You have a keyword you want to rank for, and you already have an old post that sort of covers it. The choice to update old page or create new page content comes up on almost every content team, and the honest answer depends on one thing: whether the keyword points to the same reader need your old page already serves. If it does, keep one destination and improve it, even if that means rewriting most of it. If it points to a different need, a different task, or a different kind of page, give it its own URL.
There's no rule from Google that settles this for you, so what follows is editorial judgment, built from search intent, performance data and what Google says about duplicate pages and URL changes. It won't give you a score or a cutoff. It will give you a few questions for any refresh vs new content call, in an order that tends to make the answer clear.
The quick version
Here is how the options line up on the axes that matter for a content refresh decision. Each one is covered in more detail below.
| Axis | Refresh or rewrite the existing page | Publish a new page |
|---|---|---|
| Underlying need | Same reader, same task | A different reader or task |
| What the search results look like | Similar page types and formats to your old page | A different page type or format than your old page |
| Overlap with what you have | The old page is the natural home | The old page would have to change what it is |
| Fit of the old outline | Fits, or can be rebuilt on the same URL | Can't fit both promises honestly |
| Page and URL history | Has value worth keeping | Little to keep, or it already earns for other queries |
What refreshing an existing page means
A refresh means you revise the page you already have so it keeps serving the same topic and the same reader. How much you change is a separate matter. You might fix a few outdated points, or you might rebuild the whole argument, and in both cases the page stays the same destination.
It helps to separate two sizes of work here. A light refresh updates facts, examples and sections that have gone stale. A rewrite on the same URL replaces the outline, the order of the argument or most of the copy, but the page keeps its address and its role. People often treat a full rewrite as the same thing as starting a new URL, and it isn't. If you have already worked out how much effort a page deserves, our guide on choosing between a high-effort and low-effort refresh covers that side of it.
One more point that trips people up is that a new keyword isn't evidence that you need a new page. People type the same need in many different ways. The question is whether the keyword stands for a separate searcher with a separate task, or whether it's one more way of asking what your existing page already answers.
What publishing a new page means
Publishing a new page makes sense when the reader, the task, the content type or the search intent is meaningfully different from what your old page should keep serving. The word "meaningfully" is doing real work there. A new angle on the same question doesn't count, and neither does a different phrasing.
A useful way to test it is to try writing the promise of each page in a sentence. If you can state both promises clearly and honestly, and they point at different people or different jobs, you probably have two pages. If the two sentences end up saying nearly the same thing, you probably have one page and a second keyword.
Compare intent first, because it settles the most cases
Search intent is the reason behind a search, what the person wants to get done or understand. The usual buckets are informational, commercial investigation, transactional and navigational, and plenty of queries mix more than one.
A more practical way to compare is the "three Cs" that Ahrefs describes in its guide to search intent. Look at the current results for your keyword and note three things about them: the content type (a blog post, a product page, a category page or a landing page), the content format (a how-to, a list, a comparison, a review) and the content angle, meaning the focus or promise that makes those results useful.
Then put your old page next to that. If the results are blog posts in the same format and angle as your old page, one continuing page usually makes sense. If the results are a different page type or format than what you have, say product pages where you have a how-to, a new page becomes easier to defend. The results are evidence about what searchers seem to expect, not a template to copy.
Some results are mixed, and when they are, you shouldn't pretend there's a single mechanical answer. Pick the audience and task the page is meant to satisfy, then check whether two pages could each serve a distinct need without simply rephrasing the same article. If you can't show that, a separate page for a wording variation isn't worth it. Our piece on writing for different search intent types goes deeper on how each type tends to look on the page.
Cannibalization is a clue, not a verdict
Most people reach for cannibalization at this point, and it's worth being careful with it. The common description is that several pages target the same or similar keywords. A more useful test asks whether the pages target the same query and satisfy the same or very similar intent, and whether having both seems to hurt how the site performs.
Shared terms alone don't establish a problem. Pages can rank for many related queries, and pages with different intents can use overlapping language without getting in each other's way. The reverse is also true, because two pages with different target keywords can still be near-duplicates in purpose and content.
Signs worth looking into include two of your URLs showing up for the same target query, your URLs taking turns in visibility over time, or two pages whose titles and outlines make essentially the same promise. These are things to investigate and not proof. A site-restricted search can help you find candidate pages, but it can return loosely related pages, so it isn't enough to diagnose anything by itself. If you do suspect real overlap, how to diagnose and fix keyword cannibalization walks through the checks.
The question to ask upstream is whether both pages could be useful to different readers or for different tasks. If the answer is yes, separate pages may be fine. If the answer is no, a second page risks becoming a competing or duplicate destination, and the fact that it can hold another variation of the keyword isn't a reason to write it.
How far the old page's structure is from what the reader needs
Here you look at two kinds of fit on their own. One is the editorial structure of the article, and the other is the value and history the page has built up.
Start with structure. An edit or refresh is the natural choice when the page's core topic, audience and purpose still match the target, and the gaps can be fixed without changing what the page fundamentally is. A clean rewrite on the same URL is reasonable when the need is the same but the old outline, framing or organization is too far from what the reader wants now. You are keeping the destination and rebuilding what's inside it. If you have already judged a page to be low-value and are weighing consolidation, redirects or deletion, that is a different decision, covered in how to decide whether to refresh, rewrite, prune or redirect a page.
Knowing when to write new content instead of updating comes down to this. A new page becomes more compelling when making the old page satisfy the new target would mean changing its core audience, task, content type or promise so much that it would stop serving its original readers. That's an editorial judgment and not a published Google threshold. A good check is whether the old promise and the new promise can sit on one page without fighting each other. Once you land on keeping the page, the steps for refreshing old SEO articles without losing rankings cover the execution.
