How Do Visible Update Dates and dateModified Schema Signal Content Freshness?

How are visible dates and dateModified used?

The visible date matters more than many teams assume. Tests of AI assistants found that when they fetch a page directly, they read the visible content and ignore JSON-LD. So a date that exists only in markup is invisible to them. Put a clear, labelled date in the page text, and keep the markup consistent with it.

Do last-modified dates affect citations in AI search answers?

They are linked, though dates alone are not the cause. Seer Interactive found that in March–June 2026, 75% of the pages cited by ChatGPT, Gemini and Perplexity had been updated within the past year. Only 12% were older than two years. An Ahrefs study of about 17 million citations found AI assistants cite pages that are, on average, 25.7% fresher than those in Google's organic results.

Two things drive this. Engines prefer current information for time-sensitive questions, and pages that are maintained are usually better. A visible, accurate date helps an engine recognise a current page, but the gain comes from what was updated.

How do freshness signals differ between Perplexity, ChatGPT, and Google AI search?

The differences reflect what each engine tends to cite and how it retrieves, more than a single "freshness setting". For time-sensitive questions every engine leans towards current pages; for evergreen questions all of them will cite older authoritative ones. Perplexity's behaviour is covered in more detail in whether Perplexity favours recently updated pages.

Do LLMs avoid old-dated content, and can refreshing dates improve citation share?

Old content is cited less often for time-sensitive questions, but it is not avoided outright: in Seer's data, around a quarter of cited pages had not been updated for over a year. Refreshing the date alone does not help. Google's helpful content guidance treats changing a page's date to make it seem fresh, without substantial changes, as a warning sign. And an engine that reads a "2026" date above figures from 2022 gains nothing useful.

A genuine refresh can improve citation share: new figures, current examples, corrected steps and newly answered questions. Update the visible date and dateModified when you make those changes, not otherwise. What to refresh and when is covered in how often to refresh old content to improve AI visibility.

Do current-year titles and headings signal freshness to AI search engines?

A year in the title helps match searches that include a year, such as "best payroll software 2026", and tells readers the page is meant to be current. It is not proof of freshness. If the content has not been updated, the title misleads readers and the page's facts give it away. Use a year in the title only for content that genuinely changes by year, such as comparisons, pricing guides, statistics roundups and annual reports, and update the content when you update the year. Evergreen explanations do not need one.

Do fresh content and multiple formats like PDFs improve AI visibility?

Fresh content helps, as above. Publishing the same content in several formats does not add a freshness signal. PDFs can be indexed and read by many engines, but they often lack clear dates, headings and links. They are also harder to update in place, so old versions linger. For content you want cited, publish it as an HTML page with a visible date. Offer a PDF as a download if readers need one, and keep the HTML version as the main, updated source.

Does changing XML sitemap lastmod update facts in Gemini and Perplexity?

Not directly. <lastmod> tells crawlers a page has changed, which can prompt a recrawl. Google says it uses <lastmod> only if it is consistently and verifiably accurate, compared with the actual changes on the page. Once a page is recrawled, answers grounded in search can use the new facts: Google's AI features, Gemini when it searches, and Perplexity. Answers that come from model memory, without a search, keep the old facts until a future model update. Changing <lastmod> without changing the page teaches crawlers to distrust your dates.

How should you set dates properly?

1. Show a labelled, visible date: "Published 3 March 2025 · Updated 12 September 2026".
2. Add datePublished and dateModified to Article, BlogPosting or other relevant structured data, matching the visible dates.
3. Update both only for significant changes, such as new facts, sections or corrections, not typo fixes.
4. Keep the original published date; do not reset it on update.
5. Never use future dates, or the date of the event a page describes instead of when the page changed.
6. Keep sitemap <lastmod> in step with real changes.
7. Consider a short change note for important pages, such as "Updated September 2026: new pricing for 2026–27".

Google's byline date guidance recommends labelled dates and consistency between visible and structured values, and warns against future dates. Where structured data conflicts with visible content, see whether schema that conflicts with page content reduces AI trust.

What do people ask about dates and AI freshness?

Should every page show an updated date?

Pages whose accuracy depends on time should: guides, prices, comparisons, statistics and policies. Timeless pages such as an About page can do without one.

Does removing dates avoid looking outdated?

It removes a signal engines and readers use to judge currency. For time-sensitive content, an undated page is harder to trust than a dated one.

Will AI engines notice if I fake update dates?

Google says it ignores lastmod values that are not consistently and verifiably accurate, and readers notice stale facts under a new date. It risks trust without adding anything useful.

Does Bing use dateModified?

Bing supports schema.org dates and recommends accurate lastmod values in sitemaps; as with Google, accuracy matters.