A content audit is a full review of everything published on a site, checked against what it’s actually doing (traffic, rankings, backlinks, business value) and sorted into one of four calls: keep it, update it, merge it into something stronger, or delete it.

That’s the whole job. Everything below is how to do that job without it turning into either a week lost to a spreadsheet or a purge you regret in March.

Most sites don’t need one because something is broken.

They need one because the site has quietly accumulated pages nobody has looked at since the year they were published, and the pages actually driving traffic are, more often than not, the ones getting the least scrutiny.

That second part is the real reason to run one, and it’s worth sitting with before you open a spreadsheet.

A content audit is a decision engine, not a cleanup checklist

Here’s where the term gets muddy, and Nielsen Norman Group draws the line cleanly: an inventory lists what exists. An audit judges it.

Confuse the two and you end up with a very tidy spreadsheet of URLs and no actual decisions attached to any of them, which is busywork wearing an audit’s name.

Term	What it actually is	Output
Content inventory	A complete list of every URL, with basic facts attached (title, author, date, word count)	A spreadsheet. Nothing decided yet.
Content audit	The inventory plus performance data, judged against a goal	A call on every page: keep, update, merge, or delete
Content strategy	What you do next, informed by what the audit found	A prioritized plan, not a one-time cleanup

Scope, before anything else: every published blog post, every landing and service page, and every category or hub page an outside visitor could land on from search.

Product pages on an ecommerce site technically qualify too, though they usually get their own pass with different thresholds (a product page with modest traffic but real conversions is a keep, no matter what its pageview count looks like next to a blog post).

Your best-performing page is the one most overdue for this

Success shields content from scrutiny. Nobody opens the page that’s ranked in the top five for three years and asks whether the advice inside it still holds, because it’s clearly working.

That’s exactly the page running on the most inherited assumptions: the competitor set it beat in 2023 isn’t the competitor set it’s up against now.

The screenshots are three redesigns out of date, and the internal links pointing to it were built for a site structure that no longer exists.

Say your top landing page has held position four or five for a keyword since 2023, and nobody on the team has opened it in months because why would you touch something that works. That’s the page to open first, not last.

Google’s helpful-content systems are built to reward pages that show ongoing care, real expertise, and a point of view that’s still current, not pages that happened to get there once and were left alone.

A page that hasn’t been touched in two years is quietly telling both readers and search engines that nobody’s minding it, even while it still ranks.

Your move: before you audit anything else, pull up the three pages driving the most organic traffic right now and check the byline date.

If any of them predate your last product launch, pricing change, or major industry shift, that’s your starting point, not an afterthought at the end of the list.

Before you touch a spreadsheet, pick one goal

The goal decides your thresholds later, which is why skipping this step is the single most common way an audit stalls out at “well, some of these pages seem fine, I guess.”

Without a stated goal, every scoring decision downstream is a guess dressed up as a metric.

• Recover declining traffic: the goal is finding which pages lost the most ground over the last 12 months and why, before you touch anything that’s still stable.
• Fix keyword cannibalization: the goal is finding every case where two or more of your own pages compete for the same query, then merging the weaker one into the stronger.
• Prep for a redesign or migration: the goal is deciding what’s worth carrying to the new site and what finally gets left behind, before a developer migrates every URL by default.
• Raise the site’s overall quality bar: the goal is identifying thin, outdated, or low-value pages that drag down how a search engine or a reader judges the rest of the domain.

Pick one as the primary goal even if two or three apply. The others will surface anyway once you’re in the data; they just won’t be what decides your keep-or-cut line.

Build the inventory: what to log for every URL

Get the full list of live URLs first, before any performance data touches it.

A free crawler (Screaming Frog’s no-cost tier handles up to 500 URLs) or an export straight from your CMS both work; an XML sitemap export is the fastest option if your sitemap is actually kept current, which, worth checking, it may not be.

Log these fields for every row before you add a single traffic number:

• URL and page title, exactly as they appear live, not as they appear in a CMS draft that never got renamed.
• Publish date and last-modified date, separately. A page “updated” by a template change isn’t the same as one with genuinely refreshed content, and your CMS usually can’t tell the difference on its own.
• Author and content type (blog post, landing page, product page, category hub), so you can filter and compare like against like later.
• Category or topic, even a rough one. This is what lets you spot patterns by section instead of judging 400 rows one at a time.
• Word count, useful mostly as a thin-content flag, not as a quality signal on its own.

When a URL doesn’t obviously fit a category, don’t guess from the title alone. Check which queries it actually ranks for in Search Console first; the real answer is usually in the data, not in what you assumed the post was about when it published.

Pull the metrics that actually tell you something

Three free sources cover almost everything you need. None of this requires a paid tool to start.

What Google Search Console tells you

Pull the Performance report for the last 12 months, filtered by page, and export clicks, impressions, average position, and CTR for every URL. Four signals worth flagging as you go:

• Impressions dropping while position holds steady: usually means the query itself is losing volume, not that the page is doing anything wrong.
• Position sliding from page one to page two: the clearest single decay signal there is. A page on page two of results is invisible to most searchers regardless of what it says.
• CTR well below what its position should earn: the ranking is fine, the title and meta description aren’t earning the click.
• Two or more of your own URLs ranking for the same query: keyword cannibalization. Neither page is likely to outrank the other while they’re splitting the same relevance signal.

What GA4 tells you

Sessions and engaged sessions by landing page, year over year, plus whatever conversion event actually matters to the business: a form fill, a newsletter signup, an add-to-cart.

A page with respectable traffic and zero conversions isn’t automatically a failure, but it’s worth separating from a page that both drives traffic and does something with it.

