41% of Pages Show a Date. Only 3% Have Anything Behind It

For the surrounding account decisions, compare the full benchmark and the method.

In short

40.9% of pages carry a freshness signal. 3.0% pass the check asking whether it is load-bearing. That thirteen-fold gap is the clearest example in the benchmark of a signal being adopted in form and not in substance.

What the two checks separate

freshness_signals asks a mechanical question: is there a date on this page that a machine can read? A dateModified in schema, a visible “last updated” line, a published date in the byline.

content_freshness_depth asks the harder one: does the page show evidence that its content has actually been maintained? Specific recent references, a changelog, dated statements about what changed, version-specific detail that would be wrong if the page were stale.

Almost everything that passes the first fails the second.

The pattern this reveals

Sites learned that freshness is a ranking and retrieval signal, and responded by making the signal true in the cheapest available way: automating dateModified so it updates on deploy. The date is now technically accurate — the file was modified — and carries no information about whether the content is current.

This is a rational response to being measured, and it is exactly why a second check exists. Any signal cheap to fake gets faked, and the useful measurement moves to whatever is expensive to fake. Here that is specificity: a page that names what changed, and when, and why, cannot be produced by a deploy pipeline.

Why this matters more for advertising content than most categories

Meta’s platform behaviour changes frequently and materially. Within the documentation we checked on 2026-09-22, Meta records that the existing-customer budget cap for Advantage+ sales campaigns is no longer available, and that from June 2024 dynamic creative could no longer be selected for new ad sets using the sales or app promotion objectives.

A guide written before either change reads as confidently correct and is wrong. A “last updated: this month” stamp on that guide makes it worse than an honestly stale one, because it converts a recognisable old page into a trusted current one.

For a category this volatile, freshness theatre is not a neutral optimisation. It actively misleads.

What load-bearing freshness looks like

• Date the claim, not just the page. “As of September 2026, Meta’s documentation states…” puts the expiry inside the sentence, where it survives extraction and where a reader in 2028 knows what they are holding.
• Keep a visible changelog on pages that change. Three lines at the bottom saying what was revised and when is worth more than any automated timestamp.
• Let stale pages be stale. A page that has not been revised should not claim it has. If the content is still correct, that is a fact about the content, not something a date needs to assert.
• Cite versioned sources. Linking to a specific documentation page that itself carries a date transfers the freshness burden to something that is maintained.

Where we sit and what it cost

Our corpus passes freshness_signals on 100% and content_freshness_depth on 97.2%, against the industry’s 40.9% and 3.0%. That is not a subtle margin, and it comes from one decision: every post carries dated, specific statements about platform behaviour rather than undated generalities. The cost is that those statements have to be re-checked when the platform moves, which is ongoing work rather than a template change.

Interpretation boundary

content_freshness_depth infers maintenance from textual specificity. A page written today with rich dated detail passes without ever having been revised, and a genuinely maintained page whose updates were silent fails. It measures whether a page demonstrates currency, not whether it is current — and those come apart in both directions.

Can software help?

Bach.ai audits your connected Meta account, estimates the revenue impact of what it finds, and proposes specific fixes. It applies a change only after you approve it. Think of it as an automated audit layer that surfaces issues and proposed fixes for your review — not a replacement for your team’s judgment, and creative production is not its core job, though the Pro and Agency plans can generate a limited number of variants.

FAQ

Does a last-updated date help AI visibility?

Only when something changed. 40.9% of pages we measured carry a freshness signal and 3.0% pass the check asking whether it attaches to genuinely revised content. An automated dateModified that moves on deploy carries no information.

Why is stale content worse in advertising than other categories?

Because the platform moves. Meta’s own documentation records that the Advantage+ existing-customer budget cap has been withdrawn and that dynamic creative was closed to new sales ad sets from June 2024. A guide predating either is confidently wrong.

How do I show real freshness rather than fake it?

Date the claim rather than the page — as of September 2026, Meta’s documentation states — keep a short visible changelog on pages that change, cite sources that themselves carry dates, and let genuinely unrevised pages stay unrevised.

Is it bad to leave a page without a recent date?

No. An honestly stale page is easier to judge than a fresh-looking one that has not been revised. The damage from an automated timestamp is that it converts a recognisable old page into a trusted current one.

Method and sources

“Only if something behind the date actually changed.”

Source: the Bach.ai AI Extractability Benchmark, run 2026-09-22. Sample: 132 pages returning HTTP 200, drawn from the published sitemaps of 12 advertising and analytics tools, plus our own 392 published posts. Every page was rendered in a headless browser and scored by the same 17-check engine the product runs against customer sites, so each figure is reproducible against the live web rather than asserted. Read the per-check rates as an industry signal, not a precise ranking of any one vendor: between 3 and 12 pages were sampled per tool, spread across the whole sitemap rather than drawn from the newest posts. The method is written up in full at how we measure AI extractability.