I run a blog that has crossed a few hundred published posts, and somewhere around post two hundred the same question started eating an hour of my week: this topic needs coverage, so do I open the old article and fix it, or start a fresh one?
For a long time I guessed. I would refresh a post, watch nothing move, then rebuild it from scratch two months later and wish I had done that first. Other times I published a shiny new article and quietly killed a page that was already sitting on page one for the exact same search. Both mistakes cost traffic. Both were avoidable.
What follows is the process I use now, after making most of the wrong calls at least once. It is not a gut feel. It comes down to a handful of signals you can check in about ten minutes inside Google Search Console, plus a few rules about search intent that Google has been unusually blunt about. Update the right posts and the math is genuinely good. Update the wrong ones and you have changed a date, nothing more.
The short version One rule you can use today Update when the old post already ranks and still matches what searchers want, and only the facts have gone stale. Write new when the angle, the audience, or the search intent is genuinely different from anything you already have. When in doubt, check whether the two would compete for the same query. If they would, you have one post, not two. |
Why it matters
Content does not hold its position on its own. The agency Animalz mapped the pattern years ago and it still holds: a successful post moves through a spike, a trough, a growth phase, a plateau, and then decay. Most of a post's traffic, close to 90 percent, arrives after it has already been live for six months, so the long tail is where the value lives, and the long tail is exactly what erodes.
That erosion is slow enough to miss on a monthly dashboard. Animalz measured an average decay of roughly 1.21 percent of organic traffic per week before a refresh. It compounds. A page pulling 1,000 visits a week at that rate lands near 531 a year later, a broken asset hiding inside a report that still looks fine because ten new posts are papering over it.
106% Average lift in organic views on old posts HubSpot refreshed HUBSPOT | 50% Of clicks lost moving from position 1 to 2 ANIMALZ | ~1.2% Average weekly traffic decay before a refresh ANIMALZ | 454% Keyword growth from full rewrites vs cosmetic edits PORTENT |
The pattern in that data is the whole game. HubSpot did not double leads by touching dates. Portent's study of 60 posts found that surface edits barely registered while full rewrites drove enormous keyword gains. The lever is the depth of the change, not the timestamp on it.
Update the old post, or write a new one
Run your topic through these two lists. Most posts land cleanly on one side. The ones that do not are usually a cannibalization problem in disguise, which is the next section.
Update the existing post when
▪ It already ranks, roughly positions 4 to 20. You are buying back rankings you already earned, plus the age and any links that URL carries. A brand new address starts from nothing.
▪ The intent is unchanged, only the facts aged. Same question, stale answer: old prices, dated screenshots, a best-tools list from two years ago, a year in the title.
▪ Traffic is decaying, not dead. A steady slide over 8 to 16 weeks, where impressions fall before clicks and position drifts toward page two, is the textbook refresh candidate.
▪ The URL has backlinks pointing at it. Links point to that exact address. A new post throws that equity away unless you redirect, and you still lose a little in the move.
▪ There is one strong page on the topic, and it is this one. Pour the effort into the page that already has momentum instead of splitting your attention and your authority across two.
Write a brand new post when
▸ The search intent is genuinely different. How to do X, X vs Y, and X pricing are three different jobs. One page cannot win all three well, so the new intent earns its own page.
▸ The topic outgrew a single post. What used to be one section now deserves a full article. Spin it out and let the parent page link down to it, a hub and spoke.
▸ Nothing you own actually fits. If there is no existing page that targets this query and intent, there is nothing to update and nothing to protect. Write it.
▸ The old post is obsolete or off brand. If a rewrite would change what the URL is about so completely that the slug no longer makes sense, a clean new URL is tidier. Then 301 the old one.
▸ You are chasing a keyword the old page was never built for. Forcing a second primary keyword into a focused page blurs it. Give the new target its own home instead.
