**AI engines strongly favour fresh content — and to benefit, you have to both *be* fresh and *signal* it.** This check looks for published and modified dates exposed in machine-readable form (`datePublished`/`dateModified` in schema, `article:modified_time` meta, plus a visible date) and whether the content is reasonably current. A genuinely updated page that doesn't expose a date gets no credit for it.
It checks both that you expose dates in a form machines can read and that the content itself is reasonably current. Specifically:
- Machine-readable dates — datePublished and dateModified in Article schema, or article:modified_time meta tags.
- A visible date — a human-readable "published" or "updated" date on the page.
- Actual currency — content that reflects current facts, not a recent date stamped on stale information.
A machine-readable date plus reasonably fresh content passes; a date that's only human-readable, or present but on stale content, is a warning; no machine-readable date is a fail.
GEObubbly looks for datePublished/dateModified in structured data, article:modified_time meta, or a visible date, and weighs it against signs of currency. It's a core GEO / LLM Readiness check worth 3 points, because freshness is one of the signals answer engines weight most heavily when choosing which sources to cite.
AI platforms favour recent content, often dramatically. Analyses of AI citations have found that fresher pages can be cited far more than older ones, especially for commercial and fast-moving topics. To benefit, you have to do two things: actually be fresh, and signal it in a way machines can read. Expose your dates in machine-readable form — datePublished and dateModified in Article schema, or article:modified_time meta — plus a visible date on the page; a page that was genuinely updated but doesn't expose a date gets no credit for it. And keep the content itself current: refresh stats, dates and claims, because a machine-readable "updated" date on stale content is worse than nothing if the content contradicts what's now true. Freshness compounds with the rest of GEO / LLM Readiness — a regularly updated, dated, well-structured page is exactly what an answer engine reaches for when assembling a current answer.
A lot, for many topics. Analyses of AI citations have found that fresher pages are cited dramatically more than older ones, especially for commercial, news and fast-moving subjects where currency matters. Answer engines aim to give up-to-date answers, so they lean toward sources that appear recently published or updated. The effect isn't uniform — evergreen topics are less time-sensitive — but for anything where facts change, keeping content fresh and signalling that freshness clearly is one of the higher-impact things you can do for AI visibility.
Machine-readable dates are publish and update dates exposed in a structured form engines can parse, rather than only as text a human reads. The main ways are datePublished and dateModified inside Article (or similar) JSON-LD schema, and the article:modified_time Open Graph meta tag. Add these in your page's <head> or structured data, and also show a visible "Updated [date]" on the page. That way both engines and readers can see when the content was last refreshed.
No — and it can backfire. Stamping a fresh dateModified on content you haven't actually updated is a hollow signal: if the page still contains outdated facts, it now openly contradicts its own "updated" date, which erodes trust with both engines and readers. Freshness only helps when the content is genuinely current. The right approach is to periodically review and refresh the substance — update statistics, dates, examples and claims — and then update the date to reflect that real revision.
It depends on the topic. Fast-moving, commercial or news-style content benefits from frequent review — monthly or whenever the underlying facts change — while evergreen explainers can be refreshed less often, perhaps a couple of times a year, mainly to keep figures, examples and dates current. The signal that matters is genuine currency, not a fixed cadence: review on a schedule that fits how quickly your subject changes, update the substance when needed, and expose the revised date so engines and readers can see the content is maintained.
Ideally all three, because they reinforce each other. Put datePublished and dateModified in your structured data so engines can read them reliably; add article:modified_time in Open Graph meta for surfaces that use it; and show a visible "Published / Updated" date on the page for human readers. Keep the three consistent with each other and with the actual state of the content. Consistent, machine-readable, visible dates give both AI engines and people a clear, trustworthy signal of how current the page is.