A content audit is a page-by-page review of everything your site publishes, ending in one decision per URL: keep, refresh, merge or prune. On a large site it runs in five moves. Build a full inventory from your crawl, sitemaps, Search Console and analytics. Join 12 months of metrics to every URL. Judge page families before single pages. Ship the redirects, then give every page an owner and a review date so the site doesn't drift back. You don't end with a report. You end with changes.
I'm Brenden House, and I run Searchmaxxed, an SEO and AI SEO agency. A content audit sits inside every SEO audit we run, because a big site's worst problems are usually pages it already has.
In summary
- A content audit ends in decisions, not scores. Every URL leaves with one of four actions, an owner and a date.
- Large sites need the API. Search Console's interface exports 1,000 rows, while its API returns up to 25,000 rows per request and pages through the rest.
- Audit templates first. One template fix can repair thousands of pages.
- Check your data before you trust it. Bot traffic and broken conversion events can make a dead page look alive, or a good page look dead.
- Pruning to look fresh won't help. Google says so plainly. Remove pages because they're wrong, duplicated or useless.
What is a content audit?
A content audit is a structured review of a site's published content against its goals: search traffic, conversions, accuracy and usefulness. It answers one question for each URL. What happens to this page now?
It starts from a content inventory, the full list of what you have: every URL with its title, type, owner and dates. The audit adds judgement. It scores each page on performance and quality, then gives it an action that someone on your team can carry out.
| Content inventory | Content audit | |
|---|---|---|
| Question it answers | What do we have? | Is it working, and what do we do with it? |
| Output | A list of every URL with its facts | One action per URL: keep, refresh, merge or prune |
| Data | Crawl, sitemaps, CMS export | Inventory plus Search Console, GA4, links, conversions, AI citations |
| How often | Kept live, ideally from the CMS | Full audit yearly; page families on a rolling quarterly cycle |
| Who owns it | Web or content operations | SEO and content leads, with product, legal and engineering signing off |
An SEO content audit is the same work, weighted toward search: rankings, clicks, cannibalisation, internal links and crawl health. A technical SEO audit covers the machinery underneath. Think rendering, indexing and site speed. Our guide on what an SEO audit covers sets out where the two overlap and where they part ways.
When should you run a content audit?
Run a full content audit once a year on a large site, and audit page families on a rolling quarterly cycle. Run one straight away before a migration, after a merger, or when organic traffic falls for two or three months and nobody can say why.
The common triggers on enterprise and B2B SaaS sites:
- A migration or replatform. Why move 20,000 pages you'd delete anyway? The audit decides what moves, which shrinks the redirect map before our site migration SEO checklist takes over.
- Flat or falling organic traffic across one section.
- A product or positioning change. Old feature names, retired plans and dead integrations keep ranking, and they keep confusing the buyers who land on them.
- An acquisition. You inherit two blogs and two help centres, with overlapping articles on the same topics and nobody sure which version is current.
- Years of volume publishing. Many teams spent a few years publishing to fill a calendar, and those pages now compete with each other for the same searches.
- AI answers naming a competitor for questions your pages should own.
How large is large? Google's crawl budget guide sets the bar at 1 million or more unique pages changing weekly, or 10,000 or more changing daily. Most B2B sites sit well under that, so crawl budget isn't their problem. Relevance is: too many weak pages chasing the same searches.
How do you run a content audit on a large site?
You run it in seven steps. The first four gather and sort: goal and scope, inventory, metrics per URL and page families. The last three act: one decision per URL, redirects and removals, then governance so the gains hold.
| Step | What you produce | Typical owner |
|---|---|---|
| 1. Goal and scope | One sentence of intent, the sections in scope, the date window | Head of SEO or content |
| 2. Inventory | Every URL, deduplicated and normalised | SEO, with web operations |
| 3. Metrics per URL | One row per URL with 12 months of data joined | SEO or analytics |
| 4. Page families | Each URL tagged by template and topic | SEO |
| 5. Decisions | Keep, refresh, merge or prune for every URL | SEO and content, with product and legal |
| 6. Redirects and removals | A redirect map and a removal list, tested | Engineering |
| 7. Governance | An owner and review date per URL, plus publishing rules | Content operations |
Step 1: Set the goal and the scope
Start with one sentence and a number. "Lift demo requests from organic search on the product and solutions pages by a fifth this year" is a goal. "Improve our content" isn't.
Next, set a scope you can defend to the people who own each section. On a large site I'd audit every indexable HTML page, plus PDFs and docs if they pull organic traffic. Use a 12-month data window. A page that sells in March looks dead in September, and a three-month window would have you prune it.