If you aren't sure whether the old page's problems are structural, working through an audit of thin or underperforming pages can show you whether the outline is the issue or just the details.
How much the page's history should count
History is something to weigh, and it isn't a guarantee of anything. Before you give up an existing destination, look at its query and page performance and at any external links or other business value attached to it. A relevant page with useful history is a stronger candidate to keep than an identical page with no such evidence.
It's easy to overstate this, though. Neither the age of a page nor the age of a URL has been shown to be a ranking advantage in the sources behind this piece. Google does describe freshness systems for queries where people expect recent information, such as a newly released movie or a recent earthquake, but that isn't a promise of a boost for a newer date or an older address. If you are working out how freshness plays into AI answers specifically, how freshness affects AI search citations looks at that question.
URL changes are worth a mention too. Google says changing URLs can cause ranking fluctuations while it recrawls and reindexes, and that a medium-sized site may take a few weeks for Google to notice a change, with larger sites taking longer. That warning is about changing the address of an existing page. It isn't evidence that a new, distinct page will see the same delay. So fear of moving a URL shouldn't decide the editorial question for you, and keeping a URL doesn't guarantee rankings either.
What a clean rewrite gains and what it doesn't
A clean rewrite gives you room to drop a legacy outline, change the order of the argument, make the audience and promise more precise, and remove framing that no longer serves the intent. It's often the right middle choice when a small edit would leave the page structurally wrong but a new URL would split one continuing topic across two destinations.
It doesn't automatically call for a new URL, and the available sources don't show that a rewritten page ranks better than a new one would. A rewrite also doesn't fix a different-intent problem. If the new target really is a separate job for a different reader, forcing both jobs into one article can weaken the promise to each of them.
Whatever your content refresh decision turns out to be, an update should be meaningful and not just a date change. Ahrefs advises against changing only the publish date, and suggests checking whether the topic still matters to the business, whether demand remains, whether the real problem is the content or the backlinks, and whether search intent has shifted. Those points are in its guide to republishing content without just changing dates, and they are practical advice from an SEO publisher, not a Google rule. The same guide offers one example: a link reclamation post that never went above an estimated 350 organic visits a month, which was rewritten and republished in August 2024, after which traffic tripled. It's a single publisher-reported case with no comparison against a new URL, so it shows that a rewrite can coincide with growth and nothing more.
What the performance data can and can't tell you
Google Search Console's performance report shows how a page and its queries did over time, and it's a good place to start. Look at the query view and the page view together instead of reading only a keyword total, because a single overlapping query doesn't tell you the whole story for a page.
Be careful about reading a decline as a content verdict. Google's guidance on debugging traffic drops lists ranking changes, technical problems, seasonality and shifts in search interest as possible causes, and it suggests looking at longer time ranges, up to 16 months of data, to see patterns. Clicks changing without a matching change in impressions may point to how the result is presented and not to the page's content. The report describes outcomes. It can't tell you whether two pages share an intent, and it doesn't prove that a rewrite caused a change. If a page has slipped and you're not sure why, our walkthrough on diagnosing an organic traffic drop is a good next step before you decide anything.
When each option wins
Putting the pieces together, this is how the situations tend to fall out.
| Situation | Lean toward | Why |
|---|---|---|
| The page and the keyword serve the same reader and task, and the page type and format still fit | Refresh the existing page | One page can stay the clearest destination, and a second URL may repeat the same promise. |
| Same underlying need, but the outline or framing can't answer it | Rewrite on the existing URL | The job is a rebuild, but that alone doesn't justify another URL. Weigh the page's history and value first. |
| The target is a distinct task, audience, funnel need or format, and the current results support that | Publish a new page | Two separate promises are easier to serve with two pages than by forcing them into one article. |
| The old page already serves a separate intent or earns meaningful traffic for other queries | Be careful about repurposing it | One overlapping query doesn't show the full picture, so look at the whole page and query set. |
| The results are mixed or the evidence is unclear | Don't create a page just for a wording variation | Choose a specific audience and task, compare real results and existing URLs, and don't claim certainty you don't have. |
If you want one short test, first ask whether the old page and the proposed page could each serve a distinct reader need. If they couldn't, favor one destination, and then pick an edit or a full rewrite based on how far the structure is from what's needed. If they could, check that the distinction is real in the page promise and in the current results before you create the new URL.
A practical note on finding what you already have: before you commit to a new page, make sure you know every page on your site that touches the topic. DeepSmith's Content Map crawls your site, classifies each page onto a topic and funnel stage, and shows page counts per topic, which makes it faster to see whether you already have a page that could take the keyword.
Where AI search fits in
It's natural to wonder whether AI search changes any of this. Google's guidance for its generative search features says its existing SEO practices still apply, advises making useful content for people, and discourages making separate pages for every possible query variation just to influence search or AI responses. The same guidance on generative AI features says there's no requirement to break content into tiny pieces and no ideal page length. That's about Google's own generative features and not a promise about every AI system.
So the rule stays the same. Update or rewrite the page that best serves the need, and create a separate page only when there's a distinct need worth serving. A new page isn't required to earn AI citations, a rewrite won't guarantee them, and nothing says Google wants special AI-only content.
Next time you have a keyword and an old post
Run through the same short sequence for every refresh vs new content call. Look at the results and compare content type, format and angle with your old page. Check whether the old page and a new one could each serve a real, different reader. Look at the page's query and page history before you give up its address. Then choose whether to update old page or create new page content, as a refresh, a rewrite on the same URL or a new page, and write down why, so the next person on the team can see the reasoning.
If you'd like help keeping track of what you've already published as you decide, you can start a free DeepSmith trial and see how your pages map onto topics.
