How Often Should You Update Old Blog Posts? A Refresh System for Canadian Businesses

By Cody Wise, founder of Wise Media, Calgary. Last updated 6 November 2026.

Summary

• Four outcomes, not one: refresh, rewrite, merge, or delete. Most sites only ever do the first and wonder why the library keeps growing and the traffic does not.
• Sort by impressions, not by age. A two year old post holding impressions with no clicks is the highest value hour on your site. A two year old post with no impressions is not worth touching.
• Merging two competing posts into one usually beats refreshing either. Keyword cannibalisation is the most common reason a page plateaus.
• Changing the date without changing the content is a tactic search engines are specifically built to discount.
• Google says original, non-commodity content matters more than any other optimisation, and says so in its own documentation. That applies to a refresh as much as to a new post.

Table Of Contents

• How Often Should You Actually Update Old Posts?
• Why Content Decays, And What Decay Looks Like
• Which Posts To Touch: The Triage
• Refresh, Rewrite, Merge Or Delete
• Merging, With A Worked Example
• Dates, dateModified And The Thing Not To Do
• The Quarterly Refresh System
• Common Mistakes
• FAQ

How Often Should You Actually Update Old Posts?

Set the interval by how fast the underlying facts change, and review the whole library quarterly to decide who is due.

The quarterly cadence is the review, not the edit. You are not refreshing everything every quarter. You are spending an hour every three months deciding which ten posts have earned attention and which ten should quietly go away.

The 50 Word Answer

Review your whole blog quarterly and refresh by decay rate: fast-moving topics every three to six months, evergreen guides every twelve to eighteen, regulatory content on every rule change. Prioritise posts with high impressions and low clicks. Merge or delete more than you refresh. Never change the date without changing the content.

Why Content Decays, And What Decay Looks Like

Decay is not one thing. Four different failures all look like a traffic decline on a chart, and they need four different fixes.

• Factual decay. The post is still well written and the facts in it are wrong. Prices, product names, rules, the current version of something. This is the most damaging kind because a reader who catches one error distrusts the rest.
• Competitive decay. Nothing in your post changed. Three competitors published something better. Your post is the same quality it always was and is now the fourth best answer.
• Intent drift. What people mean by the query changed. A search that used to want a definition now wants a comparison, or a tool, or a price. Your page answers the old question perfectly.
• Self-inflicted decay. You published a second post on the same topic and the two now split their own signals. This is cannibalisation, and it is the one entirely within your control. We covered the detection method in how to find and fix keyword cannibalisation.

Diagnose before you edit. Rewriting a post that is suffering from cannibalisation makes the problem worse, because now you have two good posts competing instead of one good and one weak.

Why This Matters More Than It Used To

There is a second reason to prioritise by impressions rather than by age, and it has to do with what a ranking is now worth. Pew Research Center, studying 900 US adults who shared their browsing activity in March 2025, found users clicked a traditional search result on 8% of visits where an AI summary appeared, against 15% where it did not. That is American data and directional rather than a Canadian benchmark, but the implication for a refresh programme is clear enough: a page holding impressions and losing clicks is no longer always a title problem. Sometimes the answer is being read above your result. That does not make the refresh pointless, because being the cited source still requires being accurate and current, but it should change what you expect a refresh to deliver. Measure recovered clicks where you can, and accept that some of the value now shows up as presence rather than traffic.

Which Posts To Touch: The Triage

Pull a 12 month Search Console export, with page, query, impressions, clicks and average position. Then sort into five buckets. This takes about an hour for a 200 post site and it is the whole method.

The bucket people skip is the last one. A library of 200 posts where 140 earn nothing is not a library of 200 assets. Google’s own guidance for generative AI features warns that duplicate and commodity content wastes crawl resources and adds little value, and says creating content from your own knowledge and experience will likely matter more over time than any other suggestion it makes. Deleting dead weight is the cheapest way to raise the average quality of what remains.

Refresh, Rewrite, Merge Or Delete

Refresh: 30 To 60 Minutes

The structure is sound and the argument still holds. You are correcting facts and closing gaps.

• Verify every number, price, date, product name and statute reference against its primary source. Replace anything you cannot verify rather than softening it.
• Check every outbound link still resolves and still says what you said it says.
• Add the two or three things that have genuinely changed since publication.
• Add internal links to anything you have published since, in both directions.
• Tighten the opening so it answers the question in the first 50 words.
• Update dateModified, and show a visible last-updated line.

Rewrite: 2 To 4 Hours

The topic still matters but the page does not deserve to rank. Keep the URL, replace the content. Keeping the URL is the entire point: you inherit whatever authority the old page accumulated instead of starting at zero.

Merge: 1 To 2 Hours

Two or more posts compete for the same query. Pick the stronger URL, usually the one with more links and better position, fold the unique material from the others into it, then 301 the losers to the winner and update every internal link that pointed at them.

Delete: 10 Minutes

No impressions, no links, no strategic value, nothing worth folding anywhere. Redirect it to the closest relevant page if one exists, or let it 410 if nothing fits. Do not redirect everything to the homepage; a pile of unrelated redirects to the homepage is treated as a soft 404 and earns nothing.

Merging, With A Worked Example

This site is a fair example of the problem, so here is our own list rather than an invented one. Six pairs on wisemedia.io currently target the same queries as each other:

• Two posts on AI website builders versus hiring an agency, one with a country suffix and one without.
• Two on redesigning a website without losing rankings.
• Two on how long it takes to build a website.
• Three on website accessibility compliance in Canada.
• Three on what a professional website costs in Canada.
• A pair on what makes a website feel premium.