Step 2: Build the inventory
Pull URLs from at least four places, because each source misses pages the others catch:
- A crawl finds what your internal links can reach. Screaming Frog's free version stops at 500 URLs, so a large site needs the paid licence or a cloud crawler.
- XML sitemaps list what you've asked Google to index.
- Search Console shows the pages Google has seen, including orphans that no link on your site reaches anymore.
- GA4 landing pages catch old campaign pages.
- The CMS export gives you the owner, author, publish date and template for each page, which you'll need for governance later.
Search Console is where data goes missing. Google's performance data deep dive is blunt about it: "The maximum you can export through the Search Console user interface is 1,000 rows of data". The Search Analytics API returns up to 25,000 rows per request. Use it.
Normalise every URL before you join anything. Lowercase the host, settle trailing slashes, strip tracking parameters and pick one protocol. Skip this step and one page can show up as four rows, each holding a quarter of its clicks. All four look weak.
Step 3: Pull the metrics for every URL
Join one row of data to each URL using the metrics below, and note what each one decides for the page:
| Metric | Source | What it tells you |
|---|---|---|
| Clicks and impressions, 12 months | Search Console API | Whether search finds the page, and whether people choose it |
| Average position and top queries | Search Console API | Whether a refresh could move it, and which search it serves |
| Organic landing sessions | GA4 | Whether the visit lands and stays |
| Conversions from the page | GA4, CRM | Whether it brings pipeline: demos, trials, contact forms |
| Referring domains | Ahrefs, Semrush or similar | Whether removing it would throw away links |
| Internal links in | Crawl | Whether the site supports the page |
| Indexability and status code | Crawl, URL Inspection | Whether Google can index it at all |
| Word count and last updated | Crawl, CMS | Whether it's thin or stale |
| AI citations | Your own prompt tracking | Whether ChatGPT, Gemini, Perplexity or AI Overviews use it as a source |
Check the data before you trust a single row. We've seen two failures on one enterprise B2B account that would have wrecked an audit.
The first is bots. A client's GA4 showed AI-assistant visits jump from 354 to 6,050 in a month. Almost none of it was real. About 5,700 came in three one-hour bursts: same 1920x1080 desktop Chrome, all new users, no referrer, about 20 seconds each, landing on two pages. Real AI traffic was 385 sessions against 354 the month before, roughly flat.
An audit that trusted the "LLM referral" column would have crowned those two pages as AI winners and left the rest of the site unexamined.
The second is conversions that never count. A client's four enquiry types never counted as conversions. Their event names ran 41 to 50 characters; GA4 cuts at 40. The 40-character "career" event counted every time. Nobody had noticed, because the events still showed up.
Feed that into a content audit and every page driving general, partner or investor enquiries looks like it converts nothing. Those are exactly the pages you'd prune.
Run three checks before scoring:
- Look at the hourly traffic behind each channel, because bots arrive in bursts.
- Fire every conversion event yourself.
- Compare Search Console clicks with GA4 organic landings, page by page, and chase any page where they disagree by a wide margin.
Step 4: Group pages into families before you judge them
Tag each URL with its page family and topic cluster, then read the numbers by family first. A page family is every URL built from one template for one purpose. A topic cluster is the set of pages that answer one group of related searches, whatever template they use.
On a site with 30,000 URLs, a handful of families hold most of the pages: blog posts, docs, integrations, glossary terms, comparisons and customer stories. Family first finds what URL-first hides.
Say 1,400 integration pages share one thin template, with the same three paragraphs and a partner logo swapped in on each. The fix is the template. If the glossary gets impressions but no clicks, look at its titles before you touch a word of body copy. And when two families chase the same searches, you've found cannibalisation at scale.
Step 5: Decide keep, refresh, merge or prune for each URL
Give every URL exactly one action. These are the rules I use, in this order, and they're starting points: a page with a clear business job overrides them.
| Action | When to use it | What you do |
|---|---|---|
| Keep | It ranks, brings clicks or conversions, and is accurate | Leave it. Add internal links if it deserves more |
| Refresh | It sits in positions 4 to 20, or has impressions but a low click rate, or the facts are out of date | Rewrite the title and opening answer, update facts, add what the top results cover, fix internal links |
| Merge | Two or more URLs win impressions for the same searches and split the clicks | Pick the strongest URL, fold the best of the others into it, 301 the rest to it |
| Prune: redirect | No clicks or conversions in 12 months, but it has links or a close replacement exists | 301 to the closest relevant page |
| Prune: remove | No clicks, conversions or links in 12 months, no business job, and no close replacement | Return 404 or 410, and drop it from sitemaps and internal links |
| Keep, noindex | It serves users but shouldn't rank, such as gated thank-you pages or internal search results | Leave it live with noindex |
Two checks before any prune. First, has the page had time to work? I give new pages six months, and longer if they target a search with real competition. Second, does anyone outside SEO depend on it, such as sales, support, legal or investor relations, for links they send every week? Ask before you remove anything.
