“Write less new content and maintain what you have instead.” The advice has been going around the SEO industry for years, and it is right. It just does not tell you which of your two hundred articles comes first.
That is exactly where most good intentions fail. The updating itself is quick. The choosing is not.
Key Takeaways - A reworked article does not start at zero. It already has a ranking history, often a few positions and with some luck backlinks too. A new article has to earn all of that first
- The work sits in the choosing, not the changing. Replacing a number takes minutes. Finding out which article it is in is the actual effort
- Turning only the date does not work. Google's own quality questions explicitly ask whether pages get freshly dated without the content having changed
- A fixed rhythm alone is not enough. Triggers work better: a traffic drop, a new survey, a changed product
What is a content refresh?
A content refresh is the targeted rework of an existing article so that it is correct again and can keep up. It is not about rewriting the text. It is about repairing the parts that no longer hold.
Before you start, it pays to decide on the scope. There are three routes, and they differ more in effort than in outcome:
| Route | When | What happens |
|---|---|---|
| Refresh | The article is sound at its core, individual details are out of date | Replace numbers, swap dead links, add missing sections. Structure and URL stay |
| Rewrite | The topic has changed fundamentally | The URL stays, the content is created anew. Backlinks and history are preserved |
| Retire | No traffic, no connection to your topic any more | Delete with a 301 redirect, or merge into a stronger article |
By far the most common case is the refresh. A complete rebuild rarely pays off, because you end up doing the work twice.
Why not just write a new article?
Because the existing article brings things with it that you would otherwise have to build from scratch. It is indexed, Google has experience with it, it may already sit on page one, and others may link to it.
How big the difference can get is something HubSpot measured on its own blog. For the old posts they reworked and republished, monthly organic views rose by an average of 106%.
The figure comes from a single blog and does not transfer to you directly. The direction still holds, and it matches what we see with clients.
Before you plan the next article: check whether you already have one on the same topic. If you do, reworking it is almost always the faster route to the same goal.
Why does a refresh work at all?
Because you get two things at once that normally pull against each other. You keep everything the article has earned over the years, and you still deliver the current state of things.
Google has a system for it
In its ranking systems overview, Google explicitly describes freshness systems that show more current content for queries where that is expected. Google’s example is a film that just came out: people want current reviews there, not pieces written during production.
For you that means recency is not a general bonus but tied to the query. What sits behind it is covered under what content freshness means.
AI systems take your figures over
ChatGPT, Perplexity and Google’s AI Overviews build an answer out of your text. A 2021 number on your page becomes a wrong answer in theirs, which is why these systems check before they cite.
What gets checked is more than the date above the article. The age of the studies you rely on counts too, and whether your outbound links are still reachable.
What’s the difference between a high-effort and a low-effort refresh?
A low-effort refresh corrects what has become wrong. A high-effort refresh changes what the page is trying to do. The distinction decides how much time a page gets and who has to be involved.
| Low-effort refresh | High-effort refresh | |
|---|---|---|
| Typical work | Replace figures, fix dead links, update screenshots | Restructure, add sections, rewrite the argument |
| Time per page | 15 to 45 minutes | Half a day to two days |
| Who does it | Anyone who knows the subject | The subject expert, usually with an editor |
| Trigger | The facts moved | The intent moved, or competitors pulled ahead |
| Risk | Almost none | Can lose what already worked |
Most pages need the first kind and get neither. A page that is factually wrong but structurally fine does not need a rewrite. Treating every refresh as a project is the main reason refresh backlogs never clear.
The tell is where the problem sits. If you can list the sentences that are wrong, it is a low-effort refresh. If you can no longer explain what the page is for, it is a high-effort one. The honest question then is whether it should be merged or removed instead.
Step 1: Find the candidates
This is where the actual work sits. Four patterns reliably show you which articles deserve a rework.
Falling traffic with no technical cause. When load time and indexing are fine and the page keeps sliding anyway, the content is the cause. We describe the pattern behind it under content decay.
Visibly outdated details. Old years, discontinued products, links into nothing. Passages like these undermine the credibility of the whole article, not just the paragraph they are in.
Positions four to ten. These articles have proven they can rank. The way forward is shorter here than for anything sitting on page two.
Many impressions, few clicks. The position is fine, the entry in the results is not. Usually it comes down to a title with an old year in it, or a meta description that no longer matches the content.
What you search with
| Tool | What for |
|---|---|
| Google Search Console | Traffic and position trends, click-through rates year over year |
| Ahrefs or Semrush | What competitors are touching on the same topic, backlink development |
| Screaming Frog | Dead links and technical findings across the whole site |
What none of these tools can do: read the text. Take the claim “most vendors charge extra for this”. No crawler can see whether it still holds, because there is neither a year nor a broken link in it.
That is what the fact check is for. It works through the claims in an article, flags the ones that may have expired, and delivers a source to read up on for every finding.
Step 2: Check what is actually missing
You have your candidates. Now you go through each one, in this order, because the first two points produce the most findings.
Numbers and statistics. Search for years in the text. For every figure, check whether the study named has a successor survey by now.
