Content Audit Guide (2026): A Working Process, Run on This Site's Own 241 URLs

Most content audit guides describe the process without ever running it. Consequently, the step that breaks real audits goes unmentioned. This refresh runs the process on Anobee's own site, and the first pass hit that step immediately. Search Console reported 241 URLs with impressions over 16 months, against 94 in the sitemap, and nine of the ten highest-performing URLs were addresses that no longer exist as pages.

The previous version of this page claimed ten "proven" steps and invented a framework called RANK. It also carried a case study about a 120-article blog that cannot be sourced, and left eight sections as empty headings with no table or content under them. All of that is gone. Instead, what is left is the process, the decisions, and this site's own numbers.

What a content audit is, and what it is for

A content audit is a structured review of every page you have published, ending in one decision per URL. The decision set is small: keep, update, merge, redirect, noindex, delete.

The reason to run one is not tidiness. Google's guidance on the 2022 helpful content update describes "a new site-wide signal". It states plainly that "any content, not just unhelpful content, on sites determined to have relatively high amounts of unhelpful content overall is less likely to perform well in Search" [1]. The same post says that "removing unhelpful content could help the rankings of your other content". However, it warns that the classification is reapplied "over a period of months" rather than lifting the moment you press delete [1].

There is a limit to that logic, and Google states it in the helpful-content guidance. Adding or removing a lot of content mainly because you think it will make the site look fresh does not work [2]. Therefore an audit that turns into a deletion spree is optimising for the appearance of a clean site rather than for readers.

Step 0: normalise the URLs, or the audit will lie to you

This is the step that separates a content audit from a spreadsheet. It is missing from most guides because most guides never pull real data.

Search Console's Performance report groups by URL string, not by page. After any migration, slug change or trailing-slash policy change, the same page can appear under several strings. Meanwhile, old addresses keep accumulating impressions for months.

Anobee's own 16-month Pages report makes the point better than an explanation can. The ten URLs with the most clicks were:

Nine of ten are not pages. They are strings Google still has on file from before this site moved off WordPress, each redirecting to the live URL. Consequently, an audit that treats that table as a ranking of pages will conclude the strongest content is a tools section that no longer exists.

So before analysing anything:

1. Export the Pages report for 16 months.
2. Strip the protocol and host, then resolve redirects. A quick crawl or a batch curl -sI over the list gives you the final URL for each string.
3. Group rows by final URL and sum clicks and impressions.
4. Mark any string whose final URL is a 404 as an inventory item in its own right. Those are redirect or removal decisions, not content decisions.

Only then does the spreadsheet describe pages.

Step 1: build the inventory and reconcile the three counts

Here are the three numbers to have in front of you before any judgement:

• URLs you publish. Anobee's sitemap: 94.
• URLs Google knows. The Page indexing report [4]: 518 known, 119 indexed and 399 not indexed, of which 174 are redirects and 135 are 404s.
• URLs earning impressions. The 16-month Pages report: 241.

The gaps are the content audit's opening findings. 241 earning URLs against 94 published means legacy addresses are still carrying a meaningful share of search visibility. Similarly, 119 indexed against 94 published means Google is indexing things outside your sitemap. Finally, the sitemap-filtered view of the indexing report, which on this site shows 73 of 94 submitted URLs indexed, tells you how the pages you actually care about are doing.

The Search Console setup guide covers getting these reports, and a crawler fills in the rest. Screaming Frog is free up to 500 URLs per crawl [5]. It returns titles, headings, status codes, canonical tags and word counts in bulk, which is everything the inventory needs that Search Console does not supply.

Step 2: attach the data that supports a decision

For each normalised URL, the minimum useful columns:

Sixteen months is not arbitrary. It is the longest window Search Console keeps, and it lets you compare a full year plus the run-up.

Step 3: find decay before it shows up as a traffic drop

Content decay rarely announces itself. The reliable early signal is impressions falling while clicks stay flat. Impressions move first when rankings slip from position 8 to position 15, while the page still catches a few clicks from brand or long-tail queries.

In the Performance report, set 16 months, switch on comparison with the previous period, open the Pages tab, and sort by impression change. Flag:

• Pages losing impressions with flat or rising clicks. Highest priority; they are still ranking but sliding.
• Pages that have lost five or more average positions. Second priority.
• Pages with high impressions and near-zero CTR. Often a title or intent problem rather than a content one.

Anobee's own case for the last category: this page recorded 17 impressions and 0 clicks at an average position of 58.8 over three months. At position 58 the CTR question does not arise, because the page is not being seen. That distinction matters. Ultimately the fix for position 58 is topical authority and links, not a new title tag.

