How to do a website content audit

At the most basic level, a content audit is a review of the information on your website. Working page by page, you assess how your content is performing against qualitative criteria and quantitative metrics. The objective is to understand what content you have, what state it’s in, and what you can do to improve it.

It sounds straightforward. And in principle, it is. But spend five minutes planning one and the complexity starts to reveal itself:

• How do you find every single page on your site? 
• Do you need to audit all of them, and if not, how do you decide which ones to include? 
• How do you judge pages consistently and objectively? 
• And which criteria should you use to make sure you’re looking at content from every angle?

AI offers a very appealing response to this complexity. There’s loads of tools out there now that promise to crawl your site, flag errors, and spit out a report in minutes. So why would you bother doing a content audit the old way at all? 

The honest answer is that those tools can be useful – but they have limits. They can tell you a page has too few words, but not whether those words are accurate. They can flag a low traffic page, but not that this page is there for a small, high-value audience so low traffic is absolutely fine. You need human intelligence for those kinds of insights.

If anything, all the change in the last few years makes a content audit more valuable, not less. You need to know what you’ve got, whether it’s still working, and where to focus your energy. Plus if you want to optimise your website to be found and referenced by AI search and large language models (what’s being called generative experience optimisation or GEO) a content audit is a great first step. That’s what this guide is for. It will walk you through the process, from planning, to execution, to turning your findings into action.

Why audit your website content

Most websites grow in the same, organic, unplanned, slightly chaotic way. Someone adds a page for a campaign. A team member leaves and their content stays. A strategy changes but the old pages don’t. Marketing absolutely must have a microsite. Years pass. Before long you’re sitting on thousands – or hundreds of thousands – of pages, and nobody is entirely sure what’s working, what’s embarrassingly out of date, or what’s quietly doing damage.

A content audit changes that. It gives you an honest, evidence-based picture of what you’ve actually got – not what you think you’ve got – and a clear sense of what to do about it. In my experience, most organisations are surprised (both happily and unhappily) by what they find. 

The benefits tend to fall into four areas:

• Strategic: An audit helps you prepare for major changes like redesigns or migrations, align content with organisational goals, and identify gaps and opportunities. If you’re about to rebuild your site, an audit stops you carrying dead weight into the new version.
• Operational: An audit helps you get control of sprawling, unmanaged content, create a baseline for ongoing maintenance, and get new team members up to speed quickly. If you’ve just inherited a website, it’s the fastest way to get the lay of the land.
• Risk management: Audits surface outdated or inaccurate information, accessibility failures, and compliance gaps. That page from 2018 with incorrect legal information? An audit will find it before your users do.
• Environmental: Content has a carbon cost. Every page needs to be hosted, and hosting requires energy. An audit identifies content you can delete – which is better for your users and for the planet.

Choosing your website audit approach

Content audits aren’t one-size-fits-all. You need to choose an approach that matches your goals, resources, and timeline. In my experience, there are 4 main ways you can approach an audit:

1. Full audit

A comprehensive review of all (or nearly all) pages on your site. This works well for smaller sites (100-500 pages), or when you need a complete picture before a redesign. 

2. Focused audit

A narrow audit of pages related to a specific user journey, product, service, or campaign. Use this when you’re launching something new, optimising a particular section, or working with limited time. 

3. Rolling audit

An ongoing process where you audit a few pages every week or month. This suits organisations that want to maintain content quality continuously rather than in big, infrequent projects. 

4. Sample audit

A representative sample of content from across the site. This works for large sites (500+ pages) where a full audit isn’t feasible, or when you need directional insights quickly. A well-designed sample of 100-200 pages can reveal patterns across thousands of pages.

You can combine these approaches. For example, you might use a tool to do a quantitative-only full audit of all your content, a qualitative sample audit of your highest-traffic pages, and then move to a rolling audit for ongoing maintenance.

Choosing a sample of content strategically

Choosing a sample of content is tough, and it’s one of the things I get asked about most. It’s part science, and part art. You can use sampling methods and make some rational choices in advance about what to focus on and what you want to cover. But you’ll also want to leave yourself enough flexibility to dig into anything interesting or surprising that comes up.

Some different sampling methods you can use are:

• Traffic-based sampling: Audit your highest-traffic pages first. A small percentage of pages typically drive the majority of traffic. Start with pages that get 80% of your visits.
• Representative sampling: Take samples from each major section or content type to understand patterns across the site. If you have 1,000 blog posts, audit 30-50 representative examples rather than all of them.
• Stratified sampling: Divide content into categories (by topic, audience, content type, or business unit) and sample proportionally from each category.
• Risk-based sampling: Prioritise pages with legal, compliance, or reputational implications. Medical information, financial advice, legal disclaimers, and policy pages warrant careful review.
• User journey sampling: Audit all pages along critical user journeys, from landing page through to conversion.

