Content Freshness: When to Update vs. Republish

There is a moment every site owner hits eventually. You look at a page you wrote two years ago, and it embarrasses you. The advice is stale, the screenshots show an interface that no longer exists, the tone is off. The instinct is to delete it and write a fresh one. That instinct is almost always wrong.

The reason is that a page accumulates things you can’t see by reading it. It has a URL that other sites may link to. It has a history in search engines, which have crawled it many times and formed an opinion about what queries it answers. It may show up in someone’s bookmarks, in an old newsletter, in a forum thread from 2024. When you delete the page and publish a new one at a different URL, all of that evaporates. The new page starts from zero, and the old page starts returning 404s to everyone who trusted the old address. You’ve traded an aging asset for a debt.

So the first rule is simple: keep the URL. Almost everything else about a page can change — the title, the body, the images, the date — but the URL is the page’s identity to the rest of the web. If you must change it, redirect the old address to the new one with a 301, which tells browsers and crawlers alike that the move is permanent. But treat that as surgery, not routine maintenance.

Given that you’re keeping the URL, the real question is smaller than it first appeared: do you update quietly, or do you republish loudly? The distinction is worth making precise. An update means you fix what’s wrong — correct facts, replace dead examples, tighten the prose — and the page goes on living as itself. Republishing means the revision is substantial enough that you change the visible date, maybe announce it, and present the page as effectively new work at the old address.

I’d draw the line at whether a past reader would learn something by reading it again. If someone read the page in 2024 and your changes only fix what was broken, that’s an update. Change the content, leave the presentation of its age honest, and move on. If your changes are large enough that the 2024 reader would get real new value — new sections, a changed conclusion, rethought advice — then republishing is legitimate. You’re not gaming anything; the page genuinely is new work.

What you should not do is the thing a lot of sites do, which is bump the date without changing anything. It’s tempting because dates are visible in search results and readers prefer recent ones. But a reader who clicks through expecting 2026 thinking and finds 2023 thinking notices, and the noticing costs you more than the click gained you. Freshness that exists only in the metadata is a small lie, and small lies on websites compound the way small lies in person do.

There’s a related trap worth naming: freshness for its own sake. Not every page decays. A page explaining what a canonical tag is doesn’t need an annual revision, because canonical tags haven’t changed. Pages decay at the rate their subject changes. Pricing pages decay in months. Tutorials tied to a specific tool version decay when the tool does. Essays about how people think decay slowly or not at all. If you schedule updates on a calendar instead of watching the subject, you’ll waste effort polishing pages that were fine and miss the ones that quietly rotted.

How do you find the rotted ones? The mechanical signals are the easiest. When I run automated audits on sites — this is the kind of thing I built GazeSite to do — the stale pages tend to announce themselves through side effects. They link to things that no longer exist. They reference products by old names. Their titles and meta descriptions were written for a positioning the company abandoned two pivots ago. A page can have perfect grammar and still be visibly dead if you look at what it points to and what it promises.

The harder signal is intent drift. A page ranks, or used to rank, for a query — but what people mean by that query has shifted. Five years ago someone searching for a term might have wanted a definition; now they want a comparison, because the field matured and everyone knows the definition. The page didn’t change, but the question did. This is the case where a heavy revision is most worth it, because you already have the URL’s history working for you; you just need the content to answer the question as it’s now asked.

One last thing about deletion, since I said the instinct is almost always wrong: almost. Some pages should die. Thin pages that exist only because someone once thought more pages meant more traffic, duplicate near-copies of each other, announcements of events long past that no one will ever search for again. Killing these is fine, and pruning a bloated site is genuinely useful, because it concentrates attention — yours and the crawlers’ — on the pages that matter. But even then, prefer redirecting a dead page to its closest living relative over letting it 404. A 404 is an answer to nobody.

The general principle underneath all of this is that a website is not a feed. Feeds are append-only; you publish and move forward. A website is more like a garden, and most of the work of a good garden is tending what’s already planted. The sites that feel authoritative are rarely the ones publishing most often. They’re the ones where you can land on a page from four years ago and find it still true, because someone went back and made it true.