Updating an old post without losing what ranked
The post that’s already working is not a blank canvas
I’ve killed rankings by “improving” a post. Not once, not on some throwaway test site, but on pages that were pulling steady clicks for two years before I touched them. The pattern is always the same: the post looks dated, someone says it needs a refresh, and the refresh turns into a rewrite. New title, new intro, new structure, half the original sections gone. Three weeks later the impressions in Search Console are down 40% and I’m trying to figure out which change did it, except now I can’t, because I changed six things at once.
An old post that ranks is not a draft. It’s a page Google already trusts enough to serve for a specific query, with a specific set of backlinks pointing at a specific URL, and internal links from across your site reinforcing it. Treat it like new content and you’re gambling all of that against a guess that your new version is better. Sometimes it is. Often the thing that was actually ranking was never the part you rewrote.
Find out what’s actually ranking before you touch anything
Open Search Console, filter to the URL, and look at the query list for the last 3 to 6 months. Not just the top query, all of them. A post you think ranks for “updating old blog posts” might be pulling half its clicks from a long-tail variant you never targeted on purpose, buried in paragraph four. If you rewrite paragraph four because it “reads clunky,” you can lose that traffic without ever touching the part you thought was the point of the page.
Do the same check in your rank tracker if you run one, and look at which sections of the post earn backlinks. A tool like Ahrefs or a manual look at referring pages will usually show you exactly which anchor text people used and which part of the page they linked to. That paragraph is now load-bearing. You can expand it, but don’t delete it or move its core claim somewhere else, because the link equity is tied to more than just the URL, it’s tied to context.
Write these findings down before you start editing. It sounds like overhead, but it’s the only way to know afterward what you can safely attribute a traffic change to.
Decide if this is a refresh or a replacement
Most posts that “need updating” fall into two very different buckets, and mixing them up is where the damage happens.
A refresh means the core angle and structure are still correct, but facts are stale: a tool got discontinued, a screenshot shows an old UI, a stat is from three years ago, a competitor you named is gone. This is low-risk work. Update the facts, leave the structure and the ranking sections alone.
A replacement means the whole premise is wrong or the page is thin compared to what’s now ranking above it. This is a real rewrite, and it carries real risk, but it’s still not a blank slate. Keep the URL. Keep the internal links pointing at it. Keep the title close enough that the query intent doesn’t shift out from under existing backlinks. You’re allowed to restructure the argument. You’re not allowed to pretend the page has no history.
If you genuinely can’t tell which bucket you’re in, that’s usually a sign the post is a refresh with one weak section, not a full replacement.
Never change the URL if you can avoid it
I’ve done this one wrong too. Changing a slug because the new one “reads better” means every backlink, every internal link, and every bit of accumulated trust on that exact URL now has to funnel through a 301 redirect instead of hitting the page directly. Redirects work, but they’re a hop, not a teleport, and consolidating signal through one takes time you don’t control and can’t schedule around. If the old slug is embarrassing or keyword-stuffed, you can decide the tradeoff is worth it, but do it as a conscious call, not a side effect of a content refresh. Most of the time the fix is to leave the URL exactly as it is and improve everything else.
Edit sections, not the whole document
Work paragraph by paragraph against your notes from Search Console. If a section is pulling queries or holds a backlink, you’re editing it for accuracy and clarity only, not rewriting its claim or its structure. If a section is dead weight, gone, that’s fair game. New sections go in as additions, not as replacements for something that already had traction.
This is slower than a top-to-bottom rewrite and it feels less satisfying. It’s also the only way to make a change and actually know afterward whether it helped, because you didn’t touch ten things at the same time.
Update the date, but be straight about it
If you’re materially updating a post, updating the visible date and the lastmod field in your sitemap is fine and honest, that’s literally what happened. What I won’t do is bump the date on a post I didn’t meaningfully touch just to look fresh. Search engines have gotten a lot better at noticing when a “updated” date doesn’t match any real content change, and even if it didn’t matter for rankings, it’s a cheap trick that erodes trust with actual readers who click through expecting something new.
Watch internal links and anchor text too
When you restructure a post, check what anchor text other pages on your site use to link to it. If three other posts link to it with the phrase “how to update old content” and you retitle the H1 to something unrelated, you’ve created a mismatch between what your own site says the page is about and what the page now says about itself. Go update those internal links to match, or better, keep the core topic phrase somewhere prominent so the mismatch never happens.
Don’t add new keyword targets you don’t need
The riskiest thing I see people do on an “update” is try to make one post rank for three more keywords than it did before, on top of keeping the old one. Sometimes that works. More often it dilutes the page’s focus enough that it gets worse at the thing it was already good at, and you end up with a post that half-ranks for four things instead of solidly ranking for one. If you have a genuinely new keyword to go after, that’s usually a new post with its own internal link back to this one, not a graft onto an existing winner.
Check the result over weeks, not days
Search engines don’t re-crawl and re-rank instantly, and a dip in the first week or two after a real content change is common and doesn’t tell you much on its own. Give it a real stretch, keep watching the same Search Console query list you pulled before you started, and compare like for like. If the queries that mattered are still showing impressions at a similar rate, the edit was safe. If one specific query cratered, go back and check whether the paragraph tied to it survived the edit intact.
There’s no fixed number of days that guarantees a clean read, and anyone telling you otherwise is guessing same as the rest of us. What you can control is not making five changes at once, so that when something moves, you have a shot at knowing why.
The boring version wins
None of this is exciting. It’s checking a report before you touch anything, editing in small pieces, leaving URLs alone, and resisting the urge to make an old post look brand new just because it feels stale to you personally. The posts on my own sites that survived a “refresh” without losing traffic are the ones where I treated the existing rankings as something to protect first and improve second. The ones I regret are the ones I treated as first drafts.
If you want more of this kind of unglamorous, specific rundown of what actually holds up in SEO, The SEO Desk covers it in more depth, along with the keyword research, link building, and technical SEO work that sits underneath posts like this one.