Three of those are three-way splits. A three-way split on a commercial query like website cost is not a small inefficiency. It is the site competing against itself for its highest intent term, and no amount of refreshing any one of the three fixes it.

The Merge Procedure

1. Pick the survivor on evidence, not preference. Highest impressions, best average position, most referring domains, cleanest URL. If one has external links and the other does not, the linked one usually wins regardless of which reads better.
2. Take an inventory of what is unique in each loser. Most of the content will be duplicative. Usually only 20 to 30 percent is genuinely new material worth moving.
3. Fold that material into the survivor properly. Not appended as a new section at the bottom. Integrated where it belongs, so the merged page reads as one piece rather than three stapled together.
4. 301 every loser to the survivor. One hop, no chains.
5. Update every internal link that pointed at a merged URL. If you skip this you build a site full of internal links that all take a redirect hop.
6. Update dateModified on the survivor and resubmit the sitemap.
7. Wait six weeks before judging it. Merges dip before they recover.

Honest expectation setting: a merge is not guaranteed to produce the sum of the two pages. It usually produces more than either alone and less than both added together, and the gain shows up as a higher ceiling over the following months rather than as a jump.

Dates, dateModified And The Thing Not To Do

Three separate things get confused here.

• The visible last-updated line. For readers. Worth having on anything evergreen, because a reader deciding whether to trust a pricing guide looks for it first.
• dateModified in your Article schema. For machines. It should reflect a real, substantive edit.
• The displayed publish date. Leave it alone. Changing the original publication date to today is not a refresh, it is a claim that the post is new.

Bumping dates without changing content is a well-known tactic and it is the kind of manipulation search systems are explicitly designed to discount. Google’s guidance warns against producing content at scale to manipulate rankings, and a date change with no substantive edit sits in exactly that family. It also actively costs you something: once you have marked a page as updated, you lose your own ability to tell which pages genuinely are.

The honest version is simple. If you changed something real, update dateModified and say so. If you changed a sentence, do not.

The Quarterly Refresh System

Four hours a quarter for a site under 300 posts. Put it in the calendar or it will not happen.

1. Hour one, pull and sort. Export 12 months of Search Console data by page and query. Sort into the five buckets above.
2. Hour one, continued, pick ten. Five from the high-impressions-low-clicks bucket, three from the position 5 to 15 bucket, two merges. Write the list down.
3. Hour two, title and meta pass. Rewrite titles and descriptions for the five click-starved posts. This is the highest return work on the list and it does not require touching the body copy.
4. Hours three and four, the real edits. Refresh the three near-miss posts and execute the two merges.
5. Fifteen minutes, the delete pass. Identify everything with zero impressions over 12 months. Redirect or remove.
6. Log what you did and the date. Next quarter you compare against this, which is the only way to know whether the system works.

Measure at 90 days, not at 14. Refreshes take a recrawl plus a reassessment, and the first two weeks after an edit tell you nothing.

Pros And Cons Of A Refresh Programme

Common Mistakes

• Refreshing by age. Age is not decay. A three year old post holding position four needs nothing.
• Changing the date and nothing else. Discounted by search systems, and it destroys your own audit trail.
• Refreshing a cannibalised page. Diagnose first. Two competing pages need a merge, not two edits.
• Changing the URL during a refresh. The URL’s history is the asset. Keep it and replace the content.
• Adding 800 words for the sake of length. Google states there is no ideal page length. Padding a page makes it worse.
• Never deleting anything. A library where most pages earn nothing is a liability you keep paying to crawl.
• Judging at two weeks. Nothing has been reassessed yet. Ninety days is the honest window.

FAQ

How Often Should I Update Old Blog Posts?

Review quarterly, refresh by decay rate. Fast-moving topics every three to six months, evergreen guides every twelve to eighteen, anything regulatory on every rule change. Most posts in a given quarter should be left alone.

Does Updating Old Posts Actually Improve Rankings?

It can, when the update genuinely improves the answer or corrects decay. It does not work as a mechanical trick. The gain comes from the page being better, not from the fact of an edit, and a refresh cannot rescue a page targeting a query nobody searches.

Should I Change The Publish Date When I Update A Post?

No. Keep the original publish date and update dateModified plus a visible last-updated line. Rewriting the publish date misrepresents the post’s age and removes your own ability to audit what has actually been maintained.

Is It Better To Update An Old Post Or Write A New One?

If a page on the topic already exists and ranks anywhere at all, update it. Writing a second post on a topic you already cover is how cannibalisation starts. Write new only for genuinely uncovered questions.

How Many Old Posts Should I Delete?

However many have earned no impressions in twelve months, carry no external links, and contain nothing worth folding elsewhere. On most sites that is a larger share than the owner expects. Redirect each to the nearest relevant page where one exists.

Will Deleting Posts Hurt My Traffic?

Deleting pages that produce no impressions cannot cost you impressions. The risk is deleting something that quietly carried a backlink, so check referring domains before removing anything, and redirect rather than remove when in doubt.

Does A Refresh Help With AI Citations?

Accurate, current content is more likely to be a safe thing for an answer engine to quote, so keeping facts right matters. But be careful about the claim that freshness alone drives citation. The larger factor is off-site presence, which we argue in detail in AI citations versus backlinks.

Conclusion

A refresh system is mostly a triage system. The editing is the easy half. The discipline is sitting down every quarter, sorting by impressions rather than by date, and being willing to merge or delete more pages than you polish. Do that four times a year and a library that was quietly decaying starts compounding instead.

If your blog has grown past the point where you can hold it in your head, our website growth packages include the quarterly audit and the merge work. Send us the site through our intake form and we will tell you how many of your pages are competing with each other before you commit to anything.