Refreshes are where the money is. A page with impressions and no clicks usually has a snippet problem, not a ranking problem. Its title and description miss what the searcher wanted, and rewriting those two lines is the cheapest fix on the whole list.
Step 6: Ship the redirects and removals
Turn every merge and prune into a redirect map. One row per old URL, with its new URL, status code, reason and owner. Engineering ships it and someone tests every row after release, because a redirect map with typos does more damage than no map at all.
The rules, most of them straight from Google's documentation:
- Use permanent server-side redirects. Google's redirect guide recommends them, and 301 and 308 both mean a page "has permanently moved".
- Redirect to the closest relevant page. Sending every old URL to the homepage tells the reader nothing and wastes the links that pointed at the old page.
- Point straight at the final URL. Google's crawlers follow up to 10 redirect hops (Google), and its crawl budget guide says to avoid long chains.
- Return 404 or 410 for pages you remove for good. A 404 "is a strong signal not to crawl that URL again", says the crawl budget guide.
- Keep redirects for at least a year. Google's site move guide says "for as long as possible, generally at least 1 year".
- Update internal links and sitemaps. Nothing on your own site should point at a redirect or a dead page.
Why redirect instead of adding a canonical tag? Google's duplicate URL guide lists redirects first among canonical signals, as "a strong signal that the target of the redirect should become canonical". A merged page should stop existing, not linger as a duplicate.
Step 7: Set governance so the site doesn't drift back
Governance means the rules that stop the next audit finding the same mess. Every URL gets an owner and a review date in the inventory, and the inventory stays live.
| Rule | What it looks like |
|---|---|
| An owner per URL | A named team or person, recorded in the CMS |
| A review date per URL | Fast-changing pages every six months; evergreen pages yearly |
| A publishing check | Before a new page goes live, search your own site for one that already targets the same search |
| A retirement rule | When a product, plan or integration ends, its pages get a redirect decision that week |
| A monthly data check | Clicks, conversions and the bot filter, compared with the same 28 days last month |
The publishing check matters most. Two minutes before each launch saves a merge project a year later.
What goes in a content audit template?
A content audit template is a spreadsheet with one row per URL and a column for every fact you need to decide that page's action. Here are the columns we use, in order. Copy them into a sheet, and you have a working template for a site of any size.
| Column | Example value |
|---|---|
| URL (normalised) | /blog/sso-for-saas |
| Page family | Blog post |
| Topic cluster | Single sign-on |
| Owner | Product marketing |
| Published / last updated | 2023-04-11 / 2024-02-02 |
| Status code / indexable | 200 / yes |
| Clicks, 12 months | 2,140 |
| Impressions, 12 months | 96,000 |
| Average position | 8.4 |
| Main query | sso for saas |
| Other URLs on the same query | /docs/sso-setup |
| Organic conversions, 12 months | 11 demo requests |
| Referring domains | 14 |
| Internal links in | 6 |
| AI citations seen | Perplexity, 2 of 10 runs |
| Action | Merge into /solutions/sso |
| Redirect target | /solutions/sso |
| Reason | Same search as the solutions page; that page converts |
| Review date | 2027-04 |
I invented the example values to show the format. Most templates leave out one column, and it's the one I'd fight for: "Other URLs on the same query". Pull query and page together from the Search Console API, then flag every query where more than one of your URLs earns impressions. That column finds your merges.
When should you merge or delete pages?
Merge when two or more of your URLs show up for the same searches and split the clicks between them. Delete when a page has no clicks, conversions, links or business job after a fair trial. And no close page to send it to.
Cannibalisation is the usual reason to merge, and it grows with templates. It happens anywhere a template multiplies pages, and most enterprise sites have several cases in plain sight. Integration pages fight their own docs. Industry pages fight solution pages, and glossary entries fight blog posts on the same term. When Google can't tell which of your pages answers a search, none of them tends to rank well.
My view is simple. A small set of excellent pages beats a large set of average ones. One strong page on a topic will usually outrank three weak pages that split its clicks, links and internal support between them.
Does deleting old content help SEO?
