What is a content audit?
A review of the pages you already have, ending in a decision about each one. Written for somebody who has three hundred pages and a suspicion that most of them are not earning their place.
A content audit is a review of every page you have already published, ending in one of four decisions for each: keep, update, merge or delete. It is not a technical check. It asks whether a page still deserves to exist, which is a judgement about your business rather than a measurement of your markup.
- Answers
- Should this page exist, and if so what does it need?
- Output
- A decision per URL, not a spreadsheet of metrics.
- Cadence
- Annually for most sites. Quarterly if you publish weekly.
- Most common correct answer
- Delete or merge. No other audit in this discipline says that.
Why the pages you already have are the cheapest thing you own
Every published page carries a cost that nobody records. It is crawled, so it spends budget. It competes with your other pages for the same queries. It represents your judgement to anybody who lands on it, including a model deciding whether you are worth quoting. And it accumulates: a site that publishes weekly for four years has two hundred pages, most of which nobody has read since the week they went live.
The reason to audit rather than to keep publishing is arithmetic. A page that already exists, already has some history, and already ranks somewhere respectable is far cheaper to move than a new page is to launch. Updating twenty of those is usually a better quarter than writing twenty more, and it is almost never what a content plan proposes, because writing something new is easier to scope and easier to celebrate.
The uncomfortable half is that a real audit ends with deletions. Not archiving, not noindexing, not "we will get to it": removing pages that will never be good and never mattered. That is the part that makes this exercise different from every other audit, and it is why so many stop at the inventory.
What to put in the inventory, and what to leave out
The inventory is the easy half and it is where most audits spend all their time. Pull one row per URL and resist adding columns you will not use to make a decision.
The URL, the title and the date it was published
Three columns that let you sort by age, which is the first cut on most sites: anything untouched in three years is either evergreen or abandoned, and you know which by reading it.
Sessions or clicks over the last twelve months
Twelve, not one, so seasonality does not sentence a page to death in its quiet quarter. A page with zero over a full year is a decision waiting to be made.
Impressions and average position from Search Console
The single most useful pair, because it separates two very different zeroes. Zero clicks with impressions means the page is found and not chosen. Zero clicks with no impressions means it was never in the running.
Conversions, if you can attribute any
Attribution is imperfect and it does not have to be perfect to be decisive. A page with traffic and no conversion in a year is not necessarily a delete; it is a page you should stop treating as commercial.
What it is for, in your own words
One free text column, filled in by a person. This is the column that makes the audit useful and the one most inventories skip, and a page nobody can write a purpose for has answered the question.
The four decisions, and the questions that produce them
Every URL gets exactly one. Allowing a fifth option, in any form, is what turns an audit into a document: "revisit later" is not a decision, it is the absence of one, and it is where content audits go to die.
Three questions, in order, settle almost every page.
Does anybody search for what this is about? If the topic has no demand, no amount of rewriting produces traffic. That is a delete or a merge, and it is not a criticism of the writing.
Is this the best page on this topic that we could publish? If a competitor covers it better and we have nothing to add, we are publishing a worse version of something that already exists. Google says the same thing in its own words: content restating what is already available, however competently, is what it calls commodity content.
Would we publish this today? The most useful question on the list and the least technical. If the honest answer is no, the page is either an update or a delete, and the difference is whether the topic still matters to the business.
The trap: an inventory is not an audit
The most common failure is a beautiful spreadsheet with four hundred rows, fifteen columns, conditional formatting, and no decisions in it. It gets presented, admired and never opened again.
This happens because the inventory is mechanical and the decision is not. Pulling metrics can be automated, so it feels like progress and it can be finished. Deciding that a page somebody wrote should be deleted is a conversation, and it is easier to add another column than to have it.
Two rules prevent it. Fill the decision column before you add another metric column: if a metric would not change any decision, it is decoration. And cap the audit at what you will actually act on this quarter. Fifty URLs decided and executed is worth more than four hundred URLs described, and the four hundred row version is why nobody trusts the exercise.
What to do with each decision
Keep means untouched, and it should be the majority on a healthy site. Note the date so the next audit knows it was looked at rather than missed.
Update means the topic is still right and the page is not. Refresh what is stale, add what is missing, and change the modified date only because the content changed. Bumping a date on a page nobody edited is a pattern that is easy to detect and gains nothing.
Merge means two or three pages are competing for one query and none of them is winning. Combine them into the strongest URL and redirect the others to it permanently, which is a 301 rather than a 302 because they are not coming back.
Delete means the page has no future. Return 410 if nothing links to it and 404 is fine too; redirect only where there is a genuine equivalent, because a redirect to something unrelated is a soft 404 with extra steps. Deleting pages nobody reads does not harm a site, and leaving hundreds of them does not help it.
