Of 27 pages ranking for content advice, 8 show the reader a date. Eighteen show one to Google.
Every content audit template has a line for freshness. I went looking for where that freshness is actually published, and found it is mostly published to robots.
- 27 pages checked. 8 put a date where a reader can see it. 18 put one in structured markup for Google.
- Seven pages carry no date anywhere at all, neither on screen nor in markup.
- Two pages disagree with themselves. One shows readers September 2023 and tells Google August 2026.
- Median page was last modified 227 days ago. Five have not been touched in over a year and still rank.
- Median length is 2,930 words, which is four times what the same pages recommend as a starting point.
There is a line in every content audit template about freshness. Last updated, flag anything over a year, schedule a refresh. I have built that template. I have handed it to clients.
What I had never done is check where the freshness actually gets published once you have decided it matters. So I took the top results for four queries about content strategy and auditing, one page per domain, and looked at twenty-seven of them.
The reader mostly does not get told
Eight pages out of twenty-seven put a date somewhere a person can see it. Eighteen hand one to Google in structured markup, quietly, where no reader will ever look. Seven carry nothing at all, on screen or in code.
Read that ordering again. On pages about writing for an audience, the machine is the better informed party by more than two to one.
Two pages contradict themselves
One shows readers September 2023 near the byline and tells Google it was modified in August 2026. Nearly three years apart, on the same page, at the same moment.
I do not think that is a lie anyone typed on purpose. A template prints the original publication date, a build pipeline stamps dateModified on every deploy, and nobody puts the two side by side. But a reader who scrolls to the top and sees 2023 makes a decision about whether to trust the advice, and that decision is based on the wrong number.
Open your own article and look
Is there a date on screen where someone would actually see it? Now view source and find dateModified. Do those two say the same year?
A minute per page. It is the cheapest line on any audit list and the one nobody runs, because freshness gets treated as a scheduling problem rather than a publishing one. If the two disagree, pick the true one and make the other match it. If there is no date anywhere and your topic moves, put one where the reader lands, not only where the crawler looks.
The pages that never got touched
Median last-modified in my sample is 227 days ago. Five pages have not changed in over a year, and they are sitting in the top results anyway.
That is worth holding onto next time someone proposes a refresh sprint. Something in the top ten has been untouched since last year, and it is still there.
The number I did not expect
Median length across these pages is 2,930 words. Several of the same pages recommend starting somewhere around seven hundred.
They rank with four times what they tell you to write. That is either a lesson about the format or a lesson about advice in general, and after staring at the numbers for a while I genuinely cannot tell you which one it is.
An afternoon that went nowhere
I tried to line up modification date against position in results to see whether fresher pages sat higher. Four queries and twenty-seven pages cannot answer that, I knew it while I was doing it, and I did it anyway because the data was already sitting there.
What I still want to know
Seven pages carry no date at all, and two of those are Google’s own documentation.
If the largest publisher of technical guidance on the web deliberately omits the thing every audit template demands, either they know something about how readers use dates that the templates do not, or documentation is simply a different genre with different rules.
I lean towards the second. I have not tested it, and the test would be a different measurement: not who shows a date, but whether showing one changes what a reader does next.
Questions people ask about this
How did you decide what counts as a visible date?
A date in the first two thousand characters of page text. Below that you start catching dates from inside the article itself, like a study year, and those are not publication dates. It is a rough rule and it will miss a date placed at the very bottom of a long page.
Where does the markup date come from?
From datePublished and dateModified in JSON-LD, or from the article:published_time and article:modified_time meta tags. That is what Google reads and what feeds the date it sometimes shows in results.
Is showing no date bad?
Not automatically. Evergreen reference pages often drop it on purpose, and Google's own documentation pages in my sample carry no date either. It becomes a problem when the topic moves and the reader cannot tell whether they are reading something from this year or from four years ago.
What about the two pages that disagree with themselves?
That is the part I would fix first. A reader sees one year, Google sees another, and one of them is being told something untrue. In the worst case here the gap is close to three years.
Does updating an old page actually help it rank?
My sample cannot answer that, and I am not going to pretend otherwise. Twenty-seven pages across four queries shows what is published, not what causes rankings. Five pages untouched for over a year are sitting in the top results, which at least argues against panic.
- Google Search Central, Article structured data. Checked 2026-08-26.
- Google Search Central, Dates on your pages. Checked 2026-08-26.
- Schema.org, dateModified. Checked 2026-08-26.
- Google Search Central, Creating helpful, reliable content. Checked 2026-08-26.
- Google Programmable Search Engine, Custom Search JSON API. Checked 2026-08-26.
- Nielsen Norman Group, Trustworthiness in web design. Checked 2026-08-26.
- Own measurement, 27 pages, 26 August 2026, script in the Milestones research folder. Checked 2026-08-26.
Build the brief before the draft. Type a topic, an audience and a goal, and the generator gives you the questions to answer, the sources to find and the number to measure.
Build a brief free → Open the glossary