Which old posts to refresh, kill, or redirect
You have a decade of published work sitting on your own domain. Some of it still brings people in. Some of it argues a position you have moved on from.
A few pieces rank for terms you no longer want to be found for.
The advice you will find says to prune. Delete the dead weight, redirect the rest, watch the site get stronger. Part of that is right, and the part that is wrong can cost you links you spent years earning.
Everything here is about pages on your own domain, since the listings and bios elsewhere run on a different order.
Triage runs on two questions per URL. What does this page still earn, and what does it say about me now.
Four outcomes follow from the answers.
Clicks are the wrong first number to look at
Traffic used to be the whole triage. A page with no visits was dead weight, and the decision made itself.
That reading broke when answers moved into the results page. Pages now hold their rankings while their clicks fall. A quiet page may still be doing work your analytics never shows you.
Look at four things instead, in this order.
Links from other sites come first. They are the one item on the list you cannot rebuild by writing something better. Impressions come second, since they show the page is still being served for a query somebody types.
Citations come third: whether other writers, or an assistant answering a question, still point at the page. What the page says about you now comes fourth. It overrules the other three when the answer is bad enough.
A page with no clicks and 30 referring domains is an asset rather than a candidate for deletion.
Build the list in one sitting
Export 16 months of pages and queries from Search Console. Then export referring domains by URL from Ahrefs or whichever link tool you have.
Hand both files to Claude or ChatGPT with the four criteria above. Ask for one row per URL with a recommended action. You will have a first-pass triage of several hundred pages in the time it takes to make coffee.
Then read the shortlist yourself. An assistant can see what a page earns. It cannot tell you whether the page still sounds like the person you are now. That is the half of the decision that matters most.
The four outcomes, and how to tell them apart
| What you have | What to do | Why |
|---|---|---|
| The topic still matters and the page still earns links or impressions | Refresh it, keeping the URL | You inherit the history the page has already built |
| Three or four thin pages circling one topic | Merge into the strongest one, redirect the others to it | One solid page competes better than four partial ones |
| A page you will not maintain, with links or traffic and a close successor on your site | Redirect it one to one | The successor inherits what the old URL earned |
| No links, no impressions, no successor, and nothing you want attached to your name | Let it go, and return a clean gone status | Nothing is lost, and the index clears |
Most of a ten-year archive lands in a fifth category the table leaves out. Those pages are fine as they are, dated but honest, and not worth an afternoon.
Refreshing means new substance, not a new date
Changing the publish date and pressing update does nothing. What moves a page is information that was not there before. That means current figures, a corrected claim, or a section answering the question readers now arrive with.
Freshness does carry weight in AI answers. Ahrefs studied nearly 17 million citations across the assistants. Cited pages run about 25.7% fresher than the pages ranking organically for the same queries.
Read the rest of that study before you rewrite your whole library. Measured by last update rather than publication date, the gap narrows to 13.1%. Google's own AI Overviews lean older than the third-party assistants do.
So freshness is a tilt rather than a rule. It pays on pages where the answer has changed since you wrote it. It pays almost nothing where the answer has not.
Keep the URL when you refresh. Changing it breaks the links, the campaign destinations, and the indexed history you were trying to keep.
Redirect one to one, or not at all
A redirect earns its place when the new page answers the question the old page answered. That is the whole test.
Sending a retired page to your homepage looks tidy and buys nothing. Google treats an unrelated redirect as a soft 404. The URL drops out of the index anyway, and no signals reach the destination.
A page with no successor is better off gone. Return 410 where your platform allows it, which tells Google the removal was deliberate. Return 404 where it does not.
After a merge, repoint the internal links that fed the old pages. That step gets skipped more than any other, and skipping it wastes the merge.
The post you no longer agree with is usually worth keeping
Every long archive holds a piece the author has outgrown. The instinct is to delete it quietly.
Consider the correction instead. Add a dated note at the top. Say what you got wrong and what changed your mind, then leave the original text below it.
A visible change of mind is evidence of someone who kept reading. To a careful buyer it reads better than an archive with no history in it.
Search engines and assistants also read your record of thinking on a subject. That record is stronger with the turns still in it.
Delete rather than correct in two cases. The first is a piece wrong in a way that could hurt someone who acts on it. The second is a piece that belongs to a business you no longer run.
What pruning will not do
Cutting pages will not lift the rest of the site. Google's John Mueller has said that removing content does not make the remaining pages rank higher.
Google's own guidance on unhelpful content reads a little warmer than that. The effect is modest either way, so prune for the reader's sake rather than for a ranking.
The doubled-traffic numbers in refresh guides deserve the same caution. Most of them trace back to one software company's account of its own blog program. It describes a version of search that has changed a great deal since.
What triage does deliver is a library you can stand behind. Every page either still earns something, still represents you, or is gone.
Work down the list once. Refresh what the answer has outgrown. Merge what is thin, redirect what has a successor, and let the rest go.
Then leave the archive alone until something you wrote stops being true.