Most effective audits combine several methods. You might audit all top-traffic pages, a representative sample from each section, and all pages along your top 3 user journeys.

How to judge content quality: the 10 heuristics

If you’re an experienced content professional, you could probably look at a page and immediately identify what’s not working. But instinct alone won’t give you an audit you can stand behind. Without a defined set of criteria, your judgments will shift from page to page, your scores won’t be comparable, and it becomes very hard to explain your decisions to stakeholders. Criteria give you a system. They give you objectivity. And they make the difference between an audit that drives change and one that just reflects your opinions on a given day.

I use 10 heuristics – shortcuts, rules of thumb, guidelines – to help me evaluate content quality:

1. Strategic: Does it align with your organisation’s goals and strategy?
2. Relevant: Does it meet user needs? 
3. Informative: Does it contain enough useful information? 
4. Clear: Is it easy to understand? 
5. Structured: Is information organised logically with headings, bullets, and white space? 
6. Engaging: Does it draw people in and hold their attention? 
7. Consistent: Does it follow your brand voice, house style, and design patterns?
8. Accessible and inclusive: Can everyone access and understand it, regardless of ability or background? 
9. Findable: Can people discover it through search and navigation?
10. Actionable: Does it help people take the next step? 

You can read more about the thinking behind these heuristics here.

It’s a good idea to build out guidance on each heuristic you want to use, to give yourself key signals and red flags to look out for as you audit.

Scoring and judging content in an audit

Once you’ve chosen your heuristics, you need a consistent way to apply them. Rather than a simple yes/no, I recommend scoring each heuristic on a 0-3 scale – it gives you enough nuance to distinguish between content that’s genuinely failing and content that just needs a light touch.

• 3 (Excellent): Meets or exceeds expectations, no issues
• 2 (Good): Mostly good with minor issues, low-priority improvements needed
• 1 (Poor): Significant issues that must be addressed
• 0 (Fails): Doesn’t meet the criteria at all, critical problems

Those scores then feed into the most important decision of your audit: what to do with each piece of content. Everything gets categorised as one of the following:

• Keep as is: Content that scores mostly 2s and 3s across all heuristics. It’s working well and doesn’t need immediate attention.
• Improve: Content that serves a user or business need but has quality issues. Typically this means it scores well for strategic and relevant criteria but poorly on others.
• Merge: Content that isn’t substantial enough to justify a standalone page, or that duplicates information found elsewhere, but contains something worth keeping.
• Delete: Content that doesn’t serve user or organisational needs, is seriously out of date, or gets almost no traffic. This is likely to be a lot of your content, in my experience.
• Archive: Content that’s business/org-critical or legally has to be retained but has no public-facing role. Annual reports from 2015, old board meeting minutes, historical records, that sort of thing.

How to create a content inventory

Before you can audit content, you need to know what content exists. A content inventory is a complete list of your website’s pages, along with metadata like URLs, titles, publication dates, and performance metrics.

The fastest way to create an inventory is to use a crawler like Screaming Frog, Sitebulb, or one of the newer generation of AI tools. These tools spider your site and extract data from every page. Some tools can pull in data from other platforms like GA4 and Search Console, calculate readability scores, spot missing alt text, and lots of other very helpful things.

You can also export a list of pages from your CMS, or from Google Analytics or Search Console, if you don’t have the budget for a crawler.

The goal is a master spreadsheet where each row represents one page, and the columns contain all the data you need to make informed decisions. This inventory becomes the foundation for your audit.

Doing the actual audit

Here’s where most guides make it sound easy. Open each page, score it against your heuristics, move on. And yes, that’s essentially what you’re doing – but the reality is that auditing is slow, repetitive, and surprisingly tiring. If you go in expecting that, you’ll be fine. If you don’t, you’ll burn out halfway through and end up with a half-finished spreadsheet that haunts you for months.

Before you start, set up your audit spreadsheet. It should include: 

• a column for each heuristic you’re measuring
• an overall score that calculates automatically (by adding up heuristic scores)
• a column for your keep/improve/merge/delete/archive decision
• a notes column for observations, issues, and ideas

Then work through row by row, page by page: 

• Open the page
• Assess it against each heuristic
• Record your scores
• Note anything worth flagging

I find that it takes me an average of 10 minutes to audit a page. Some pages will only take a minute or two, others will take 30 minutes or more. That means that looking at 100 pages in detail might take you the best part of two full working days. 

