What to Do With Old Blog Posts: Update, Merge, Redirect, or Delete

Published July 27, 2026

Part of: Traffic & Audience — our full guide on this topic.

Disclosure: Some links below are affiliate links. If you sign up through them we may earn a commission at no extra cost to you. We only recommend tools we'd genuinely suggest to a friend. See our full disclosure.

Every guide on this site takes a piece of content forward. Pick the topic, plan it, write it, batch it, repurpose it.

They all end at publish. The post does not.

Two or three years in, you are no longer running a content plan — you are running an estate. Posts that name a tool that renamed itself. Posts with screenshots of an interface that no longer looks like that. Two posts that cover almost the same thing because you forgot you had written the first one. A post about a service that shut down. And the whole pile sits there with no process attached to it, because the entire content conversation is about what to publish next.

The advice that fills the gap is almost always “prune ruthlessly” — and it has been imported from a place where it is true for a reason that does not apply here.

That is worth untangling first, because it decides your default. Then there is the more useful question underneath: an old post can get five different verdicts, they are not the same action, and most people only know two of them.

The quick version

• “Prune ruthlessly” is email-list advice wearing SEO clothes. Dead subscribers damage every other email you send. A quiet post does not work that way.
• Look before you judge. Impressions, clicks and average position for that URL, free, in your own search console. Zero impressions and low impressions mean opposite things.
• There are five verdicts, not two: leave it alone, update it, merge it, redirect it, delete it.
• Leave it alone is a real answer and it is the right one most often. An accurate post that nobody visits costs you nothing.
• Update means change the content. Moving the date without changing the post is not an update, it is a lie with a timestamp.
• When you merge, merge into the stronger URL — the one with the visits and the links — not the one with the nicer title.
• A redirect must land somewhere the visitor actually wanted. If there is no genuine replacement, letting the URL 404 is more honest than sending them somewhere random.
• Do it once a year, with a bounded list. An open-ended archive review never finishes.

First: where “prune ruthlessly” actually comes from

If you have read anything about maintaining a list of things you publish or send, you have met the ruthless-pruning school. Cut the dead weight. A smaller, stronger set beats a big weak one. It is stated as a general principle of running any collection.

It is not a general principle. It is a specific, correct piece of email advice, and it is correct for a mechanical reason: inactive subscribers drag down your sender reputation, and sender reputation is a single shared meter. Every address that never opens is making it slightly harder for your next email to reach the people who do open. That is why pruning a list genuinely raises your open rates — you are not tidying, you are removing something that was actively taxing everything else.

An archive of blog posts has no equivalent shared meter that a quiet post drains. A post that nobody reads is not making your good posts harder to find, in the direct mechanical way a dead address makes your emails harder to deliver. The two situations look alike — a big collection, some of it inert — and the fix from one has been carried over to the other on the strength of the resemblance.

This matters because it sets your default. If you believe old posts are actively harmful, deleting is the safe move and you will delete finished work on a hunch. If you accept that a quiet post is mostly just quiet, then the burden of proof flips: you need a specific reason to remove something, and “it gets no traffic” is not one on its own.

There are specific reasons. They are just narrower than the advice implies.

What an old post actually costs you

Four real costs, and one that gets invoked far more often than it applies.

It can be wrong. This is the only cost that is genuinely urgent. A post that recommends a tool that shut down (a sweep worth doing deliberately on the week one of them actually does, while you still remember every place you mentioned it), quotes a price that has doubled, or describes a process that changed is not neutral — it is actively misinforming someone who trusted you enough to read to the end. Every day it stays up, it does that again.

It can be off-topic for what you now do. Businesses drift. If a third of your archive is about something you have quietly stopped doing, a new reader landing there gets a confusing picture of what you are. That is a positioning cost, not a traffic one, and it is real.

It can point at things that no longer exist. Old posts are full of links, and links rot. An internal link to a page you deleted, a link to a product you retired, a call to action for an offer that is gone. The post still reads fine; the exits are all dead ends.

It can be embarrassing. Not “I would write this better now” — everyone feels that about everything. Genuinely embarrassing: a take you no longer hold, a tone you have grown out of, a claim you would not make today. If you would wince at a prospective client reading it, that is a cost.