Deleting old content helps only when the pages are wrong, duplicated or useless. Deleting to look fresh does nothing. Google's guide to helpful content asks whether you're "removing a lot of older content" because you think it'll make your site "seem 'fresh?'" Its answer: "No, it won't."
The same guide warns against changing a page's date when the content hasn't changed much. A refresh needs real change: new facts, a better answer, a clearer structure. Our guide to content refresh covers how to run one.
What pruning does well:
- It ends cannibalisation, so one page collects the clicks and links that three used to share.
- Redirects move link value from dead pages to live ones.
- Sales and support stop sending buyers to retired products.
- Your team has fewer pages to keep accurate, which matters more every year a site grows.
Google's spam policies call it scaled content abuse when "many pages are generated for the primary purpose of manipulating search rankings and not helping users". If an old publishing push looks like that, the audit is where you clean it up.
How does a content audit change for AI search?
A content audit for AI search adds one question per page: do ChatGPT, Gemini, Perplexity and Google's AI Overviews use it as a source? Track a fixed set of buyer questions in each engine. Record which of your URLs they cite, and add that as a column in the audit.
That column changes some decisions. A page with modest Google clicks that Perplexity cites for a category question may be worth keeping and improving. And a page AI answers never cite, for a question they answer every time, needs a clearer answer in its first lines.
Two cautions from our own work. AI referral traffic in GA4 can be bots, as the 6,050-session month above shows. Filter it before it drives any decision. And Google says AI Overviews and AI Mode carry "no additional technical requirements" (Google). The same audit serves both. One accurate, answer-first page per topic, crawlable and well linked, does the job for Google and the AI engines alike.
A worked example: a B2B SaaS site with 40,000 URLs
Take a venture-backed B2B SaaS company with 40,000 indexable URLs. Most are help-centre articles, about 34,000 of them, with 3,000 blog posts, 1,400 integration pages and 900 glossary terms on top. A few hundred product, solution and comparison pages carry the sales load. These are example figures, there to show the method.
Organic sign-ups have been flat for a year, and the board wants to know whether the content team is helping or hurting. Here's how the audit plays out by family:
| Page family | What the data shows | Decision |
|---|---|---|
| Product, solution, comparison pages | Most of the demo requests; several sit at positions 5 to 12 | Refresh, starting with the pages closest to page-one top three |
| Blog posts | A small share of posts bring most clicks; hundreds have none in 12 months; many overlap | Keep the winners, merge the overlaps, prune the dead with redirects |
| Integration pages | One thin template; impressions, few clicks | Fix the template once: add setup steps, use cases and the partner's own terms |
| Glossary terms | Impressions with a low click rate; some duplicate blog posts | Rewrite titles and opening definitions; merge duplicates into the stronger URL |
| Help centre | Serves customers; few conversions, as expected | Keep; exclude from pruning; fix only broken pages and redirect chains |
Notice what doesn't happen here, even though both moves would look decisive in a slide deck. Nobody rewrites 34,000 help articles. And nobody deletes the help centre for failing to convert, because converting was never its job. The effort goes where the pipeline is. One template release does what 1,400 separate page edits would have done, in a fraction of the engineering time.
How long does a content audit take?
The first full pass on a large site takes weeks, and most of that time goes on decisions rather than data. With the API, the data is fast. Sign-offs are slow, because product, legal and engineering each need to agree before pages move or disappear.
| Stage | Effort | What slows it down |
|---|---|---|
| Inventory and data | Low, with the API | Broken tracking, URL variants, missing CMS fields |
| Decisions | Highest | Pages nobody owns; debates over pages with traffic but no conversions |
| Redirects and removals | Medium | Engineering queues; redirect rules split between CDN and CMS |
| Refresh work | Ongoing | Writing and subject-matter review |
Slow sign-off has one fix. Agree the rule for a page family once, then apply it to every URL in that family without reopening the debate page by page.
How does Searchmaxxed run a content audit?
Searchmaxxed runs the content audit inside our SEO audit service, the AI SEO Audit. It takes five business days. It covers technical SEO, on-page SEO, GEO and AEO, and it ranks every fix by what it's worth to you.
The content audit lives in the part we call "what the site is doing to itself". We crawl the site and read it page by page. We find crawl blocks, index faults, pages competing for one search and thin buying pages, then tie each fault to the searches it touches. A fault on a page with no demand drops down the list, while a fault on a money page moves up.
You get one document in four parts:
- A map of where buyers search without finding you, with the competitor that wins each search.
- A read on what the site is doing to itself, family by family.
- Every fix, ranked by what it's worth shipping first.
