Content Freshness: Why AI Forgets Old Pages

Content freshness refers to a page’s original publish date and how recently it was last updated. Classic Google search uses this as one ranking signal, but AI search engines — ChatGPT, Perplexity, Gemini, and the like — weight it far more heavily. The reason is simple: when someone asks an AI a question, they generally expect an answer that’s true “right now,” not a price or a statistic from two years ago. In this piece I explain why freshness has become this critical, and how to maintain it.

This page is part of my GEO Guide. For the basics, start with What Is GEO, How Is It Different From SEO; if you’re curious how AI search is affecting your organic traffic, Click Loss in AI Search might interest you.

Why AI Cares So Much About Freshness

An AI model draws on two sources. The first is training data — frozen at a certain point in time, everything the model “knows” up to that cutoff. The second is real-time crawling — the model going out to the web right before answering and reading current pages.

Systems that do real-time crawling — Perplexity, ChatGPT’s search mode, Google AI Overviews — use a page’s last-update date directly as a trust signal. A pricing page or a compiled statistic written two years ago gets deprioritized if a more current alternative exists — even if the page is technically still accurate. In classic Google, that used to mean dropping two or three spots. In AI search, it can mean never making it into the answer at all.

Which Content Types Are More Freshness-Sensitive

Not every page is equally exposed. Here’s the pattern that shows up in the field:

• Pages with pricing information — if a page shows a service’s current cost, it becomes inaccurate within a few months.
• Pages sharing statistics/data — numbers like “X% of users do this” age fast, and AI notices.
• Titles with a year, like “Best X in 2024” — if the title has a date in it, AI takes that date seriously, and relevance drops as the year goes by.
• Fast-moving tech/tool comparisons — a “best AI tool” list from a year ago is very likely incomplete today.

By contrast, timeless, evergreen definition pages — content that explains a concept without numbers or dates — carry much less freshness pressure. This very page is one of them: the definition of “what is content freshness” will still hold years from now, but the examples and methods inside it need to be updated over time.

How to Keep Content Fresh

I follow a four-step method:

1. Add a visible “last updated: [date]” note. Near the top of the page, somewhere the reader sees it right away. It signals to both the reader and to crawling systems.
2. Check old statistics periodically. Set a calendar reminder for every page that contains numbers — every 3 to 6 months, depending on the industry.
3. Fill in the “modified date” schema field correctly. A visible date alone isn’t enough — the dateModified field in your JSON-LD needs to match the real update date, or AI systems end up trusting the structured data over the visible date and get it wrong.
4. Keep the page “alive” with small additions and fixes, even without a major overhaul. A new example, an updated number, a corrected sentence — all of these produce a freshness signal.

Something I Saw in the Field

On one of the sites I regularly work on, there was a price comparison page written three years earlier — the numbers were still roughly accurate, but the page had no date information anywhere. We added a visible “last updated” note, swapped a few numbers for real current values, and fixed the dateModified field — barely touching the rest of the content. Within a few weeks, that page started appearing as a cited source in an AI search result it had never shown up in before. The page itself was nearly identical; the only thing that changed was the freshness signal.

You Can Ask Me to Do This

Tracking content freshness — picking the right pages and updating them regularly — is time-consuming but not technical work. You can hire me for this: usually a remote, hourly job that runs anywhere from 2 hours to 2 weeks. We can work out together which pages on your site carry freshness risk and set up an update calendar. For my full AI search visibility work, take a look at the AI Search Visibility service page.

Frequently Asked Questions

Is content freshness an SEO ranking factor, or is it AI-search-specific?

Freshness has long been a signal in classic Google search too — especially for “query deserves freshness” topics. But its effect is bigger in AI search, because the model produces a single direct answer, and freshness plays into which source it picks for that answer.

How often do I need to update each page?

There’s no fixed rule — it depends on the type of page. For pages with pricing and statistics, 3–6 months is reasonable. For timeless definition pages, a once-a-year review is usually enough.

Is it enough to just change the visible date?

No. Changing the visible date while leaving the content unchanged misleads both readers and AI systems, and creates a noticeable inconsistency. Both the visible date and the dateModified field in the page’s schema need to be updated alongside a real change.

How do AI search engines even know a page’s update date?

From several sources: the page’s visible text, HTML meta tags, the dateModified field in structured data, and the server’s Last-Modified HTTP header. Having these agree with each other increases trust.