And the one people wrongly invoke: it makes the rest of your site worse. Maybe, in narrow cases — mostly where you have several posts genuinely competing for the same query, which is a real thing and covered below. But as a blanket claim about any low-traffic page, it is the reason people delete work they should have left alone. If you cannot describe the mechanism by which this specific post is hurting that specific post, you are tidying.

Notice what is not on the list: hosting cost (a text page costs effectively nothing), and your own discomfort at seeing a long list of posts with small numbers next to them. That second one is not a cost to your business. It is a cost to your mood, and it is the single most common reason archives get culled.

Before you decide: three numbers and one fork

You cannot judge a post from memory. You need to look, and the looking is free — connect your site to a search console (Google’s and Bing’s are both free, and if you have set up traffic tracking you are most of the way there already) and pull three numbers per URL:

• Impressions — how often the page appeared in results at all.
• Clicks — how often anyone chose it.
• Average position — roughly where it sat when it did appear.

Those three tell you very different stories, and the fork that matters most is at the very start.

A note on reading these three numbers, since they are the same three that diagnose a site-wide problem: impressions, position and clicks only mean something together. Impressions down with position steady is falling demand; position down is lost ground; both steady with clicks falling is the result page changing around you. That distinction matters here per-post and it matters much more when traffic drops across the whole site, where reading one number alone is the usual cause of a wrong diagnosis.

Zero impressions. The page is not appearing in search results at all. Before concluding anything about its quality, check whether it was ever indexed — a page that search engines never picked up has not been judged and found wanting, it has been missed. That is a technical problem with a technical fix (is it in your sitemap, is it linked from anywhere on your own site, is it accidentally marked no-index), and it has nothing to do with whether the writing is any good. Deleting an unindexed post because it gets no traffic is deleting an experiment that never ran.

Impressions but almost no clicks. People are seeing it in results and choosing something else. That is a title and description problem far more often than a content problem, and it is the cheapest fix in this entire article — you can rewrite a title in two minutes and you cannot rewrite a post in two minutes.

Impressions, clicks, and a position that has slid over time. Something changed around the post: better answers appeared, or the post aged while the topic did not. This is the classic update candidate and usually the highest-value work in the whole pass.

Steady clicks. Leave it alone. Your best-performing old post is the one people most want to “improve”, and a rewrite that changes what the page is about can lose you the thing it was working for. If you touch it at all, correct the facts and fix the dead links. Do not restructure it.

One honest caveat about all of this: search data lags, and a post published recently has not had time to show you anything. Judge posts that have been up long enough to have a record, and leave the recent ones to run.

The five verdicts

Here is the part the “delete or keep” framing hides. An old post can get five different outcomes, they cost wildly different amounts of effort, and picking the cheapest one that solves the actual problem is most of the skill.

1. Leave it alone. Accurate, on-topic, not competing with anything else. Traffic is irrelevant to this verdict. This is the correct answer for the large majority of an archive and it is free.

2. Update it. The post is right about the topic and wrong about the details. The bones are good; the facts, examples, links or screenshots have aged. Keep the URL, keep the title if it still fits, change the content.

3. Merge it. Two or more posts are doing one post’s job. Combine the good material into one, redirect the others to it.

4. Redirect it. The post should not exist any more, but there is a genuine replacement for the person who was looking for it. The URL keeps working and takes them there.

5. Delete it. The post should not exist and there is nothing to send anyone to. Rare, and worth being honest that it is rare.

The mistake almost everyone makes is jumping from verdict 1 to verdict 5 with nothing in between, because the two middle options take actual work and the two ends take none.

Updating: what it means and what it doesn’t

An update is a change to the content. That is the whole definition, and it needs saying because the most widespread version of this advice is to open old posts and change the date.

Changing the date is not an update. It is a claim, made to a reader, that this page reflects the present. A visitor who sees “updated this month” above a screenshot of an interface that was redesigned two years ago has learned something specific about you, and it is not what you wanted them to learn. The trust you spend there is worth more than any presentation benefit.