External links. Two cases, and the second is the more unpleasant one. A dead link is noticed immediately. A link that still works but leads to a page that has since changed survives every technical check.
What has arrived since publication. New tools, new procedures, new requirements. That concerns the basic premise of the text more often than individual facts.
What the pages ahead of you have. Open the top three results for your keyword and compare. Usually you find two or three points that are missing on your page.
Step 3: Set the order
You cannot rework everything at once, and the most common mistake is starting with the page that has the most findings. That feels productive and gains little if hardly anyone visits the page.
Sort by impact and effort instead:
- High traffic, small change. An article in position five that is only missing two current figures. Start there.
- High traffic, large change. An important article that needs new sections. That belongs in the editorial calendar, not in a spare hour.
- Low traffic, small change. Collect these and work through them in one block each quarter.
- Low traffic, large change. Here the more honest question is whether the page should stay at all.
Two things weight the order on top of that. An article that fell from 5,000 to 2,000 visitors has more potential than one that never passed 100. With the first, you know it can be done. And pages closer to a conversion deserve more attention than side topics with the same traffic.
Richard Roth Founder CiteRanger, SEO Strategist The order sounds banal, but it is the point where most refresh projects tip over. Anyone sorting by gut feeling almost always lands on the article that annoys them most, and that one rarely has the most traffic.
Step 4: The rework
Now the actual work. In this order, because each step makes the next one easier.
Replace numbers. Take the new figure from the original study, not from an article quoting it. Numbers slip as they get passed along: “38% of the pages examined” becomes “38% of all pages”, a median becomes an average.
Renew links. Swap dead targets, read changed sources again. This step pays off twice, because you almost always find the outdated numbers along the way.
Tighten the structure. Add missing subheadings, split long paragraphs, bring the introduction up to today’s state.
Add sections. What has become important since publication belongs in the article. For many older pieces, that is now the part about AI search.
Relink internally. New articles have appeared since publication that fit here. And the other way round.
Adjust the date. For genuine content changes the modification date belongs updated, the publication date stays. How to mark both up cleanly is covered in Google’s documentation on dates.
What not to do while you are at it
Turn only the date. Google’s own content quality questions include one about this. Are you changing the date of pages to make them seem fresh, when the content has not substantially changed? What makes it onto that list is not a shortcut.
Change the URL. That loses you the backlinks that made the article strong in the first place. If it is unavoidable, then with a 301 redirect.
Rebuild everything at once. Not because Google would otherwise fail to recognize the article, that is industry folklore. But because after a complete rebuild you can no longer tell which change did the work.
How do I maintain backlinks during a content refresh?
Keep the URL and keep the passage people linked to. Those two rules cover almost every case.
A backlink points at a page because of something on it: a statistic, a definition, a chart, a specific claim. A refresh that removes that thing leaves the link pointing at a page that no longer supports the reference. That is the same failure as content drift, except this time you caused it.
Four practical rules:
- Never change the URL during a refresh. A new address needs a redirect, and a redirect costs a little at every hop, permanently.
- Check what the links point to before you cut. Pull the referring pages for the URL and look at the anchor text and the context around it. That tells you which passage earned the link.
- If the linked claim is outdated, update it rather than delete it. A current figure keeps the reference valid. Removing the section does not.
- Keep headings and anchor IDs stable. Deep links to a section break silently when the heading is renamed.
Where a passage genuinely has to go, leave a short line saying what replaced it. A reader arriving from an old link then lands somewhere that still makes sense.
Step 5: Follow up
A refresh is only finished once you know whether it worked.
Watch position and traffic against the weeks before. How long you wait for a reaction depends on how often Google crawls the page, and that varies a lot. On a page that gets visited daily anyway, you see something sooner than on an article deep in the archive.
If nothing moves after several weeks, three questions help. Was the change substantial enough, or was it cosmetic? Did the pages ahead of you also improve in the same period? And is there another cause, a technical finding or new competitors?
How often should you rework content?
It depends on the topic. Tech every three to six months, marketing every six to twelve months, legal topics with every change in the law, fundamentals once a year.
Those are starting values, not rules. More useful than the calendar are triggers, because they hang on the substance and not on the date:
- A page’s traffic drops. The earlier you get to it, the better your chance of seeing the cause
- A new survey appears. Then every page quoting the old one is due
- A product changes that you recommend or compare
- A competitor improves and overtakes you on an important keyword
- A QDF event means Google suddenly prefers fresh content on your topic
The difference between a calendar and a trigger is who gives the nudge. As long as you have to think of it yourself, the upkeep of your archive hangs on your memory, and memory loses against every urgent task.
Conclusion
A refresh process has four parts: find, sort, rework, follow up. The first part is where it fails, and also the one you can most easily hand off.
Start with your ten strongest pages. Find the old numbers, click the links, replace what no longer holds. After that you know how big the issue really is on your site, and you can decide whether a quarterly date is enough.
For teams, the content update playbook turns this into a routine with roles and a checklist. How to date a refresh honestly is covered in last updated dates and dateModified.