The decision table
Read the row that matches what the data says, and the decision is usually already made. The last column is the one to look at when two rows nearly fit.
| What the page looks like | Decision | What you actually do | The judgement call |
|---|---|---|---|
| Traffic, conversions, current | Keep | Nothing. Record the date it was reviewed. | Resist improving pages that are working. |
| Impressions, few clicks, position 5 to 20 | Update | Rewrite the title and the opening. Answer the question earlier. | The cheapest win available on most sites. |
| Real traffic, factually out of date | Update | Fix the facts, then change the modified date. | Never the date without the facts. |
| Three pages, one topic, none ranking | Merge | One strong page, permanent redirects from the rest. | Choose the URL with the most history, not the best draft. |
| No impressions, no links, no purpose | Delete | 410 or 404. Redirect only where there is a real equivalent. | This is the row people negotiate with. Do not. |
| No traffic, but it exists for a reason | Keep | Nothing, and stop measuring it as a traffic page. | Legal pages, support answers and proof pages are not failures. |
The last row matters more than it looks. A content audit run purely on traffic will recommend deleting your refund policy, your accessibility statement and the page that answers the question your support team is asked most often. Traffic is the input; purpose is the veto.
Running one, in an order that finishes
Two sittings for most sites: one to build the inventory, one to decide. Do not attempt both in the same afternoon, and do not let the first one expand to fill the time.
- 01
Export the URL list from your sitemap, not from your CMS
The sitemap is what you are asking to have indexed, which is the set that matters. A CMS export includes drafts, revisions and screens nobody can reach, and cleaning that up is an hour spent on rows you were going to ignore.
- 02
Join twelve months of Search Console data to it
Clicks, impressions and average position per URL. This is the only data source in the exercise that tells you what happened in search rather than what happened on your site, and the impressions column is what separates a page nobody found from a page nobody chose.
- 03
Rule out the technical explanations first
A page with no impressions may be excellent and unreachable. Before judging content, confirm the page is indexable and its canonical resolves, because a third of professional sites fail the second of those. Deleting a good page for a robots problem is the worst outcome this exercise can produce.
- 04
Sort by age and read the oldest fifty
Read them, do not skim the titles. Most decisions become obvious within a paragraph, and the ones that do not are the ones worth discussing.
- 05
Write one decision per row, with no fifth option
Keep, update, merge or delete. If a row genuinely cannot be decided, the missing information is usually purpose rather than data, and one sentence from whoever commissioned the page settles it.
- 06
Execute the deletions and merges before the updates
They are faster, they are the ones that get talked out of if you leave them until last, and the merges tell you which updates were about to be written twice.
How to tell whether the audit worked
The wrong measure is total pages, and it is the one most reports lead with. Removing eighty pages is not an achievement and neither is keeping them; the number of URLs on a site is an input rather than an outcome.
The useful measures are three, and all of them are in data you already have. Clicks on the pages you updated, compared against the same pages before, over a period long enough to survive noise. Average position on the merged topic, where the question is whether one page now does better than three did. And the share of your indexed pages that got any impressions at all, which is the closest thing to a health figure this exercise produces.
Give it a quarter. Search Console reports on a delay, recrawling takes time on any site and considerably longer on a small one, and merges need the redirects to be followed and the signals consolidated before anything moves. An audit judged after three weeks will look like it did nothing.
One thing not to measure: whether the pages you deleted are missed. They will not be, and looking for it invites somebody to argue for restoring them. The evidence for the deletion was gathered before it happened, which is the point of doing the inventory first.
The technical half, which you need before you judge anything
A content audit assumes a page could be found. These are the checks that tell you whether that assumption holds, and they are the difference between deleting a weak page and deleting a blocked one.
Related reading: the SEO report template, with a content actions tab, what an SEO audit actually checks, the technical SEO checklist, how to do keyword research, the Search Console analysis prompt.
Questions people ask about content audits
What is a content audit?
A review of the pages a site has already published, ending in a decision about each one: keep it, update it, merge it into another page, or delete it. It is a judgement about whether a page deserves to exist rather than a technical check of whether it works.
How often should I run a content audit?
Once a year for most sites, and quarterly if you publish weekly or more. The trigger is usually the same in either case: the moment nobody on the team can say with confidence what is on the site, an audit is overdue.
What should a content audit include?
One row per URL with the title, the publish date, twelve months of clicks and impressions, any conversions you can attribute, and a free text column saying what the page is for. Then a decision column. Any metric that would not change a decision is decoration.
Does deleting old content hurt SEO?
Deleting pages that get no traffic, have no links and serve no purpose does not harm a site. What does harm one is deleting pages that are earning quietly, so check impressions and referring links before removing anything, and merge into a stronger page rather than deleting where the topic still matters.
What is the difference between a content audit and an SEO audit?
An SEO audit asks whether a machine can reach, read and understand your pages. A content audit asks whether those pages should exist. The first is largely automatable and answers in minutes; the second is a judgement about your business and cannot be delegated to a tool.
How long does a content audit take?
Two sittings for a site of a few hundred pages: one to build the inventory, one to decide. The decisions take longer than people expect and the inventory takes less, which is the opposite of how most audits are scheduled and part of why so many stall.
Should I noindex old content instead of deleting it?
Sometimes, and it is often a way of avoiding the decision. A noindex is right for pages that must stay reachable for people but should not be in search. If the page has no reader and no purpose, noindexing it leaves a page nobody maintains and nobody will ever revisit.
What is the fastest win a content audit produces?
Pages with impressions and few clicks, sitting somewhere on the first two pages of results. The topic is proven, the page is already found and not chosen, and rewriting the title and the opening paragraph is an afternoon rather than a project.
Rule out the technical explanations first
A page with no impressions may be excellent and unreachable, and deleting it for a robots rule is the worst outcome an audit can produce. The free audit checks indexability, canonicals and crawler access across your site in a couple of minutes, before you judge a single word.