Here are a few tips that help me maintain quality and sanity while doing this detailed, repetitive work:

• Follow a consistent order: Work through the heuristics in the same order every time. It becomes automatic.
• Work in small batches of 5-10 pages, then take a break. Audit fatigue is real, and your scores will become less reliable if you push through for hours.
• Audit similar content types together. If you’re auditing blog posts, do them in one session. Product pages in another. This helps you maintain consistent standards.
• Keep your heuristic definitions visible. It’s easy to forget exactly what you’re measuring, especially for heuristics like ‘engaging’ or ‘strategic’.
• Note patterns as you go. If you’re seeing the same issue on every page in a section, you probably don’t need to audit every page in that section in detail.
• Don’t overthink scores. If you can’t decide between a 1 and a 2, pick one and move on. Perfection isn’t the goal – useful insights are.
• Screenshot as you audit: When you spot a good or bad example, screenshot it immediately label with the page URL or title and save it to a folder. You’ll thank yourself later when writing the report.
• Start your findings document now. Keep a running document of patterns, trends, and quotes. For example: ‘Product pages: all missing CTAs’ or ‘Blog: consistently scoring 0-1 for Structure’. This makes report-writing infinitely easier.
• Copy scores for identical pages. If you find pages that are near-identical (same template, same issues, same metrics), score one thoroughly, then copy the scores down for the others. Note what you did: ‘Rows 45-67: identical blog template, scores copied from row 45’.

Using AI to support your content audit

I’m including this section with a big caveat: AI tools have real limitations when it comes to content audits (and everything else, tbh). I’ve tried out lots of different ideas and prompts, with varying degrees of success. The issues I’ve run into are:

• Page volume. Most AI tools can only process a limited amount of data in one go. For large sites, you’ll need to work in batches, which takes more time than you might expect. Results can also become less consistent the more data you add to a single conversation. For most audits, this is the main issue that means doing things myself is actually faster.
• Consistency. AI can be inconsistent when scoring content, particularly for nuanced criteria. Always review a sample of AI-generated scores against your own judgement.
• Context. AI doesn’t know your organisation, your audience, or your strategy unless you tell it. The more context you provide upfront, the better the results.
• Accuracy. AI regularly generates plausible-sounding but inaccurate information. Always check outputs before including them in a report.

That said, here are some things that have worked reasonably well for me, and that you might want to experiment with:

1. Inventory creation and data extraction

AI tools might help you get more out of your crawler exports, CMS data, and other sources. Rather than cleaning data manually, you can ask an AI to:

• Identify and flag missing metadata – for example, pages without title tags, meta descriptions, or H1s
• Flag duplicate title tags or meta descriptions across your inventory
• Categorise content by type or topic based on URL patterns or page titles
• Clean and reformat exported data – for example, stripping tracking parameters from URLs or standardising date formats

Example prompt:

‘Here is a spreadsheet of pages exported from our website crawler. Please identify any pages that are missing a title tag, H1, or meta description, and flag any duplicate title tags. Return the results as a table.’

Note: Tools like Sitebulb already do a lot of this, and can handle large page volumes with ease. 

2. Pattern identification and content clustering

AI might be able to help you spot patterns that you miss when reviewing pages one at a time:

• Group similar pages together to identify merge candidates
• Flag near-duplicate content – pages covering the same topic in similar ways
• Identify which content characteristics tend to correlate with higher or lower scores

Example prompt:

‘Here are 30 pages from our website, with their titles and a brief description of each. Please group any pages that cover similar topics and could potentially be merged, and explain your reasoning.’

3. Choosing a sample of content to audit

If you’re doing a sample audit, deciding which pages to include is one of the most time-consuming parts of the process. AI might be able to help you build a sampling plan based on your site structure, audit goals, and available time.

To get a useful recommendation, you’ll need to give the AI some context upfront:

• Your site structure – paste your URL list or spreadsheet, organised by section and content type
• Why you’re doing the audit – for example, taking stock, preparing for a redesign, or focusing on a specific area
• How much time you have – including number of people auditing and hours available
• Any priorities – for example, high-traffic pages, known problem areas, or specific user journeys

With this information, AI can identify which content types are templated (where auditing one or two examples covers all similar pages), suggest how many pages to audit from each section, flag content that can likely be excluded, and give you a realistic assessment of whether your goals are achievable in your available time.

Example prompt:

‘I’m planning a content audit and need help selecting which pages to audit. I have [X] hours available across [X] people. My audit goal is [taking stock / focused audit / redesign prep]. Here’s my site structure: [paste your spreadsheet — columns: URL, Section, Subsection, Page type]. Please recommend how many pages to audit from each section, identify any content types I can assess as templates, flag content I can safely skip, and tell me whether my goals are achievable in the time I have.’

4. Findings synthesis and report drafting

AI might be able to help you pull your findings together once the audit is complete:

• Summarise patterns in your audit data
• Draft an initial findings section based on your scores
• Generate a list of quick wins from your completed spreadsheet
• Spot patterns and trends in your notes
• Help you structure your recommendations into now, wow, and how categories

Example prompt:

‘Here is a summary of the results from my content audit. [Paste summary data.] Please draft a findings section that identifies the three or four most significant patterns in the data, and suggest a list of quick wins based on these findings.’