What a genuine update looks like, in the order worth doing it:

• Fix what is wrong. Prices, tool names, features that moved or vanished, steps that changed. This is the part that actually matters.
• Fix the exits. Check every link. Internal links to pages you have moved or removed, external links to sites that have gone, and — the one everybody misses — the call to action at the bottom. A three-year-old post is often still promoting an offer that no longer exists.
• Add what you now know. The genuine advantage of updating over rewriting from scratch is that you have three more years of experience about this exact topic. Put some of it in.
• Cut what has not aged. Sections that were padding when you wrote them are still padding.
• Then, and only then, the date. If you changed a lot, move it. If you fixed two links, leave it. A visible “last checked” line alongside the original publish date is more useful than either, because it tells a reader both things they want to know: this is old, and it has been looked at.

One rule that saves a lot of pain: do not change the URL. If the post has any visits, links, or history at all, the address is the most valuable thing about it. A better slug is not worth the trade.

Merging: how not to lose the one that works

Merging is the verdict that most often gets done backwards.

You have two posts on nearly the same topic. Before doing anything, confirm it is a real problem: look at your search data and see whether both URLs are actually picking up the same queries, and whether both are stuck as a result. Sometimes two similar-sounding posts genuinely serve different readers and are both fine. In that case the fix is not a merge — it is making each one clearer, in its title and its opening paragraph, about who it is for.

If it is a real overlap, then:

1. Pick the survivor by evidence, not by preference. The survivor is the URL with the visits, the inbound links and the history. Not the one with the title you like better. If you prefer the weaker post’s title, change the survivor’s title — titles are editable, a page’s accumulated history is not.
2. Move the genuinely good material across. Not all of it. The point of a merge is one good page, not one long page. If the second post’s only unique contribution is three sentences, three sentences is what moves.
3. Redirect the loser to the survivor. Not to your homepage, not to your blog index — to the specific page that replaced it.
4. Fix your own internal links. Every place on your site that pointed at the merged-away post should now point at the survivor directly. A redirect covers you, but a direct link is cleaner and it is a five-minute job.

Removing: where a dead URL should point

This is where an old article behaves differently from an old product, and the difference is worth being explicit about, because the instinct carries over wrongly.

When you retire a digital product, you should essentially never let the URL die. Someone who bought it has a relationship with that specific address — a receipt, a bookmark, a download link they expect to keep working — so the page has to keep saying something, even if what it says is “this is retired, here is what I would recommend instead”.

An article’s visitor has no relationship with the URL. They arrived from a search result or someone else’s link, and what they wanted was the topic. So the test is different and simpler:

• Is there a page on your site that genuinely serves the person who was trying to reach this one? Redirect to it. That reader gets what they came for and nothing is lost.
• Is there not? Then let it 404, and make sure your site’s “page not found” screen offers a real way onward — your main topic pages, a search box, a link home. A plain “not found” plus a helpful page is more honest than a redirect to something unrelated.

That second point runs against the common instinct, so it is worth defending. A redirect to a page the visitor did not ask for is not a kindness. They clicked expecting one thing, landed on another, and now have to work out what happened — and the most common version of this, redirecting everything you remove to your homepage, is the worst one: someone who wanted one specific answer and lands on a generic front page simply leaves.

Redirects also accumulate. Every one is a small permanent rule you are carrying, and a site with a hundred of them pointing at loosely-related pages is harder to reason about than a site with twenty good ones and some honest 404s.

It is worth being clear that this rule is specific to article URLs, because two neighbouring cases work differently and for stated reasons. A retired product’s URL must never die, because a buyer holds that exact address as a receipt or a download link. And a social handle you stop using should not be closed at all, because unlike a URL it does not simply stop existing — it goes back in the pool for someone else to take. The question in all three is the same: what happens to the address after you let go of it.

When deleting is actually right

The honest list is short:

• It is factually wrong and you are not willing to fix it. Both halves matter. “Wrong” alone is an update. “Wrong and I will never get to this” is a delete.
• It is off-topic for what you now do, with no replacement, and no way to make it fit.
• You would not want it associated with you — a claim you no longer stand behind, or work you would not show anyone.
• It is a genuine duplicate whose material has already moved into the survivor. This is the tail end of a merge, and it should still be redirected rather than deleted outright.

