A content audit is a review of every page you have published, scored against what it is actually doing for you, ending in a decision for each one: keep it, improve it, merge it into something else, or remove it.
The hard part is not collecting the data. It is deciding, page by page, which of those four a page gets. The sections below give the thresholds for each decision.
What a content audit is not
It is not a technical site audit. Broken links, redirect chains, page speed and crawl errors are a different job with different tools, and that work is covered in our SEO audit checklist.
It is also not a content inventory. An inventory is the list of what exists. The audit is the judgement you apply to that list. You need the inventory first, and having one is not the same as having audited anything.
Keeping the two apart matters, because mixing them produces a spreadsheet with ninety columns that nobody finishes.
When it is worth doing
Running one too early wastes a week. The honest triggers are these.
- You have more than about fifty published pages. Below that you can hold the whole site in your head.
- Traffic has been flat or falling for two quarters while you kept publishing. That pattern usually means new work is being cancelled out by old work decaying.
- Your indexed count is well below your published count. Google is already auditing you and rejecting things.
- You inherited the site. Nobody can improve what they have not read.
- Annually, as maintenance. Not more often, because pages need time between changes for you to learn anything.
The data to pull, and nothing more
Four sources cover it. Resist adding a fifth.
- Your page list. Export the sitemap rather than trusting your memory or your menu, since orphaned pages do not appear in either.
- Search Console, by page. Clicks, impressions and average position for each URL. Note that Search Console keeps only the last 16 months, so if you want year on year comparisons you need to have been exporting before now.
- The indexing state of each URL, from the Pages report. A page that is not indexed is not underperforming, it is absent, and that is a different problem.
- Publication or last-updated date. A page from last month and a page from 2019 with identical numbers are not in the same situation.
Conversions are worth adding if you have enough of them to be meaningful. If you get three a month, they will mislead you more than they help.
The decision table
This is the part that is usually missing. Sort every URL into one of four outcomes using the numbers you just pulled, over a twelve month window.
Leave alone. Clicks are steady or rising and position is inside the top ten. The page is working. Resist improving it, because the most common way to lose a ranking is to rewrite the page that had it.
Improve. Impressions are healthy but position sits between roughly 11 and 30. Google already thinks the page is relevant and is not convinced it is the best answer. This group gives the fastest return in the whole audit, and it is the same logic behind striking distance keywords.
Merge. Two or more pages draw impressions for the same queries, and none of them ranks well. They are competing with each other. Pick the strongest URL, fold the useful parts of the others into it, and redirect them.
Consider removing. No clicks, almost no impressions, older than twelve months, no links pointing at it, and no business reason to exist. Every one of those five has to be true, not just the first.
Anything that does not land cleanly in one of the four goes in a fifth pile marked “decide later”, and you move on. Arguing over edge cases is how audits stall.
Auditing a site that has almost no traffic
Every method above assumes you have enough clicks for trends to mean something. On a small or new site you do not, and applying those rules will tell you to delete nearly everything.
Use different signals at low volume.
- Indexed or not. This is the single most useful column on a small site. If a page is not indexed, no amount of editing matters until that changes, and our notes on crawled, currently not indexed explain why Google declines.
- Impressions rather than clicks. A page with 400 impressions and no clicks is being shown and passed over, which is a content and title problem you can fix. A page with four impressions is not being shown at all, which is a different problem.
- Query count. A page appearing for thirty queries has found a subject. A page appearing for one is a coin toss.
- Internal links in. Count them. On a small site, pages nothing links to are usually the ones Google ignores too.
And at low volume, lean heavily towards improving and merging rather than deleting. You do not yet have the evidence to say a page failed, only that it has not been given a chance.
Why deleting is riskier than it sounds
Pruning gets recommended confidently, and the downside gets mentioned rarely. Four things go wrong.
You delete the links. A page with no traffic may still be the only page on your site another website chose to link to. Check for external links before removing anything, because that link disappears with the page.
You create 404s at scale. Every removed URL that something still points at becomes an error. These accumulate: our own Search Console currently reports 92 pages under “Not found (404)”, and that is without a pruning exercise.