Analysing your findings

Finishing your page by page audit is probably going to feel pretty satisfying. But this isn’t the end – it’s the start of stage 2. That spreadsheet full of scores and notes isn’t an insight – it’s raw material. The analysis phase is where you turn that data into something you can actually act on, and it’s worth giving it proper time rather than rushing straight to recommendations.

Start with the big picture. Which heuristics scored lowest across all your content? Are the problems concentrated in one section or spread across the whole site? A simple pie chart showing the distribution of keep/improve/delete/archive decisions, or a bar chart of average scores by heuristic, can tell a compelling story very quickly. And it will be useful when you come to present your findings.

Then try cutting the data in different ways. Do blog posts perform better than service pages? Is the About section stronger than the Resources section? These comparisons reveal where you should focus improvement efforts and – just as importantly – where things are actually working well. It’s easy to forget to say that in a report, but you should take any opportunity you can to highlight good work.

One comparison is particularly worth making: the relationship between content quality and performance. Do your highest-scoring pages also get the most traffic and engagement? Sometimes yes, sometimes no – and both answers are interesting. High-quality content that nobody finds suggests a findability problem. Low-quality content that drives significant traffic means you need to improve it carefully, not delete it.

Finally, go back through your notes. Scores tell you what, but notes tell you why. Look for recurring themes, patterns you noticed across multiple pages, and anything that surprised you. Group similar observations together. That’s where your most useful insights will come from.

Turning your findings into action 

The most thorough audit in the world is pointless if nothing changes as a result. And getting recommendations implemented requires just as much thought as doing the audit itself.

Start with the recommendations themselves. Vague ones like ‘improve content quality’ won’t get you anywhere. Good recommendations are specific, explain why the change matters, and make clear who should do what. I use this formula:

In order to [outcome], we should [specific action] because [evidence/reasoning].

For example, rather than ‘improve product pages’, you’d write something like: 

'In order to improve organic search performance, we should add structured product information to all product templates, because 73% of product pages are currently missing this information and pages with complete product data rank on average 12 positions higher.'

Once you have your recommendations, prioritise them using an effort/impact matrix. The idea is simple: plot each recommendation on two axes – how much effort it will take to implement, and how much impact it’s likely to have. Where a recommendation falls on that matrix determines how urgently you should act on it. I use a version of this called the How/Wow/Now framework:

• Wow recommendations are low effort, high impact – your quick wins. Fixing broken links, deleting zero-traffic pages, adding missing calls-to-action. Do these first. 
• Now recommendations are low effort, lower impact – worth doing soon but not urgent, like standardising inconsistent formatting or improving meta descriptions. 
• How recommendations are high effort, high impact – the strategic initiatives that need proper planning, like restructuring your information architecture or rebuilding entire sections. These go on the roadmap. 
• Oww recommendations: The fourth quadrant – high effort, low impact – are things that shouldn’t really get done at all, unless there’s a compelling legal or compliance reason.

When it comes to presenting your findings, tailor what you share to who you’re sharing it with. Senior leadership need the headline story: impact, risk, and resource requirements. Content teams need the detailed recommendations they’ll actually implement. Keep your quick wins visible and start delivering them early – it builds credibility and momentum for the bigger asks.

Finally, assign ownership to every recommendation. Without it, even the best recommendations become suggestions that sit in a document and go nowhere.

Moving forward

Content audits aren’t one-and-done projects. Content degrades over time. Sites grow. Strategies evolve. User needs change.  If you don’t build in a way of keeping on top of it, you’ll find yourself back where you started in a year or two – sitting on a pile of content nobody is quite sure about.

The most sustainable approach is to make auditing part of how you work rather than a big, occasional event. Larger organisations might do comprehensive audits annually or every two years, with focused audits in between. Smaller teams often find a rolling audit – reviewing a portion of content every month – more manageable. Either way, the important thing is that it’s on the to-do list and someone owns it.

The other shift worth making is upstream: building your audit criteria into your content creation process. If you know you’ll eventually measure content against these heuristics, it changes how you create it in the first place. That’s not a small thing. Done well, it means your content starts better and stays better for longer.

In conclusion

A lot has changed since I first started auditing content, and a lot has changed since I first wrote this guide in 2022. The web is harder to navigate than it used to be – for organisations publishing content and for the people trying to find it. But that’s exactly why this work matters. An audit won’t fix the internet, but it will give you something increasingly rare: a clear, honest picture of what you’ve got, what it’s doing, and what to do next. That’s worth a lot.

If this guide has given you enough to get started, the free content audit planner will help you structure your approach before you dive in. Get the free content audit planner

And if you want a complete, step-by-step system – templates, scoring guides, worked examples, and everything else – the full toolkit has it all. Find out more about the content audit toolkit.

First published: 12th July 2022