Reading time: 11 minutes
A content audit is a survey of every page on a blog, with the data for each one. It helps you decide the fate of each post: keep, update, merge, redirect or delete. Today, it also covers citations in AI answers. This guide shows the method for anyone who took over a large blog.
In short
- A content audit starts with a URL inventory and combines data from Search Console, GA4 and AI citations.
- Each post gets a fate: keep, update, merge with another, redirect or delete.
- For Google, deleting content is a last resort, and removing posts just to make the site seem fresh doesn’t work.
- When you merge two posts, use a permanent redirect to the page that stays.
What is a content audit and when should you do one?
A content audit is a data-driven, post-by-post review of a blog that ends with a decision for each URL. It differs from a technical audit because it looks at the text, the topic and the performance of each page. Our suggestion is to do the first one when you take over a blog and then repeat it once a year.
Sometimes the review can’t wait. For example, when organic traffic falls for several months, when two posts compete for the same query or when the brand changes its offer. Similarly, audit before you reorganize the blog by topic, because the new map depends on what already exists.
The difference from a technical audit is the object. Crawling, indexing and bot access belong in a technical SEO audit. A content audit, however, asks whether each post still answers a query well, has current data and deserves to stay online.
AI has also changed the list of criteria. Ahrefs, which sells SEO tools, analyzed 16.975 million URLs cited by AI and by Google Search. According to the study, content cited by ChatGPT, Perplexity, Gemini and Copilot was, on average, 25.7% fresher than content in organic results. The study shows a trend, but it doesn’t prove cause. Even so, the date of the last review deserves a column in the audit.

What data should you gather: Search Console, GA4 and AI citations?
Gather four groups of data for each URL: search (Search Console), behavior (GA4), AI citations and the links the page earns. We suggest the same time window for all of them, ideally the last 12 months, so you don’t confuse seasonality with decline. The goal is one row per URL, with everything the decision needs.
How do you build the URL inventory?
First, list every URL on the blog, not just the ones that bring traffic. The simplest sources are the sitemap and the CMS post list. Then add what they miss, such as tag pages and old posts with no links. Finally, note the title, the last review date, the topic and the main query for each URL.
Which Search Console and GA4 metrics should you use?
In the Search Console performance report, use the four metrics on the Pages tab: clicks, impressions, CTR and average position. Compare two equal periods. That’s because falling clicks with stable impressions suggest a different problem than both falling together. Next, open the Queries tab for each URL to see which queries it actually serves.
In GA4, the landing page report shows the first page each visitor landed on. You’ll find it under Reports, Engagement, Landing page. Use the sessions and the engagement rate of each post. According to Google Analytics, an engaged session lasts longer than 10 seconds, has a key event or has 2 or more page views. To separate visits from assistants such as ChatGPT, see how to measure AI traffic in GA4.
How do you include AI citations in a content audit?
Use the two official reports that show AI data by page. On Google, the Search Console generative AI performance report shows impressions in AI Overviews and AI Mode. It also has a Pages tab. The step-by-step guide covers AI Mode data in Search Console.
On Bing, the AI Performance report in Bing Webmaster Tools counts the citations for each page in Copilot, in AI summaries in Bing and in select partners. According to Bing’s help page, however, a citation doesn’t represent a click or traffic. Still, it shows which posts AI uses as a source, as the guide to the Bing AI Performance report explains. Finally, the backlinks for each URL come from an SEO tool and weigh on the decision to redirect.
How do you decide whether to keep, update, merge or delete?
The criterion we propose combines two factors: does the post still have value for someone, and does it still answer its query well? If it has value and answers well, keep it. When it has value but has aged, update it. If it repeats another post, merge and redirect. Finally, delete only what has no value and can’t be saved.
| Post status | Decision | What to do |
|---|---|---|
| Traffic, citations and current data | Keep | Just check links and date |
| Good topic, old data or decline | Update | Rewrite, replace sources |
| Same topic as another post | Merge | Move the best parts and redirect |
| Off-topic for the business, with links | Redirect | Point to the closest page |
| No value, no links, no visits | Delete | Let it return 404 or 410 |
This table is our own proposal, a starting point for the content audit, and not a Google rule. For example, a post with no traffic may be the brand’s official answer on a topic, in other words, a source that AI uses. So check the citations before you cut. According to Bing’s help page, a page may not appear in AI answers when it is outdated or outperformed by other sources. In those cases, therefore, the decision is usually to update, not delete.
Updating, by the way, means changing the content, not just the date. On its page about helpful content, Google questions changing the date of pages when the content has not substantially changed. A separate guide covers how to update outdated content step by step, from the title to the sources.
When should you merge two posts and redirect?
Merge when two posts answer the same question and neither is complete. Move the best parts of the weak post into the strong one, update the text and then redirect the old URL. According to Google’s page on redirects, a permanent redirect signals that the target should be canonical. Google recommends a server-side 301 or 308.
