How to Do a Manual Content Audit: Method, Template and Tools

A content audit answers one question for every page on your site: should it stay as it is, be updated, be merged, be redirected or be removed? Done once a year, it keeps a site accurate and easy to maintain. Done only after a traffic drop, it becomes an emergency.

This guide gives you a manual method that works well up to a few hundred pages. We learned its limits on FunInformatique, the French-language tech site we have run for more than 15 years: listing 2,000 articles and sorting them by traffic was the easy part. Checking what each article actually says was the part that did not scale.

How long a manual audit really takes

Building the inventory and the spreadsheet takes hours. Reading and checking the pages takes weeks, and the time depends less on the number of pages than on what they contain.

When we analyzed 23 of our own tutorials claim by claim, a typical article contained about 16 statements a reader relies on (a version, a menu path, a price, a rule, a recommendation), 14 of them important. Each one has to be compared with a current, official source. Count your pages, multiply by that number, and you have a realistic idea of the work in step 3.

1. Build a complete URL inventory

Before analyzing anything, you need to know what exists. You will usually find more URLs than expected: old tag and archive pages, test pages, forgotten landing pages. Cross-reference three sources:

2. Build the analysis grid

Some columns fill in from your exports; the last ones require reading the page.

Column	Source	What it tells you
URL and title	Crawl	The page and its topic
Clicks and impressions (12 months)	Search Console	What the page brings today, and the visibility it still has
Change in clicks over a year	Search Console (Compare)	Pages in decline
Publication and last update dates	CMS	Where to look first, not proof of anything
Backlinks and internal links	Search Console Links, Ahrefs or Semrush, crawl	What you would lose by deleting the page
Claims checked	Reading the page	How many statements you verified, and how many were wrong
Source for each correction	Official documentation	The proof you will need when you edit the page
Decision	You	Keep, update, merge, redirect or remove

The two columns in bold are the ones most audits leave out, and the ones that changed our own decisions. Dates and traffic told us which articles to open first. They did not tell us what was wrong: in our articles, only one error in four came from something that had changed since publication. The rest had been there from the start, or came from a sentence that promised more than the facts allow. The full data is here.

3. Read, check and classify each page

This is the longest step. For each page, list the statements a reader relies on, check each one against the primary source, and record the result in the grid. Then classify the page:

Set your own thresholds for traffic and dates according to your site. A threshold that suits a news site will not suit a site of evergreen guides.

4. Prioritize

Sort the pages to update with two questions: does the page still matter, and how much can its errors cost the reader?

• First, errors that can cost money, data or security, whatever the page's traffic. In our own articles, the most serious corrections were a free trial announced at 30 days when the publisher now lists 7, a plugin no longer maintained still recommended, and software to install on the very computer you want to examine.
• Then, pages that still bring traffic and contain wrong facts: readers are reading them today.
• Then, pages in decline on topics people still search for.
• Last, merges and removals, which take time to plan but rarely affect a reader directly.

5. Correct and measure

• Work in batches. Small batches let you see the effect of each change and undo a mistake before it spreads.
• Record the date of every change in the grid. Without it, you cannot link a traffic change to your work.
• Request indexing of your most important corrected pages with the URL Inspection tool in Search Console. Recrawling "can take anywhere from a few days to a few weeks" (Google), and Google says its systems can take several months to recognize larger improvements.
• Check again later. We re-analyzed two of our corrected articles three weeks after the fix, with a newer version of our engine: both were still accurate. That second check is what tells you a correction holds.

---

Summary: the five steps

Step	What you do	What you get
1	Inventory from Search Console, sitemap and crawl	One row per URL
2	Grid with traffic, dates, links, claims and sources	A working document
3	Read, check against official sources, classify	A decision for each page, with its proof
4	Prioritize by risk for the reader, then by traffic	An ordered action plan
5	Correct in batches, date each change, check again	Measured results

To compare this method with an automated check, see EEATClean vs manual content audit. To find the outdated pages before a full audit, start with how to find outdated content.

Tools and sources

• Google Search Console: performance and page indexing reports.
• Screaming Frog SEO Spider: crawl and metadata export.
• Ahrefs or Semrush: backlink data, to check before any removal.
• Google: Page indexing report, Ask Google to recrawl your URLs, Google Search's core updates.