Step 4: check intent match against the live SERP

Now open the target query in a private window and compare result types. If the top ten are step-by-step tutorials and your page is a definition piece, the intent has moved. If they are comparison pages and yours is a how-to, same conclusion.

Mark each page: match, partial, mismatch. A mismatch on a page with impressions is usually the highest-value rewrite in the audit, because the demand is proven and only the format is wrong. Meanwhile, the competitor content analysis process covers how to read the SERP properly rather than glancing at it.

Step 5: find cannibalisation, including the kind you created yourself

Two pages targeting one intent split signals. The usual check is to filter the Performance report by a query and see how many of your URLs appear for it. Then search site:yourdomain.com "target keyword" for a second view.

There is a second kind of cannibalisation that is entirely self-inflicted, and this site has it. The Search Console export for this very article lists two rows: /digital-marketing/content-audit-guide/ with 17 impressions at position 58.8, and /digital-marketing/content-audit-guide with 1 impression at position 99. One page, two URL strings, two positions. The slash version redirects and the canonical points at the no-slash form, so this will consolidate. Nevertheless, until it does the audit sees two entries.

Check three things agree for every page: the URL in the sitemap, the canonical tag, and the form your redirects settle on. Otherwise you get duplicate rows in every report you build for the rest of the content audit.

Step 6: score every page, then decide

Scoring exists to stop the content audit becoming a series of opinions. Five dimensions, 1 to 5 each:

Totals map to actions, and every URL gets exactly one:

Before deleting anything, confirm four things: no referring domains, no internal links you would break, no conversions in analytics, and no role in covering a topic your cluster needs. A page with zero traffic and two external links is a redirect, not a delete. Also keep Google's warning in view: content removed mainly to make a site look fresh does not earn rankings [2].

Step 7: execute the content audit, link, and re-check

Order the work by effort against expected return. Updates to pages already ranking on page two come first, merges second, deletions last. Deletions are the least reversible and the slowest to pay off [1].

When a refreshed page goes live, add internal links to it the same day from two or three related pages that already rank. This is the step most audits skip, and it is the one that gets the page re-crawled. Additionally, the technical SEO mistakes guide covers what else can hold a refreshed page back, and the topical authority guide covers where the page should sit in the cluster.

Then set the re-check. Record the date of each change in the audit sheet and look at the refreshed pages at 8 and 12 weeks against their own pre-change baseline. Keep the sheet rather than starting a new one next time. Ultimately, two dated columns on the same sheet turn an audit into a trend.

The content audit checklist

Before you start

• Write the goal in one sentence
• Export the sitemap
• Export 16 months of Search Console Performance data, Pages tab, with comparison
• Export the Page indexing report, both all-known and sitemap-filtered
• Crawl the site for titles, headings, status codes, canonicals and word counts
• Normalise URLs and resolve redirects before analysing anything

Per page

• Group all URL strings that resolve to it, and sum the metrics
• Record demand, trajectory, intent match, quality and business value
• Flag decay: falling impressions, flat clicks
• Flag cannibalisation, including duplicate URL forms
• Score and assign exactly one action

Before removing a page

• No referring domains
• No internal links that matter
• No conversions
• Not needed for topic coverage

After executing

• Internal links added to refreshed pages the same day
• dateModified updated only where content meaningfully changed
• URLs resubmitted and sitemap current
• Results checked at 8 and 12 weeks against the pre-change baseline

Common content audit mistakes

• Analysing URL strings instead of pages. The mistake this article's own audit hit first, and the reason nine redirecting URLs looked like top performers.
• Deleting on traffic alone. Links, internal links, conversions and topic coverage all outrank a zero in the sessions column.
• Pruning for freshness. Google says directly that this does not work [2].
• Expecting an immediate recovery. The helpful-content signal is reapplied over months [1].
• Refreshing without internal links. The page has no new path for a crawler to follow.
• Ignoring positions 5 to 20. These pages already have demand and standing; they are the cheapest wins in the sheet.
• Auditing once. Keep the sheet, add dated columns, and re-run the decay check quarterly.

Bottom line on running a content audit

A content audit is a decision-making exercise with a spreadsheet attached, not a spring clean. Normalise the URLs before you judge anything, because Search Console reports strings and a migration leaves plenty of them. Then reconcile what you publish, what Google knows and what earns impressions, and find decay while it is still only impressions. Give every URL one action, check links before deleting, and add internal links the day a refresh ships. Finally, keep the sheet so the next audit compares rather than restarts.