Skip to content
Reckomenda
en
Log in
Start free
GEO Myth-busting Freshness

The Freshness Myth in AI Search: What the Data Shows

"Publish in the last 30 days or you won't get cited" is a myth. AI engines cite established pages all the time. Here's what the data says and when freshness truly matters.

By The Reckomenda team 2 min read
A low-angle view of a dark modern building facade with layered, overlapping horizontal planes of glass and metal.
Photo: Anders Jildén / Unsplash
On this page

One of the stickiest claims in AI-search advice is that content has to be brand new to get cited — “publish in the last 30 days or you’re invisible,” “content under a month old gets 3× the citations,” “Perplexity weights freshness at 40%.” These numbers get repeated confidently and sourced to no one. The data tells a different, more reassuring story: AI engines cite established pages all the time, and freshness matters far less — and far more selectively — than the folklore claims.

Does publishing recently get you cited more?

Not as a general rule. When Ahrefs analyzed 1.4 million prompts to see what ChatGPT cites, they found the median cited page is around 500 days old (Ahrefs, 2025). Established, well-linked pages are frequently cited more than brand-new ones. If recency were the dominant signal, that median would be measured in weeks, not well over a year.

The takeaway: a good page doesn’t expire in 30 days. Age and accumulated authority are often working for you, not against you.

What the data actually shows

Freshness is a real but modest and situational signal, not a master dial:

  • For most informational and conceptual content, depth, clarity and citations beat recency. A three-year-old explainer that answers the question well out-cites a shallow new one.
  • The precise freshness “weights” that circulate (“40% for Perplexity,” “1-year half-life”) are unsourced or traced back to vague “syntheses” with no underlying study. Treat them as invented.
  • What is true: engines prefer current information for genuinely time-sensitive questions — pricing, product comparisons, “latest version of X,” news. There, an out-of-date page is a real liability.

When freshness genuinely matters

Spend your freshness effort where it’s earned:

  • Time-sensitive pages — pricing, plan comparisons, anything that describes fast-changing product or engine behavior. Keep these genuinely current.
  • A recurring data piece — a quarterly report with new numbers legitimately earns a fresh dateModified each cycle and gives engines a reason to re-fetch.
  • Real substantive updates — when facts change, update the page and its “last updated” date. Don’t flip the date to fake freshness; that fools no one and helps nothing.

For an evergreen conceptual post like what GEO is, obsessively re-dating it does nothing. Improving its substance does.

What to do instead

Stop chasing a publish-date signal and invest in the things the evidence actually supports: fact density, clear answer-first structure, credible sources, and off-site reputation (see why AI cites Reddit). Add a visible “last updated” date as honest hygiene, refresh time-sensitive pages on a real cadence, and leave your good evergreen content alone to accumulate authority.

And measure it rather than guessing. Whether an AI cites your page — new or two years old — is something you can check directly. The free audit below runs your questions across ChatGPT, Claude, Gemini and Perplexity so you can see what actually gets cited, no credit card required.

Keep reading