- The plan in ship order, naming the page, the change and who does it.
We walk you through it live. Your team can run the plan, or we can run it for you. We credit the audit to the build or first month when you start within 14 days. Five days finds and ranks the problems. Signing off and shipping every URL decision takes longer, and that pace depends on your approvals as much as ours.
Here's how we do it. AI agents connect to your Search Console and GA4, read the Google results and scrape the competitor pages that win. They also check what AI answers say about you. That feeds the Managed Search Loop, which ranks the work by how close it sits to revenue. We build the changes on a website you own, our Agentic Website, then check the live pages and measure what moved. Off your site, the Off-Page Source Layer builds the links, digital PR and mentions that back up the pages you kept.
After the audit, the managed packages keep the content in shape. Every one includes ongoing audits, a 90-day plan in weekly releases, the build work itself and a monthly decision memo.
| Package | Who it suits | What it includes for content |
|---|---|---|
| Growth System | A company growing in one or two markets | 30 page releases a month (new pages or substantial rebuilds), work on up to 4 existing landing pages, AI visibility tracking |
| Agentic Search Engineering | A competitive or national market | 50 page releases a month, new comparison, industry and integration page families, digital PR and link campaigns, AI answer tracking across ChatGPT, Gemini, Perplexity and AI Overviews |
| Custom & Enterprise | Several brands, several markets or a large site | Dedicated engineering, custom reporting and original first-party research |
You own the site, repository, content and data from day one. Searchmaxxed's work has generated more than A$10 million in revenue through search for its clients.
For the pages you decide to refresh or merge, our SEO content service sets out how we write and build them.
Is Searchmaxxed right for you?
Searchmaxxed suits marketing and growth leaders at enterprise, B2B SaaS, venture-backed, manufacturing and public-sector companies who want an audit turned into shipped changes. Want a spreadsheet and nothing more? A good crawler and this guide may be enough.
Growth System fits a company growing in one or two markets. Agentic Search Engineering fits a crowded or national category, where new page families, digital PR and links matter. Custom & Enterprise covers several brands, several markets or a site with tens of thousands of URLs.
| Searchmaxxed is a good fit if | Look elsewhere if |
|---|---|
| You want every fix ranked by what it's worth, then built | You only want a list of issues to work through in-house |
| Your site has thousands of pages and several page families | Your site has a few dozen pages; a spreadsheet afternoon will do |
| You want SEO and AI search handled together | You only need a one-off redirect map from a migration agency |
| You can approve page changes within days | Every page change takes weeks of committee approval |
| You want to own your site, repository, content and data | You need search to fill this month's pipeline on its own |
The approval row matters most on large sites. An audit is only worth the changes that ship. A five-week sign-off for each page will stall any plan, ours included.
Frequently asked questions
How often should you do a content audit?
Once a year for a full audit on a large site, with page families reviewed on a rolling quarterly cycle. Audit straight away before a migration, after an acquisition or rebrand, or after two or three months of traffic loss that nobody can explain.
What is the difference between a content inventory and a content audit?
A content inventory lists every page you have, with facts such as URL, title, owner and dates. A content audit adds performance and quality data. Then it gives each page an action: keep, refresh, merge or prune.
What tools do you need for a content audit?
A crawler, Google Search Console, GA4, a backlink tool and a spreadsheet or database. On a large site, pull Search Console through its API, because the interface exports only 1,000 rows. Add AI answer tracking if AI search matters to your pipeline.
Should I delete pages with no traffic?
Sometimes. Delete only after a fair trial, and only if the page also has no conversions, links, business job or close replacement. If it has links or a related page exists, redirect it instead. Google says removing content to look fresh won't help rankings, so pruning should never be a freshness trick.
How do I find cannibalisation in a content audit?
Pull Search Console data by query and page together through the API. Any query where two or more of your URLs earn impressions is a candidate for a merge, and the ones splitting real clicks come first. Keep both pages when they serve different intents.
Can AI do a content audit?
AI can do most of the gathering and first-pass scoring: pulling data, tagging page families, comparing pages with the top results and drafting refreshes. People still make the keep, merge and prune calls. Those depend on product, legal and sales context that never shows up in the data.
How long should redirects stay in place after a content audit?
At least a year. Google's site move guide says "generally at least 1 year". That gives Google time to transfer signals and reassign links pointing at your old URLs.
What should you do next?
Export your page-level Search Console data through the API, join it to a crawl and tag every URL by page family. Within a day, you'll see where the problems sit. If you'd like us to run the audit and ship the fixes on your site, let's talk pricing.