The same conversion metrics worth tracking after a content refresh are the ones worth pulling here first, so you have a real before number to compare against later.

What a backlink tool tells you

Ahrefs or Semrush, whichever the team already has access to, exported at the URL level for referring domains and a page-level authority score.

This is the check that stops a good delete decision from becoming a bad one: a page with no traffic can still be holding real backlinks, and deleting it without a redirect throws that authority away instead of routing it somewhere useful.

Worth calling out: if the crawl also turns up broken redirects, orphaned pages with no internal links pointing to them, or a pile of indexed pages that shouldn’t be, that’s a separate technical SEO problem sitting on top of the content one.

A technical SEO audit is the right tool for that layer; it isn’t something a content-decision spreadsheet fixes on its own.

The four moves: keep, update, merge, or delete

Every URL in the inventory ends up in exactly one of four buckets. The data you just pulled is what decides which one.

Keep

Stable or improving traffic, no obvious content gaps against what’s currently ranking above it, no cannibalization from another page on your own site. Leave it alone, but note the date you checked it, because “keep” has an expiration date too.

Revisit it at the next audit, not never again.

Update

Traffic or position is declining, but the query still has real demand and the page still has a reason to exist.

This is the largest bucket on most sites and the one that actually moves numbers, since you’re improving something that already has some authority instead of starting from nothing.

If the page is thin because there was never enough original material to begin with, that’s a different problem than staleness, and a real system for generating new angles is the fix, not a light copyedit.

Merge and redirect

Two or more of your own pages compete for the same query, or several thin posts each cover one fragment of a topic that deserves a single, thorough one.

Pick the stronger page as the survivor (more backlinks, better existing position, or simply the more complete draft), fold the useful material from the others into it, and 301-redirect the losers to it.

That’s not the same page deleted twice; the URL disappears, but the authority it built doesn’t.

Delete

Zero organic traffic, zero backlinks, and no business reason to keep it around: an old event announcement, a duplicate draft that got published by accident, a promotion page for something that no longer exists.

Redirect it to the closest genuinely relevant live page instead of leaving a dead 404, and confirm nothing internal still links to it before it goes.

“Delete” showing up on a first pass doesn’t mean removing it that afternoon; every page in this bucket still gets a human look before it’s gone for good.

Why pages that used to rank quietly stop ranking

Content decay is the pattern where a page climbs, holds a position for a while, then slides, without anyone changing a single word on it. Nothing on the page broke.

The world around it moved: competitors published newer, more thorough answers, the information itself aged, algorithm updates shifted what “helpful” looks like, and internal links pointing to the page thinned out as the site added newer content elsewhere.

The recovery after the refresh point isn’t automatic and it isn’t instant.

It happens because the update actually addresses what changed (newer examples, a more complete answer, better alignment with what’s now ranking above it), not because republishing alone resets some invisible clock.

A date change with no real content change underneath it is the kind of move search engines have gotten reliably better at ignoring.

Mistakes that quietly turn an audit into a purge

• Deleting on traffic alone. A page with zero sessions can still hold backlinks or rank for a query your reporting window missed. Check both before anything gets removed, not after.
• Treating delete as the default instead of the last resort. It’s the smallest of the four buckets on a healthy site, not the biggest.
• Skipping categorization. Without a topic or section tag on every row, you’re judging 400 individual pages instead of spotting the three sections actually dragging the site down.
• Updating two cannibalizing pages separately. That doubles the work and fixes nothing, since they’re still splitting the same relevance signal after both get refreshed.
• Running it once and calling it done. An audit is a recurring system, not a one-time cleanup project. See below.

How often to actually run this

Roughly a third of marketers audit their content twice a year, another quarter do it once, and about one in five do it more than three times.

Semrush, 2023 State of Content Marketing Report

Once a year is table stakes: a full pass through every URL, minimum, on any site that publishes regularly.

Twice a year holds up better on a fast-moving site or a competitive vertical, where a page can lose real ground in six months without anyone noticing. Neither cadence works as a one-time event.

The page you decided to “keep” this quarter is a candidate for “update” a year from now, quietly, the same way the top performer from the section above got there.

Your move: put the next audit on the calendar right now, for six or twelve months out, before this one’s momentum runs out and it quietly becomes a one-time project instead of a system.

The ACT step is where most of the real work happens: rewriting, merging thin pages, or redirecting the ones beyond saving.

Frequently asked questions

What’s the difference between a content audit and a content inventory?

An inventory is just the list: every URL with basic facts attached. An audit adds performance data and judges each page against a goal, ending in an actual decision for every row: keep, update, merge, or delete.

How often should you do a content audit?

At least once a year for any site that publishes regularly, twice a year for fast-moving or competitive verticals. It works best as a recurring system on the calendar, not a one-time cleanup project.

What tools do I need for a content audit?

Google Search Console and Google Analytics 4 cover most of it for free. A crawler like Screaming Frog’s free tier builds the URL inventory, and a backlink tool like Ahrefs or Semrush checks whether a low-traffic page still holds real authority before it gets deleted.

Should I delete or redirect a low-traffic page?

Redirect it if it has any backlinks, any historical traffic, or any internal links still pointing to it. Only remove it outright with no redirect if it has none of those and no business relevance at all.

How long does a content audit take?

Building the inventory and pulling metrics for a few hundred URLs typically takes a few days. The analysis and decision pass takes longer and scales with site size; budget it as an ongoing project with milestones, not a single afternoon.

The pages quietly accumulating from the first paragraph of this guide don’t announce themselves. They just sit there, ranking a little worse every quarter, until an audit is the thing that finally asks each one to justify its place on the site.

---