This way, the signals of the old page aren’t lost. According to the documentation on canonical URLs, naming the preferred page helps search engines consolidate the signals of similar pages, such as links, into a single URL. Then fix the internal links so they point straight to the page that stayed.
When should you use 404, 410 or noindex?
Use 404 or 410 when the post goes offline for good and there’s no similar page to send the reader to. For Google, both have the same effect. According to the page on HTTP status codes, all 4xx errors, except 429, are treated the same, and an already indexed URL drops out of the index.
In a content audit, noindex is for a page that stays useful to visitors but doesn’t need to be in search. According to the guide on blocking indexing, a noindex page can’t be blocked in robots.txt, or the crawler never sees the rule. In its help page on removals, however, Google says noindex is less secure than a 404 or 410 to remove a page for good. If an old post shows a price, address or service that changed, also see how to remove outdated content from Google and from AI answers.
Does deleting old posts help or hurt?
Google doesn’t promise that deleting posts improves rankings. On the contrary, it treats deletion as a last resort and says that removing lots of old content to make the site seem fresh won’t work. The exception is entire sections made for search engines, not people. In that case, cutting can help the rest of the site.
On its page about core updates, Google says deleting content is a last resort. According to the text, you should only consider it if the content can’t be salvaged. In addition, Google notes that entire sections you’re thinking of deleting were likely created for search engines first, not people. In that case, the page says that deleting the unhelpful content can help the good content perform better.
The same page asks you to avoid “quick fix” changes, like removing something because you heard it was bad for SEO. Also, the effect isn’t immediate. According to Google, it could take several months for its systems to confirm that the site as a whole has improved. Meanwhile, in its list of questions about helpful content, Google mentions removing lots of older content to make a site seem fresh. Its answer is “No, it won’t.”
In short, a content audit shouldn’t turn into a quota of deleted posts. The risk of deleting without criteria is losing pages with links, visits or AI citations. That’s why you should delete only what went through the data and can’t be fixed.
How do you find cannibalization in a content audit?
In an audit, cannibalization shows up when two URLs in the inventory share the same main query. To confirm it, filter the query in the Search Console performance report and open the Pages tab. If two URLs take turns in the impressions for the same query, the decision is usually to merge them.
In the spreadsheet, sort the URLs by the main query column. This way, pairs on the same topic sit side by side and the overlap stands out. Next, choose the owner page by adding up the data: more clicks, more backlinks and more AI citations. The other page goes into the merge-and-redirect queue.
However, not every similarity is competition. “What is”, “how to” and “tools” posts on the same topic can stay separate, as long as each one answers its own intent. To keep the overlap from coming back, organize the blog into topic clusters, with one owner page per query.
What spreadsheet or template should you use for a content audit?
A simple spreadsheet works: one row per URL and one column for each data point and for the decision. SEO tools help you collect the data, but the decision is still manual. The template below is our suggestion. It fits in a regular spreadsheet and works for blogs with 50 or 500 posts.
| Column | What to record | Where it comes from |
|---|---|---|
| URL and title | Address and current title | Sitemap or CMS |
| Main query | The query the post should answer | Search Console |
| Clicks and impressions | Two equal periods | Search Console |
| Sessions and engagement | Visits that land on the post | GA4 |
| AI impressions and citations | By page | Search Console and Bing |
| Backlinks | Domains that link to the post | SEO tool |
| Last review | Date of the last real change | CMS |
| Decision and owner | Keep, update, merge, redirect or delete | Content team |
How do you run a content audit on a blog with 500 posts?
To run a content audit on a large blog, we suggest splitting the work into batches. Start with the full inventory and then fill in the data in one go, with exports. After that, decide the extremes first: the posts that bring the most visits and the ones with no data at all.
- Inventory: every URL, with topic and main query.
- Data: export Search Console, GA4 and the AI reports into the spreadsheet.
- Pairs: sort by main query and flag repeated topics.
- Decision: fill in the fate column, starting with the posts with the most traffic.
- Execution: first the redirects and merges, then the updates, and finally the deletions.
- Review: note the date of each change and compare the data in the following months.
When you run a content audit this way, the team knows why each URL changed. As a result, the next round starts from the spreadsheet, not from scratch. The bigger plan, which ties the audit to the blog’s goals, is in the guide to content marketing in the age of AI.
Main sources
- Google Search Central, Google Search’s core updates and your website (2025).
- Google Search Central, creating helpful, reliable, people-first content (2026).
- Google Search Central, redirects and Google Search (2026).
- Google crawling documentation, how HTTP status codes affect Google’s crawlers (2026).
- Google Search Central, how to specify a canonical URL (2026).
- Search Console Help, generative AI performance report (2026).
- Bing Webmaster Tools Help, AI Performance (2026).
- Ahrefs, study on the age of content cited by AI (2025).



