Content freshness is how recently a page was published or substantively updated. Ahrefs found that AI assistants cite newer pages than Google’s organic results do. The gap is real but modest, and the same study warns against treating frequent edits as a strategy.
- Ahrefs’ July 2025 analysis of about 17 million citations found AI-cited URLs averaged 1,064 days since publication, against 1,432 days for organic Google results (about 25.7% fresher).
- Since last update, the gap was smaller: 909 days for AI assistants against 1,047 for organic results, about 13.1% fresher.
- Cited content still averaged roughly 2.9 years old, so long-lived pages remain heavily cited.
- Ahrefs says most brands may get better returns from new, high-quality content than from frequent refreshes, and Google’s John Mueller has warned against changing dates without changing content.
What the Ahrefs study measured
The question behind the study is a budget question: should a team spend hours refreshing old pages in the hope of earning more AI citations? Ahrefs tested it against data instead of opinion.
In an analysis published July 28, 2025, Ryan Law of Ahrefs examined about 16.975 million cited URLs, described as 17 million citations, drawn from Ahrefs’ Brand Radar tool. The sources were ChatGPT, Perplexity, Gemini, Copilot, Google AI Overviews and organic Google results. Page age was measured from the date Ahrefs’ crawler first saw the page, and last-updated age from the last published date recorded in its Content Explorer index.
The headline result was that AI assistants cited pages averaging 1,064 days since publication, compared with 1,432 days for URLs in organic results. Ahrefs presented the difference as about 25.7% fresher for assistants.
How platforms differ
The aggregate hides a spread between systems, and that spread matters because Google still handles most searches. In Ahrefs’ figures, ChatGPT and Copilot cited the newest pages, Perplexity and Gemini sat in the middle, and Google’s organic results and AI Overviews leaned toward older pages.
| Source | Average days since publication |
|---|---|
| ChatGPT (references) | 1,023 |
| Copilot | 1,056 |
| Gemini | 1,118 |
| Perplexity (in-text order) | 1,166 |
| Organic Google results | About 1,432 |
Ahrefs also reported that Perplexity and ChatGPT tend to order their in-text references from newest to oldest. Ahrefs states that Google is the least influenced by freshness. The practical reading is that a refresh program helps most where your buyers use assistants, and least for classic Google rankings.
What the data does not show
This is a descriptive study, not an experiment. It shows which pages assistants cited, not that updating a page caused a citation. Newer pages may simply cover newer topics, which assistants are asked about more often.
Ahrefs lists its own caveats. Cited content still averaged about 2.9 years old, which means assistants continue to lean on durable pages. Freshness is one of many factors, and frequently updated low-quality content will not help much. The article also notes that Google’s John Mueller has warned against changing a publish date without changing the page itself.
The sample also describes citations in aggregate, not outcomes for any single site. We found no independent replication in this review, so treat the figures as one well-documented dataset from a vendor that sells visibility tracking.
Where this changes a content budget
The decision-relevant finding is the authors’ own conclusion: for most brands, new, high-quality content may return more than frequent updates, given the opportunity cost. Refreshing is justified where a page makes claims that expire, such as pricing, statistics, product capabilities, regulations or dated recommendations.
That points to a triage approach. Pages whose facts decay deserve a scheduled review. Evergreen explainers can be left alone until something material changes. A visible “last reviewed” date is useful to readers, but only if the review was real. Our guides to llms.txt and AI citations and schema markup for AI search apply the same evidence-first test to other popular tactics.
A practical refresh workflow
The sequence below keeps the effort tied to measurable change instead of cosmetic edits.
- Export your pages that already earn traffic or appear in AI answers, and mark which contain time-sensitive facts.
- Check each flagged claim against its primary source and replace anything outdated, with the source and year named in the sentence.
- Add genuinely new material, such as a changed comparison, a new data point or a corrected recommendation, rather than rewording.
- Update the displayed date only when the substance changed.
- Re-test the buying questions in ChatGPT and Perplexity after a few weeks, using the approach in our guide to measuring AI search visibility.
Frequently Asked Questions
Should I update the publish date on every page?
No. Ahrefs notes that Google’s John Mueller has warned against changing publish dates without changing content. Update dates only when the page has materially changed.
Do older pages stop being cited by AI assistants?
No. Ahrefs found that cited content still averaged about 2.9 years old, so assistants continue to cite long-lived pages.
Is a refresh better than publishing something new?
Ahrefs suggests most brands may see better returns from new, high-quality content. Refreshes pay off mainly on pages with claims that expire.
Deciding which pages deserve a refresh and which deserve a replacement is an editorial and measurement problem together. At Grow with Hafsa we build AI marketing and content systems that keep that process evidence-led, and we are glad to talk through your content library.