Short answer: a date on its own tells you little. What matters is who wrote it down (an editor, the site’s template, the web server or an outside archive) and whether it can see the content at all. Work through the signals in the order below and stop when one answers the question you are actually asking.
Start with the question: changed, or just touched?
“Last updated” can mean two different things. A page is changed when someone rewrites a paragraph, corrects a figure or adds a section. It is touched when someone presses save, a template rebuilds, or a plugin refreshes a field. Both can leave the same date behind.
So decide first. If you need to know when the words last moved, you have to confirm a real change, and only the last method below does that. If a rough sense of freshness is enough, the quicker signals will do.
Method 1: the visible “updated” label and what it can hide
The easiest signal is the one printed on the page: “Updated”, “Last updated” or “Last reviewed” near the title or at the foot of the article. An editor typed it or the content management system (CMS) filled it in, so it is the date the site wants you to see.
Google Search Central asks sites to show a visible date, label it clearly as published or last updated, and keep it consistent with the date in their structured data. A site that follows that advice should have a label that matches what sits under the hood.
A label is still a claim. It can be bumped after fixing a typo, after changing one sentence, or with no change at all. Some sites refresh it to look current; others never touch it while the content drifts. Read it as a starting point, never as proof.
Method 2: dateModified and modified_time in the page source
Next, open the page source (right-click, choose “View page source”) and search for “modified”. Two fields are worth finding:
dateModifiedin the structured data, which Schema.org defines as the date the work was most recently modified.article:modified_time, an Open Graph property that sits alongsidearticle:published_time.
These are written for machines, so they usually come straight from a CMS field rather than from a designer’s idea of a tidy byline. The weakness is that they come from the same place as the visible label. If the editor or template is careless, the label and the source dates will be careless together, and they can agree with each other while both being wrong.
A mismatch is useful. If the label says one date and dateModified says another, one was updated without the other, which tells you the page has been edited by more than one hand or process.
Method 3: the document.lastModified trick and the Last-Modified header
Now move from what the site claims to what the server reports. Open your browser’s developer console on the page and type document.lastModified. You get a date and time back.
The HTML standard says this returns the last modification date as reported by the server and, if that date is not known, the current time instead. Keep that second half in mind; it explains most wrong answers.
To see the server’s report directly, open the Network tab in the developer tools, reload, click the first request and look for Last-Modified in the response headers. MDN describes it as the date and time the origin server believes the resource was last modified, and notes that it is less accurate than an ETag for judging whether the contents changed. MDN also notes that some CMSs use it to display when content was last modified, so on some sites the header and the label share a source.
Why a server date is often just “now”
A lot of modern pages are not files sitting on a disk. They are assembled when you ask for them: a template, a database query, a few widgets and a footer. There is no single file whose age the server can report, so it either sends no Last-Modified header or sends the moment the page was assembled.
In the first case, document.lastModified falls back to the current time. In the second, the header itself says “now”. Either way, a page nobody has edited in a long while can look as though it changed seconds ago. The date is accurate about when the server built the page and silent about when a person last edited it.
The rule is simple: if the date matches the minute you loaded the page, you have learned nothing about the content. On a plain static file the header can be close to the truth; on a dynamic page, treat it as noise until something else backs it up.
Method 4: compare two Wayback Machine captures to see what really changed
This is the only method here that gives you evidence instead of a claim, because you look at the page itself at two points in time.
- Enter the page address in the Internet Archive’s Wayback Machine.
- On the calendar, pick captures marked with blue dots. The Internet Archive’s help page says blue means the crawler got a good response; other colours mark redirects and errors.
- Open an earlier capture and a later one in separate tabs.
- Scroll to the part you care about and compare the wording, figures and structure.
Now read the result against the dates you collected. If the visible “updated” date moved but the text is identical, the page was touched. If the text moved and the date did not, the label is stale. Each capture’s address also contains its own timestamp, so you can note exactly when it was taken.
Know the limits. The Internet Archive says it cannot guarantee a site has been archived, and that archived pages can be incomplete. Pages that rely heavily on JavaScript are often hard to archive. A gap in the calendar means nobody captured the page then; it does not mean nothing changed. Treat the captures you find as spot checks.
Google’s cached copy and search-result dates: what is still available
Google’s cached copy used to be the quick way to see an older version. Google’s own documentation now says the cached link feature no longer exists in Search, so that route has gone.
The date shown beside a search result is still there, and Google calls it a byline date: its estimate of when the page was published or updated. Google says it looks at several factors because any single date can be wrong. That makes it a reasonable hint and a poor record, since it is built partly from the same labels and structured data covered above.
Which signal to trust: a short decision rule
- Need to know whether the content really changed? Trust a Wayback comparison, and only that.
- Need a rough sense of freshness? Take the visible label and check that
dateModifiedagrees. If they disagree, be suspicious of both. document.lastModifiedor the header shows the time you loaded the page? Ignore it.- No captures and no dates? Say you cannot tell, rather than guessing.
If you actually want to know when a page first appeared, that is a different question with different signals, covered in how to find when a web page was published. When the date matters because you are judging whether to trust the source, finding the publisher of a website and what legitimacy scanners miss are the natural next checks.
Tracking a page you care about for future changes
Looking back only works if someone captured the page. Looking forward, you can do the capturing yourself.
The simplest approach is to save a copy of the text now, with the date you saved it, and compare it with a fresh copy later. Note the visible date and the dateModified value at the same time, so you have a before and an after. The Wayback Machine’s Save Page Now feature will capture a single page on demand, which gives you a public, timestamped copy; the Internet Archive notes it saves that one page once and does not add it to future crawls.
If the change you care about is a price, there is a shortlist of sites that track prices. For any monitoring tool, check its price, free tier and limits on the vendor’s own page on the day you sign up, the same way our own checks are dated.
Sources
- Schema.org: dateModified
- The Open Graph protocol
- MDN: Last-Modified header
- MDN: Document lastModified property
- HTML Living Standard: document.lastModified
- Google Search Central: Influence your byline dates
- Google Search Central: robots meta tag (noarchive and the cached link)
- Internet Archive Help: Using the Wayback Machine
Questions people ask
How do I find out when a website was last updated?
Read the visible 'updated' label, search the page source for dateModified or article:modified_time, and check the Last-Modified response header in your browser's developer tools. To confirm the content really changed, open two Wayback Machine captures from different dates and compare the text.
Does the Last-Modified header show when the content changed?
Not reliably. MDN describes it as the date the origin server believes the resource was last modified, and says it is less accurate than an ETag for judging file contents. On pages built fresh for each visit it often reflects the build, or is missing.
Why does document.lastModified show today's date?
The HTML standard says that if the last modification date is not known, the browser returns the current date and time. Many dynamic pages send no Last-Modified header, so you see the moment you loaded the page.
Can I see when a page was last updated if it shows no date?
Try the page source for dateModified or article:modified_time, then the Last-Modified header. If none of those help, compare Wayback Machine captures. If the archive has no captures of the page, there is no reliable way to tell.
Does Google's cached copy still show an older version of a page?
No. Google's own documentation says the cached link feature no longer exists in Search. The Wayback Machine is the main place left to see an older copy.
How do I track when a website changes from now on?
Save a dated copy of the text, or use the Wayback Machine's Save Page Now feature to make a capture, and note the visible date and dateModified at the same time. Compare against a fresh copy later. If you pick a monitoring tool, check its price and limits on the vendor's own page on the day.