You lose impressions you never counted. A page earning no clicks can still be earning impressions across dozens of long tail queries, and those are the queries that eventually turn into clicks as a site gains authority.
You cannot undo it cleanly. Republishing a deleted page does not restore its history or its rankings.
The safer default is to merge and redirect rather than delete. You keep the link equity, you keep the content, and you remove the duplication. Reserve deletion for pages that are genuinely worthless, such as test pages, duplicate imports and tag archives nobody reads.
What improving a page actually means
“Update the content” is not an instruction. For the pages in the improve pile, work in this order.
- Check the query it already earns impressions for, not the one you intended it to rank for. These are often different, and the page should serve the real one.
- Compare it against what currently ranks. If the top results all answer a question your page skips, that gap is the job.
- Fix the title and description if impressions are high and clicks are not. That combination is a click problem, not a content problem.
- Add internal links into it from your strongest related pages. This is the cheapest improvement available and the most skipped.
- Change the date only if you changed the content. Refreshing a date without touching the page is a trick that fools nobody worth fooling.
Then leave it alone for at least eight weeks. Changing a page repeatedly destroys your ability to tell what worked.
Making it repeatable
A one-off audit decays the moment it is finished. The version that keeps paying is a short recurring check.
Once a quarter, pull the same four data points and look only at what changed category. Pages that fell out of the top ten, pages that entered the 11 to 30 band, and pages that stopped being indexed. That is a half-day job rather than a week.
Keep the spreadsheet, with the date of each audit in the filename, and record what you decided for each URL and why. The second audit is far more useful than the first, because you can finally see whether your improvements did anything. Score those outcomes against the measures in your SEO KPIs.
Mistakes that waste the week
- Auditing everything at once. Start with one section or one cluster, finish it, then move on.
- Collecting data you will not act on. If a column will not change a decision, do not create it.
- Treating zero traffic as proof of failure on a page that was never indexed, or never linked to from anywhere.
- Pruning on SEO data alone. A page can earn nothing from search and still close deals, answer support questions or exist for legal reasons.
- Improving pages that already rank. The improve pile is positions 11 to 30, not positions 1 to 5.
- Doing it again too soon. Annual is enough, with quarterly checks on what moved.
Frequently asked questions
What is a content audit?
A content audit is a review of every published page, measured against what it actually delivers, ending in a decision for each one: leave it alone, improve it, merge it into another page, or remove it. It is separate from a technical site audit, which deals with crawling, speed and errors. It also needs a content inventory first, since you cannot judge pages you have not listed.
How often should you run a content audit?
Once a year for a full audit, with a lighter quarterly check on what changed category. Running one more often does not help, because pages need time between changes before you can tell whether an edit worked. The quarterly check only looks at pages that dropped out of the top ten, entered the 11 to 30 band, or stopped being indexed.
Should I delete pages that get no traffic?
Usually not, and merging is the safer default. A page with no clicks may still hold the only backlink another site ever gave you, may earn impressions across many long tail queries, and may exist for a business reason unrelated to search. Only remove a page when it has no clicks, almost no impressions, is over a year old, has no links pointing at it, and has no reason to exist.
Which pages should I improve first?
The ones with healthy impressions sitting between roughly position 11 and 30. Google already considers them relevant and is not yet convinced they are the best answer, so the distance to a real gain is shortest there. Pages already in the top ten should be left alone, because rewriting a page that ranks is a common way to lose the ranking.
How do you audit content on a site with very little traffic?
Use different signals, because click trends are meaningless at low volume. Look at whether each page is indexed at all, at impressions rather than clicks, at how many distinct queries a page appears for, and at how many internal links point to it. At low volume you should lean strongly towards improving and merging, since you have no evidence a page failed, only that it has not had a chance.
What data do I need for a content audit?
Four things: a complete page list taken from your sitemap, clicks, impressions and average position per URL from Search Console, the indexing state of each URL, and the publication or last-updated date. Search Console only retains 16 months of performance data, so year on year comparisons require exports you made earlier. Add conversions only if you have enough of them to be meaningful.