A content audit is a systematic evaluation of every page a site publishes, ending in a decision for each one: keep it as it is, merge it into something else, rewrite it, or delete it. It is the maintenance discipline of content and what converts an ever-expanding library of never-to-be-seen thoughts, into a deliberate portfolio of expertise and sales-enablment.
It is also, on most sites past a few hundred pages, the highest-return piece of content work available, and the least popular. It is where a content strategy meets what was actually published while nobody was looking. Nobody is promoted for consolidating 40 posts into 8.
A content inventory is not a content audit
These get used interchangeably and they are different activities with different outputs.
An inventory is a complete list of what exists, with attributes: URL, title, template, author, publication date, last modified date, word count, internal links in and out, canonical target. It is descriptive and it can be assembled mechanically (a crawl joined to your CMS export).
An audit adds judgment and performance data to that list and produces a verdict per row. It asks whether each page still earns its place, which requires knowing what the page was for. That is why audits cannot be fully automated and why the ones that are read like a list of pages with low traffic, are not a finding.
The data join that makes an audit possible is three sources against the inventory: the Search Console bulk export for impressions, clicks and the queries each URL actually matches; analytics for engagement and conversion; and a backlink index for external links, because a page with no traffic and 90 referring domains is not a deletion candidate, it is a redirect candidate.
The four verdicts of a content audit
- Keep. The page performs, or serves a purpose that is not measured in traffic — a pricing page, a security overview, the thing sales sends after the demo. Note the owner and the next review date and move on.
- Merge. Several pages address the same question and none of them fully. Combine into the strongest URL, redirect the others with a
301, and rebuild the internal links that pointed at the retired versions. This is the verdict with the best return and the most work. - Rewrite. The page addresses a real question and does it badly, or did it well 5 years ago. Distinguish a refresh (updating facts, screenshots, product names) from a genuine rewrite, because refreshes are cheap and get confused with the expensive version in planning.
- Delete. The page serves nobody. It has no traffic, no links, no internal purpose, and no prospect of either. Return
410if it should be gone permanently and there is nothing to redirect to; use a301to a genuinely relevant parent when one exists. Redirecting everything to the homepage is a soft 404 and Google treats it as one.
Assign the verdict before you estimate the effort. Doing it the other way round guarantees that everything expensive becomes a keep.
Content decay
Content decay is the steady decline of a page that once performed, and it is the most common finding in any audit of a site older than three years. It has three usual causes, and the fix differs for each.
The page may have gone stale; for example, it might describe a product version, a price, a regulation or an interface that has changed. The market may have moved past it, with competitors publishing something materially better while the page sat untouched. Or the query itself may have changed shape, which is the one teams miss: the search result for a term can shift from articles to tools, or from overviews to comparisons, and a page that was the right format three years ago is now the wrong format entirely.
Look at the queries a decaying page currently matches in Search Console, not the ones it was written for. That comparison usually explains the decline in a minute, and it tells you whether the answer is a refresh, a rewrite, or accepting that the page has no future in that shape.
Content cannibalization
Cannibalization is what happens when several of your own pages compete for the same query and the search engine keeps changing its mind about which one to show. The symptom is an unstable position on a term you care about, with the ranking URL swapping between pages week to week.
To find it, pull the query-to-page report from Search Console and look for queries where more than one URL has earned impressions over the same period. Ignore the long tail as overlap there is normal and harmless. Concern yourself with the head terms where you are ranking in positions five to fifteen with two or three pages splitting the signals.
The usual fix is merging, which returns you to the verdict list. Occasionally the right fix is differentiation instead: two pages that genuinely serve different intents but were written as if they served one. The test is whether you can state, in a sentence each, what different person each page is for.
Why content audits stall
The technical work of an audit is a few days. The reason most of them end as an unopened spreadsheet is organizational.
Nobody wants to be the person who deleted something. There is no upside to being right and a permanent record of having been wrong.
The author of a page is often still in the building. The campaign that produced 30 posts had an executive sponsor. Somebody once presented that content as a win in a board deck. Deletion asks a person to accept that work they were rewarded for no longer has value, and no dashboard makes that conversation comfortable.
Three things make it move. First, get the deletion decision made in a batch with a defined rule rather than page by page. For example, a rule about traffic, links and purpose is arguable, a judgment on your colleague’s article is personal. Second, escalate the decision right to someone senior enough that it is not a peer overruling a peer. Third, do it in stages: remove from navigation and internal links first, watch for a month, then retire. That staging costs a month and removes almost all of the political risk.
The concession worth making: deletion is not, in itself, a ranking tactic. Removing 300 dead pages from a 500,000-page site is unlikely to produce a visible site-wide lift, and anyone promising one is overselling. The case for it is maintenance: a crawl budget spent on pages that matter, an archive you can actually govern, and a site where the best page on a topic is the one that ranks.
What to do first
Run the inventory and join it to the Search Console export before you form any opinion. Then sort by impressions rather than clicks: pages with impressions and no clicks are usually rewrite or merge candidates and they are the fastest wins in the whole exercise.
Take the first 20 verdicts through the full process: decision, execution, internal link rebuild, monitoring, before auditing the remaining thousand. An audit that ships 20 changes is worth more than one that catalogues everything and changes nothing, which is what most content programs have sitting in a drive somewhere.