What the Content Freshness Analyzer does
This analyser finds every date a page states about itself - in meta tags, JSON-LD, time elements, visible 'last updated' lines and the server's Last-Modified header - and lists the parts of the text most likely to have gone out of date: past years, 'as of' statements, software versions, prices and relative words such as 'recently' or 'this year'. Each item comes with the sentence it appears in and a reason to check it.
It gives evidence, not a freshness score. A five-year-old page can be perfectly accurate, and a page updated yesterday can still quote a 2021 price. The aim is a short, concrete list of things a person should verify before the page is republished.
How to use it
- Enter a public URL, or paste the page's HTML or plain text. Pasted content is analysed in your browser; a URL is read once through a2z.tools without running JavaScript.
- Check the reference date. It defaults to today; set an earlier date to see what a page looked like against the time it was last reviewed.
- Select Analyse freshness. Read the findings first - conflicting dates and missing dates - then the table of stated dates.
- Work through the stale references from the top: point-in-time claims and dated 'best of' titles come first, then statistics, versions and prices, then relative wording. Copy the update checklist into your editorial ticket.
Reading the results
Stated dates are what the page claims, not proof of when the content changed. A modified date that moves every night while the text stays the same is a signal to fix, not a sign of freshness.
Conflicting dates - different modified dates in the text, the markup and the meta tags, or a modified date earlier than the published one - confuse readers and search engines. Google's guidance asks for one clear visible date that matches the structured data.
A past year is flagged when it sits next to wording that implies it is still current ('as of', 'currently', 'best ... of 2024', a statistic or survey). Historical mentions such as a founding year are skipped.
Worked example: the buying guide example, checked on 19 September 2026
The example guide is titled The best budget laptops of 2024. Its meta tags say it was published on 10 January 2023 and modified on 1 June 2024, the visible text agrees on 1 June 2024, but the JSON-LD says dateModified 2025-02-01. That conflict is the first finding: 595 days separate 1 February 2025 from the reference date, and the two modified dates disagree.
In the text, 'As of March 2024, the cheapest model costs $499' is a point-in-time claim 30 months old, and its price needs re-checking. 'A 2023 survey found that 41% of students...' is a three-year-old statistic. Windows 11, Python 3.8 and Chrome 120 are listed as versions to confirm, and the sentences using 'currently' and 'recently' as relative wording.
'Acme was founded in 1998' is not flagged: it is history, not a claim about the present.
Documented guidance and A2Z heuristics
The date sources checked - datePublished and dateModified in structured data, a visible byline date, article:published_time and article:modified_time meta tags - follow Google's documentation on influencing the dates shown in Search, which asks for a clear visible date and consistent markup. The Last-Modified header is defined by HTTP (RFC 9110).
Everything else is an A2Z heuristic: which years count as past (two or more years before the reference date), which wording makes a year look like a present-tense claim, and the list of software names checked for versions. The tool does not know the current version of anything; it lists mentions for a person to confirm.
Limitations: what the result does not prove
- It cannot tell whether a statement is still true. It finds statements likely to date; checking them needs the original sources.
- Dates are read only in unambiguous formats (2024-03-05, 5 March 2024, March 5, 2024, March 2024). Numeric forms such as 03/05/2024 are skipped because they mean different days in different countries.
- A URL is read as served without JavaScript, so dates and text drawn by scripts are missed; paste the rendered HTML instead.
- It says nothing about rankings. Updating dates without updating content does not make a page fresher, and Google advises against it.
Privacy: where your data goes
The analysis runs in your browser. If you choose to load a public URL instead of pasting, a2z.tools fetches that address read-only and returns the text to your tab; private and internal addresses are refused and nothing is stored. Session recording and tag-manager scripts are switched off on this page.
Standards and sources
- Google Search Central - Influencing your byline dates in Google Search
- RFC 9110 - HTTP Semantics, Last-Modified
- Google Search Central - Creating helpful, reliable, people-first content
Frequently asked questions
How do I find out when a web page was last updated?
Look for a visible 'updated' line, the dateModified value in the page's JSON-LD, the article:modified_time meta tag and the server's Last-Modified header. This tool collects all of them in one table so you can see whether they agree.
Does updating the date on a page help SEO?
Not by itself. Google's guidance is to show the date content was meaningfully changed; changing dates without changing content misleads readers. Update the facts first, then the date - the checklist ends with that reminder.
Which parts of an article go out of date fastest?
Prices, statistics, software versions, 'as of' statements and 'best of the year' titles. The analyser lists each one with its sentence so an editor can re-check them against the original sources in a single pass.
Why does it ignore dates like 03/05/2024?
Because that is 3 May in the UK and March 5 in the US. Guessing would put wrong dates in the evidence table, so only unambiguous formats are read. Write dates in ISO or with the month name.
What does the reference date change?
Everything that depends on age: how many days old each stated date is, which years count as past, and how old an 'as of' statement is. Setting it to your last review date shows what should have been caught then.
Is the content I paste sent anywhere?
No. Pasted HTML or text is parsed and analysed in your browser. Only when you enter a URL does a2z.tools fetch that public page, once, so the browser can analyse it.
Last reviewed by the A2Z.Tools team against the sources listed above.
