Content refresh SEO: sort the page into one of four verdicts first

Content refresh SEO fails most often at the first step: deciding which of four things an old page needs. Refresh, merge, differentiate and delete are different operations with different costs, and doing the wrong one wastes a day and sometimes loses a ranking. This chapter is the sorting table, plus the three things each verdict obliges you to do afterwards.

Read this first

This chapter decides what to do with a page. Actually doing the rewrite is the one-page rewrite checklist, which is a worksheet you run after this decision, not instead of it.

You need Search Console for the sorting to work, because three of the four verdicts depend on impressions and on whether more than one of your URLs competes for the same search. Without that data you are guessing, and guessing tends to produce refreshes — the option that feels productive.

Content refresh SEO starts with four verdicts, not one

Sort on evidence, not on age. A page from three years ago that still earns impressions needs nothing; a page from last quarter with zero impressions and a near-twin two folders over needs a merge.

Merge and differentiate are the two people conflate, and they are opposites. Merge collapses two pages because they were always the same answer. Differentiate keeps both because they answer different questions and the writing failed to make that visible.

Start from the pages with impressions but no clicks

If you have an archive and no idea where to begin, this is the highest-yield filter. Pages with impressions and no clicks are already eligible, already being shown, and failing at the last step — which means the fix is usually one of the four verdicts rather than anything structural.

Pages with neither impressions nor clicks are a different problem, and refreshing them is the classic wasted afternoon. Nothing you rewrite changes a page that is not being shown for anything; that is a keyword decision, and it belongs upstream.

How to tell merge from differentiate

Take the two URLs and read the searches each one earns impressions for. If the lists overlap almost entirely, you have one page written twice — merge. If each list has a distinct centre and only the edges overlap, you have two pages that look alike — differentiate.

The mistake to avoid is merging on similarity of text. Two pages can read almost identically and serve genuinely different intents, and merging them loses whichever intent the survivor stops serving. Sort on what people searched, not on what you wrote.

What each verdict obliges you to do next

The follow-through is where most of the value leaks. Each verdict has exactly three obligations, and skipping any of them turns a good decision into a broken one.

A refresh has to change the page, not the date

Google's sitemap documentation is unusually direct about this. The lastmod value "should reflect the date and time of the last significant update to the page", and it gives an example of what does not count: "an update to the copyright date is not" significant.

That rules out the entire category of maintenance that only looks like a refresh — bumping a year in a title, regenerating a page with no content change, touching a file to move its timestamp. If the only thing that changed is the date, you have not refreshed anything, and you have spent a credibility signal to say so.

There is a second trap here that is not your fault. Our survey of 30 sites found that of the 24 with a readable sitemap, 13 either carry no lastmod at all or stamp every URL with the same day — so for those sites a genuine refresh produces no distinguishable signal. That write-up is sitemap lastmod, measured, and it is worth checking your own file before you rely on refresh dates to communicate anything.

Three ways this goes wrong

Two of these are follow-through failures rather than decision failures, which is why they survive review — the call was right and the work was left half done.

1. Refreshing on a schedule instead of on evidence. Quarterly refresh cycles produce edits to pages that were fine and leave the broken ones untouched, because the calendar does not know which is which.
2. Merging without moving the internal links. The redirect keeps readers moving; the internal links still point at a URL that now bounces, and every one of them is a wasted signal.
3. Deleting without deciding between 410 and 301. A redirect to an unrelated page is worse than a clean removal, and it is the default people reach for because it feels safer.

When to choose 410 over 301

This is the one row in the table with a real fork in it, so it is worth its own rule rather than a judgement call each time.

Redirect when there is a genuine successor covering the same intent. Return 410 when the page is simply gone and nothing you have answers the same question.

A 301 to a vaguely related page is the common wrong answer. It preserves a number on a dashboard and delivers a reader to something they did not ask for, and it teaches an engine that your redirects are not meaningful.

What this chapter cannot tell you

We have not measured what a refresh does to rankings or to citation, and we are not going to quote a figure for it. What is documented is the input side — what counts as a significant update — and the sorting logic here is built on evidence you can see in your own Search Console rather than on a claimed effect size.

We also cannot tell you how quickly a refreshed page gets re-crawled. Nobody publishes that, and our own attempt to measure it is blocked on server log access. The window to expect, and when to conclude nothing is happening, is in how long does it take to get indexed.

Running this sort across a whole archive, rather than on the handful of pages you happen to remember, is what QueryWin is being built to do.

Common questions

How often should I do a content refresh?

When the facts on a page stop being true, or when its impressions fall while the search itself has not. Both are triggers you can observe. A fixed cadence is not a trigger, it is a habit.

Does updating the date help?

Not on its own. Google's documentation says the value should reflect the last significant update and explicitly excludes a copyright date change, so a date moved without a content change is a claim you cannot back.

Should I keep the old URL when refreshing?

Yes. A refresh that changes the URL is a migration wearing a refresh costume, and it throws away whatever the old URL had accumulated.

Is deleting old content risky?

Deleting a page that earns impressions is. Deleting one that earns none, has no inbound links and answers nothing is housekeeping — and an archive full of those makes every other decision harder, because you cannot see the pages that matter through the ones that do not.

What if two pages both rank well?

Then they are not competing and nothing needs doing. Cannibalisation is a diagnosis for when neither ranks, not for any case where two of your URLs appear for related searches.