The common trap
If a new post and an old post target the same primary keyword and the same intent, Google does not stack them. It makes them compete, and it frequently ranks the weaker of the two. Ranking authority splits across both URLs, crawl budget gets spread thin, and positions start bouncing week to week. Before you publish anything new, run one search in Google: site:yourdomain.com "your target keyword" If a close match comes back, you have a decision, and it is usually update, not add. If two pages already overlap on both keyword and intent, merge them into the stronger URL, which is not automatically the newer one, and 301 the weaker page into it. If they genuinely serve different intents, keep both, but rewrite one until the focus clearly diverges. |
What each path actually costs
Same decision, viewed through the trade-offs that usually get skipped in the moment.
| Dimension | Update the old post | Write a new post |
|---|---|---|
| Starting authority | Inherits the URL's age, links, and existing rankings | Starts at zero and has to earn everything again |
| Time to result | Often weeks, since the page already has a foothold | Usually months of indexing, links, and dwell-time signals |
| Effort per unit of upside | Lower, the research and structure already exist | Higher, everything is built from a blank page |
| Best when | Intent is the same and only the substance has aged | A new intent, audience, or keyword has no home yet |
| Main risk | A shallow edit that changes nothing but the date | Cannibalizing a page you already rank with |
| Link equity | Preserved in place | Forfeited unless you redirect the old URL |
The decision matrix
Find the row that matches what you are looking at, then read the verdict.
| Scenario | Verdict | Why |
|---|---|---|
| Same keyword, same intent, only the facts are stale | Update | Fastest win available, and it keeps every bit of authority the page has built. |
| Ranks around position 6, a competitor recently overtook it | Deep rewrite | Reclaim the position by closing the gaps the competitor opened, in place. |
| A clearly different search intent than anything you cover | New post | One page cannot serve two intents well, so the new intent needs its own URL. |
| The topic has grown too big for a single article | New post | Spin out the subtopic and link to it from the original as a hub and spoke. |
| Two of your own pages already rank for the same query | Consolidate | Merge into the stronger URL and 301 the weaker one to stop splitting authority. |
| The old post is obsolete, off brand, or thin beyond saving | New URL + 301 | A clean slate is clearer, and the redirect carries the old link equity forward. |
| Strong backlinks, weak content underneath them | Update | Never orphan a page that has earned links. Fix the substance, keep the address. |
| An annual or seasonal guide rolling over to a new year | Update | Refresh the content and move the date honestly, not just the number in the title. |
How to update a post the right way
Google has said the quiet part out loud on this one. Changing the date without changing the content is, in John Mueller's words, noise and useless. Google keeps its own record of when a URL's content actually changed, and it reads cosmetic edits like a bare still-valid-this-year note as a spam signal, not a freshness signal. Faking a fresh date is an old trick, and the search team has been on to it for years.
So a real update looks like this.
1 Re-check intent against the live results. Search your target keyword and look at what actually ranks now. If the format shifted toward listicles, comparisons, or video, your page has to match that reality, not the one you wrote for two years ago.
2 Rewrite, do not sprinkle. Portent's numbers are blunt: full rewrites moved the needle, light edits barely did. Rework the introduction, restructure the sections, cut the parts that no longer earn their place, and add the coverage the top results now include.
3 Refresh the substance itself. New data, current screenshots, updated examples, better internal links to your newer pages, and every dead outbound link replaced. Stale sources quietly undermine the page with readers and search engines alike.
4 Move the visible date, honestly. Update the last-updated date only when the change is genuinely significant, and show both the original and updated dates. Google will often surface the modified date on its own once it sees real change.
5 Keep the URL. Do not change the slug for a refresh. Changing the address forfeits the ranking history unless you redirect, and even a clean redirect costs you a little. The whole point is to keep what the URL already has.
6 Re-index, then wait. Request indexing in Search Console and give it a few weeks. Roughly half of refreshes show a clear lift within that window, so watch the trend rather than panicking on day three.
Quick reference
| Position 4 to 20, declining | Update |
| Position 1 to 3, stable | Leave it alone |
| Page has backlinks | Update in place |
| Two URLs, one query in Search Console | Consolidate |
| Traffic down 3 months straight | Refresh candidate |
| New intent, no matching URL | New post |
The verdict
After a couple of years of running this on a real content library, here is the rule I live by. Default to updating. Most of the time the page you would compete with is one you already own, and rebuilding the authority of a URL that already ranks is slower and more expensive than fixing it. I reach for a new post only when I can name a search intent the old page does not serve, out loud, in a single sentence. If I cannot say it in one sentence, I am about to cannibalize myself.
The second rule is that an update has to be a real update. I stopped treating the date as a lever the moment I understood that Google keeps its own timeline of when a page truly changed. Now the date moves only when the substance does. That one shift, treating the timestamp as a report rather than a trick, cleaned up more ranking noise than anything clever I tried before it.
Pick the page, do the work, keep the URL. The traffic tends to follow.