Notice that none of those are about traffic. Traffic tells you where to look; it does not deliver the verdict. And before you delete anything with a history, keep a copy of the text somewhere — the effort of writing it is already spent, and a deleted post is often the fastest raw material for a newsletter issue or a repurposed piece later.

One reason that is not on the list, despite being the most common one people act on: the business it belonged to is over. A finished archive that still helps someone costs a domain a year to leave standing, and it stays a credential for whatever you do next. If keeping it running is part of what you are trying to escape, there are quieter ways to stop than deleting the work.

Making it a policy instead of an anxiety

The reason most people never audit their archive is that “review all my old posts” is unbounded, and unbounded tasks do not get done — they get started, half-finished, and then carried around as guilt.

Bound it:

• Once a year, on a date you pick. Not continuously. Not “when I get to it”.
• A fixed scope: your ten most-visited posts, plus anything you already know is out of date, plus anything that names a specific tool, price or platform. That is where the actual risk lives.
• A time box. One morning. What you do not get to waits for next year and nothing bad happens.
• Write down what you decided, even one line per post. Next year’s you will otherwise re-open the same twenty posts and re-make the same decisions from scratch. This is the same logic as writing down any process you repeat — the decision is worth more than the memory of it.

And one thing to check while you are in there, because it is the highest-value five minutes of the whole exercise: does each of your most-visited old posts still have a working way for a reader to do something? An old post pulling steady visits with a broken call to action, or none at all, is the closest thing to free money a small site has. It already has the attention. Fixing where that attention goes is far cheaper than earning more of it, and if the post has no email capture on it at all, that is the fix worth making before any rewrite.

Five mistakes worth avoiding

Deleting for hygiene. Covered at length above, and still the most common one. If you cannot say what the post is costing you, leave it up.

Rewriting your best post. The temptation to improve a page that is working is strong and the downside is asymmetric. Correct facts, fix links, stop.

Changing URLs during a cleanup. A tidy slug is worth almost nothing. A URL’s history is worth a great deal. This is the single most expensive avoidable mistake in an archive pass.

Merging into the wrong post. Evidence picks the survivor, not taste.

Turning it into a permanent project. An annual pass you actually finish beats a rolling review you never do. The archive is not an emergency; it has been sitting there for years and it will survive until your date comes round.

The decision, in order

1. Has this post been up long enough to have a record? No: leave it, come back next year.
2. Is it wrong, or off-topic for what I do now? Yes: fix it or remove it — this is the urgent lane, and it is not a traffic question.
3. Look at impressions, clicks, position. Zero impressions: check it was ever indexed before judging it.
4. Is another post on my site doing this one’s job? Yes: merge into the stronger URL, redirect the weaker.
5. Is it accurate and on-topic? Yes, whatever the traffic: leave it alone. Verdict complete.
6. Does it need content changes? Update it — facts, links, call to action, then the date if the change was real.
7. Should it stop existing? Redirect it to the genuine replacement. If there isn’t one, let it 404 honestly.
8. Write down what you decided, so next year starts where this year stopped.

The bottom line

An archive is not a list of everything you have ever published — it is the set of pages you would still stand behind if a stranger read them today. Keeping it that way is ordinary maintenance, and it is a much smaller job than the guilt around it suggests.

Start by dropping the borrowed instinct. Pruning is right for an email list because dead addresses tax every message you send; a quiet post does not tax anything, so “no traffic” is where you start looking, not what you conclude. The posts that genuinely cost you something are the ones that are wrong — stale prices, dead tools, links to nowhere, a call to action for an offer that no longer exists — and those are worth fixing whether they get a thousand visits a month or eleven.

Then use the middle verdicts. Most old posts need updating or merging, not deleting, and both of those keep the value you already built instead of throwing it away for tidiness. Protect the URLs. Protect the post that is working. Bound the whole thing to one morning a year.

And then go back to publishing, which is still where nearly all the return is — with the difference that what you publish this year will be looked after, instead of quietly ageing into something you would rather a stranger did not find.