Content Freshness

Content Freshness measures how recently your page was updated, as AI systems see it. friendly4AI reads the dated signals on your page, takes the most recent strong one, and scores its age. A page passes (100) when that signal is within 90 days, lands a partial (50) between 90 and 365 days, and fails (0) when the newest signal is older than a year or missing entirely.

What does Content Freshness measure?

The parameter answers one question: how recent does your content look to an AI system? friendly4AI gathers dated signals from four places:

• the Last-Modified HTTP header,
• JSON-LD dateModified,
• a visible "Last updated" date in the page text, and
• the server Date header.

From those, it picks the most recent strong signal and measures its age in days. The first three count as strong signals and decide the score. The server Date header is a weak fallback only — used when no strong signal exists — because it tells you when the page was served, not when the content actually changed.

Why does freshness matter for AI-readiness?

AI engines prefer current information and quietly discount anything that looks stale. Perplexity, ChatGPT search, and Gemini all weigh recency when they decide what to surface. So a page with no freshness markers, or one whose newest signal is a year old, gets skipped in favour of a dated competitor. Clear, honest freshness metadata tells a model the content is maintained and worth citing today.

The dates have to be real, though. Stamp an old page with a fresh date and leave the content untouched, and you produce exactly the signal AI systems learn to distrust.

How is Content Freshness scored?

This Authority and Trust parameter runs an age-based classifier under the v4.5 methodology.

| Result | Score | Condition | |---|---|---| | Pass | 100 | A strong signal is present and the newest signal is within 90 days | | Partial | 50 | The newest strong signal is between 90 and 365 days old, OR only a server Date header within 365 days is available | | Fail | 0 | No freshness signals are found, or the newest signal is older than 365 days |

If the server Date header is all that's available, the result caps at partial when it's within 365 days and fail when it's older. No signal at all fails.

A newer v2.2 enhancement — config-gated and off by default — swaps the binary 90/365 split for a four-band classifier (fresh, recent, aging, at-risk), adds a Perplexity-staleness flag, and guards against future-dated signals. With that flag disabled, the 90/365 bands documented here are what's live.

How do you fix Content Freshness issues?

• Send an accurate Last-Modified HTTP header that moves whenever the page content genuinely changes.
• Add dateModified to your JSON-LD (Article, WebPage, or similar) and keep it pinned to the real edit date.
• Put a visible "Last updated" date on substantive pages so readers and crawlers both see the recency.
• Never back-date or fake freshness. Tie every date to a real content update — mismatched signals erode trust.
• Re-scan and check the freshness_days, freshness_source, last_modified_header, and schema_date_modified evidence fields.

Related parameters

• Dated statistics ratio — how many in-text statistics carry a date.
• Author authority — author and expertise signals AI systems trust.
• Content depth — substance and coverage of the page.