Content audit: how to decide what to keep, update, merge or remove

A content audit is a page-by-page review of everything your website publishes, measured against what each page is meant to do. For every URL you record what it is, how it performs and whether it is still accurate, then decide one of four things: keep it, update it, merge it with another page or remove it.

It is the maintenance side of a content strategy, the part that keeps an archive useful as it grows. This guide covers when to run an audit, the six steps from inventory to action, how to make the four decisions without guesswork, what AI search changes and the columns to put in your audit spreadsheet.

What is a content audit?

An audit starts with an inventory. The Nielsen Norman Group draws the line clearly: a content inventory lists what you have, with each page’s URL, format, author and dates, while the audit judges its quality and decides whether it needs updating or removing. The inventory tells you there are 240 blog posts. The audit tells you that 60 of them still bring in enquiries, 90 need work and the rest are doing nothing.

Two other audits overlap with it. An SEO audit checks whether search engines can crawl, index and understand the site, so it looks at templates, redirects and speed. A digital marketing audit reviews every channel at once. A content audit sits between them: it looks at what each page says and whether that is still worth saying.

When to run a content audit

Most sites need a full audit once a year and a lighter check on their most important pages every quarter. Some moments call for one straight away:

• Before a redesign or migration. Moving pages you no longer need costs time and creates redirects you will have to maintain.
• After a drop in organic traffic. Google’s own guidance on helpful, people-first content suggests auditing the drops: which pages lost traffic and for which kinds of searches.
• Before a new content plan. You can’t plan new topics well without knowing which ones you already cover and how well.
• When a business changes. New prices, a dropped service or a rebrand leave old pages saying the wrong thing.

Large sites should audit in chunks. When the Government Digital Service audited 7,396 transport content items on GOV.UK with the Department for Transport, its main lesson was to split the work into smaller pieces, with a separate spreadsheet for each agency. A small business with 50 pages can do it in an afternoon. A site with thousands should go one section at a time.

How to do a content audit in 6 steps

1. Set the goal and the scope

Write down what the audit is for, because the goal decides which numbers matter. An audit to win more enquiries weighs key events and service pages. An audit before a migration weighs which URLs carry traffic and links. Then choose the scope: the whole site, the blog only or one topic. Agree who owns each section, since someone has to act on the decisions.

2. Build the inventory

List every URL in scope. Your CMS export and XML sitemap are a start, but a crawler finds pages they miss, such as old landing pages that are no longer linked from anywhere. The Screaming Frog SEO Spider crawls up to 500 URLs at no cost, which covers most small business sites. Record the URL, title, content type, author, publication date and last update for each page.

3. Add performance data

Pull twelve months of data so seasonal pages are judged fairly. Google Search Console keeps 16 months of clicks, impressions and average position for each page, and shows which queries each one appears for. In GA4, add engaged sessions and key events per page. Google counts a session as engaged when it lasts longer than 10 seconds, includes a key event or has at least two page views, which is a more honest measure than raw visits. If you have a backlink tool, add the number of linking domains, because a page with strong links is rarely one to delete.

4. Review the quality

Numbers show which pages perform, not why. Read each important page and score it against a short set of questions. Google’s self-assessment list is a good base: does the page give original information or analysis, is it clearly sourced, does it show who wrote it and does it contain easily verified factual errors? The E-E-A-T signals of authorship, sources and company details belong in the same check. Note anything out of date: prices, screenshots, laws, statistics and the year in the title.

5. Decide what happens to each page

Give every URL one of the four decisions, using the criteria in the next section. Group pages by topic before you decide, so you can see where two or three posts answer the same question.

6. Act, then set review dates

Turn the decisions into tasks in your content calendar, starting with pages that already have traffic, since improving them pays back fastest. Give each page an owner and a next review date. Fast-changing pages, such as pricing or legal guides, need a review every few months. Evergreen guides can wait a year.

Keep, update, merge or remove

Each decision follows from two questions: does the page do a job for the reader, and does it do that job better than any other page on the site?

Update first. Pages with impressions but few clicks, or rankings just off the first page, usually need a better answer rather than a new page. When HubSpot reworked older blog posts, it more than doubled the monthly leads those posts generated and raised their monthly organic search views by an average of 106%. A real update can carry a new date. Google’s advice on dates is not to artificially freshen a story without adding significant information, so changing the date alone does nothing useful.

Merge when pages compete. Three short posts on the same question split the clicks and the links between them, a problem known as keyword cannibalisation. Combine them into the strongest URL, then point the others to it with a permanent redirect, which Google treats as a strong signal that the new page replaces the old one.

Remove with care. Old does not mean bad. In 2023, when CNET deleted thousands of older articles to appear fresher to Google, Google’s Danny Sullivan warned that removing content just because it is old is “not a thing” that helps a site rank. Remove a page because it no longer serves anyone. If no close equivalent exists, a 404 is the honest answer: Google treats 404 and 410 the same way and drops the URL from its index. Redirecting every removed page to the homepage only confuses visitors.

Content audits and AI search

AI assistants add another reason to keep pages current. An Ahrefs study of 17 million citations found that ChatGPT, Perplexity, Copilot and Gemini cite content that is 25.7% fresher on average than the pages in Google’s organic results: 1,064 days old against 1,432. Google’s own AI Overviews behaved more like ordinary search in the same data, so the effect varies by platform.

Add three questions to the quality review for your most important pages. Does the page answer its question in the first paragraph? Is it structured with clear headings, lists and tables that an AI tool can quote? Does it show a named author and a visible update date? The AI Readiness Checker tests these points for a single URL, along with whether AI crawlers can reach the page at all. Pages that fail are strong candidates for the update list. For pages that lack structured data, the Schema Markup Generator creates Article and FAQ markup in JSON-LD. The wider approach is covered in the guide to generative engine optimisation.

Content audit template

A spreadsheet is enough. Use one row per URL and these columns, filled in roughly from left to right as you work through the steps:

Add a “merge into” column if you find many overlapping pages, and keep the finished sheet. Next year’s audit is much faster when you can compare against this one.

Common content audit mistakes

• Judging pages on traffic alone. A pricing page or a guide for existing customers may get few visits and still matter. Check what a page is for before calling it weak.
• Deleting pages that carry links. A post with no visits but a dozen linking domains is worth merging or redirecting, not dropping.
• Using too short a date range. Three months of data makes a Christmas gift guide look like a failure in June.
• Forgetting internal links. After a merge or removal, update the links that pointed to the old URL, and check for orphan pages that nothing links to.
• Auditing without acting. A finished spreadsheet changes nothing. Put the first ten actions in the calendar before you close it.