Guides 15 Sep 2026 6 min read by Kalum

How to update old blog posts (a refresh checklist that works)

Refreshing a post that already ranks beats writing a new one almost every time. The catch is that most "refreshes" are date changes with a paragraph bolted on. Here is how to pick the right posts and what to actually change.

Most sites have a folder full of posts that used to do well. They still rank somewhere on page two, they still get impressions, and nobody has opened them in three years.

Fixing those is the highest-return work in SEO for a small site, and almost nobody does it — because publishing something new feels like progress and editing something old feels like admin.

Why refreshing beats publishing

A new post starts from zero. No index history, no links, no behavioural data, and a wait of weeks to months before Google has any opinion about it at all.

An old post that ranks at position 12 has all of that already. It has been crawled a hundred times, it has whatever internal links you gave it, and Google has a settled idea of what it is about. Moving it from 12 to 6 takes an afternoon. Getting a new page to 6 takes a season.

There is a second reason, less obvious: content decay is the default. A page does not need to get worse to lose rankings. It just has to stay the same while everyone else's version gets updated. Most traffic losses on older sites are not penalties — they are slow decay nobody noticed.

Picking the right posts (not the oldest ones)

The instinct is to sort by date and start at the bottom. That is wrong — plenty of old posts are old because nobody wanted them in the first place.

Sort by opportunity instead. Export the last three months from Search Console (Performance → Pages → Export) and look for these four patterns, in this order:

1. High impressions, position 4–20. Google already rates the page. This is the best return on an afternoon, every time.
2. Traffic falling steadily over six months with no single drop date. Textbook decay. Compare the last 3 months with the previous 3 months in the Compare view.
3. Good position, terrible click-through rate. The page ranks; the snippet does not sell. That is a 20-minute fix, not a refresh.
4. A year in the title that is not this year. "Best X for 2023" is actively repelling clicks in 2026, and it is the cheapest fix on this list.

Take the top five. Do not take twenty — a half-finished refresh of twenty posts is worth less than five done properly.

The 20-minute audit before you edit

Do not open the editor yet. Open two things instead: your post, and the three results currently above you for its main keyword.

Write down four lists:

• Sections they have that you do not. Read their subheadings. This is usually the whole reason you are below them.
• Things on your page that are now wrong. Prices, version numbers, screenshots of interfaces that were redesigned, tools that shut down, statistics with a year attached, "coming soon" about something that came and went.
• The People Also Ask questions on that query. Any your page does not answer is a section you are missing.
• What they show that you only describe. Screenshots, a table of real numbers, a worked example, a named author who clearly did the thing.

Those four lists are your edit. Everything below is just executing them.

The refresh checklist

The snippet

• Exact target keyword near the front of the title.
• Year updated if the title carries one — and only if the content genuinely is current.
• Meta description that answers the search in one sentence, not a description of the article.
• URL slug left alone. Changing it costs you the history unless you get the redirect exactly right, and the gain is almost nil.

The first screen

• The direct answer in the first hundred words. If someone lands and closes the tab in four seconds, nothing further down matters.
• Cut the throat-clearing. "In today's fast-paced digital world" and the three paragraphs of history before the answer.
• One line saying what changed and when, if the update is substantial. Readers trust a page more when it admits it was revised.

The body

• Add the missing sections from your audit — properly, with real substance, not a sentence each.
• Correct everything on the "now wrong" list. This is the part readers notice and the part that earns the update.
• Add one piece of first-hand proof: a screenshot you took, a number you measured, a result you got. One is enough and it changes how the whole page reads.
• Cut sections that no longer serve the query. Removing 400 stale words is as valuable as adding 400 good ones.
• Answer the People Also Ask questions, using the question itself as a subheading.

Links

• Three to five internal links to this post from other relevant posts, with the keyword as anchor text, inside the body.
• Links out from this post to your newer related posts.
• Fix or remove dead external links. A page full of 404s reads as abandoned to a human, and it is trivially easy for a search engine to notice too.

Media

• Replace screenshots of anything that has been redesigned. This is the fastest way to signal a genuine update.
• Alt text that describes the image, not the keyword.
• Compress anything huge. An old post often carries 2 MB images from before you cared.

What to do about the date

Change the content, then change the date. Never the reverse.

Updating the visible date on a page you did not actually change is a short-term trick with a long-term cost: readers notice, and a site where every post claims to be updated last week while saying the same things it said in 2022 trains people not to trust it.

When you have genuinely revised the thing: update the published-or-modified date your theme displays, make sure the structured data dateModified matches, and keep the original publish date visible too if your theme supports both. "Published 2022, updated September 2026" is more credible than a bare recent date, not less.

After you publish

1. Request indexing. Search Console → URL inspection → paste the URL → Request indexing. Hours instead of weeks.
2. Write down today's numbers. Position, impressions, clicks, CTR for that page. You cannot tell whether the refresh worked if you do not know where it started.
3. Share it again. An updated post is a legitimate reason to post the link where the topic comes up — the subreddit, the forum thread, the community that asked about it. Answer properly, link once, at the end.
4. Wait two to four weeks before judging it, and compare against the numbers you wrote down.

When to delete instead

Not every old post deserves a refresh. Be honest about three cases:

• No impressions at all after years. Nobody is searching for it, or Google does not consider you a candidate. Refreshing changes neither. Merge anything useful into a stronger post and redirect.
• The topic is dead. A guide to a discontinued product. Redirect it to the closest living page.
• Thin duplicates. Four 400-word posts on near-identical questions. One good 1,500-word page that answers all four will outrank all of them, and the three redirects pass their history to it.

Cutting weak pages is not a loss. A smaller site of pages that each deserve to rank is a stronger site than a large one carrying dead weight — and every page you delete is one less thing to keep current next year.

Read next

Why did my Google rankings drop? A 30-minute checklist

Traffic falls and the first instinct is to blame an update. Nine times out of ten it is something narrower and fixable. Here is the order to check things in, so you find the cause before you start rewriting anything.

Striking distance keywords: the traffic you already own

Positions 4 to 20 are where most small sites are quietly sitting on their next 40% of traffic. Here is how to find those keywords in a free Search Console export, work out which ones are worth the afternoon, and move them.

Turn your Search Console export into an action plan

Everyone exports the CSV. Almost nobody does anything with it. Here are the six things to pull out of that file, in order, and what each one tells you to do this week.