage helps displace it.

Which assistants rely most on recent coverage?

Search-based answers in all major assistants favour recent sources. Answers from model knowledge reflect older data.

How many pieces of coverage a year are enough?

It depends on your category, but a regular rhythm matters more than volume. One solid piece a month, spread across news, expert quotes and evergreen guides, avoids the gap that opens when a single burst fades after a few months.