How to Run a Content Audit

Most content audits die in the inventory. Someone exports every URL into a spreadsheet, highlights a hundred rows in three colors, feels productive, and never makes a single decision, which was the only part that mattered.

An audit is not a census. It is a sequence of verdicts. The inventory exists only to make the verdicts possible, and there are exactly four of them, keep, merge, cut, or refresh. A page you cannot assign one of those to is a page you have not finished auditing.

An Audit Produces Verdicts, Not a Spreadsheet

The spreadsheet is a means. Somewhere along the way it became the deliverable, which is how audits turned into an annual ritual that changes nothing.

Reframe it before you start. The output of an audit is a list of actions with a page attached to each. Every row ends in a decision, and a row without one is unfinished work masquerading as analysis. If you run an audit and nothing gets rewritten, merged, or removed, you did not run an audit. You maintained a filing system.

This matters because a cluster drifts. Pages you wrote eighteen months ago now overlap with newer ones, older pieces went stale while you were not looking, and a few never linked into anything. An audit is how you keep a content strategy that ranks honest over time, by catching the drift before it spreads through the cluster. The census tells you what you have. The verdicts are what the audit is for.

Inventory Everything, Then Stop Admiring It

You do need the inventory. You just need to spend an hour on it, not a week.

List every published URL in one place. For each, capture the cluster it belongs to, or a blank where it belongs to none. Add a traffic trend rather than a raw pageview count, because direction tells you more than a single number. Add the date of the last meaningful update, not the last time a typo was fixed. Then add the column that decides most verdicts.

That column is one line answering a single question. What does this page add that the pages already ranking do not? A page with a confident answer there is usually a keep. A blank in that column is the loudest signal in the entire sheet, because a page that adds nothing is competing against the thing it summarizes, and losing. That is what information gain means in practice, applied one row at a time. Everything else in the inventory is context. This column is the verdict in waiting.

Resist the pull to keep enriching the sheet. More columns feel like more rigor and are usually procrastination. Five columns that lead to a decision beat fifteen that lead to another meeting.

Find the Thin, the Redundant, and the Orphaned

Three kinds of page do most of the damage, and each is easy to miss when you are looking at a hundred green rows.

Thin pages add nothing. They are grammatical, structured, sometimes even long, and they say what the top ten results already said. Length is not the tell, information gain is. A 2,000 word page that summarizes is thinner than a 600 word page with one original observation.

Redundant pages compete with each other. Two or three articles half-answer the same query, split the signal between them, and none of them ranks as well as one merged piece would. This is the failure people misdiagnose as a niche being too crowded, when the crowd is all their own pages.

Orphaned pages sit outside the structure. Nothing links to them, so they are a page in a folder rather than a piece in a cluster. An orphan can be individually excellent and still underperform, because the internal links are the structure and a page with none is invisible from the rest of the site. Orphans are the most recoverable of the three. Often the fix is a verdict of keep plus two internal links you should have added at publication.

Keep, Merge, Cut, or Refresh

Every flagged page gets one verdict. Not two, not a maybe. The discipline of forcing a single choice is what turns the inventory into an action list.

Verdict	What It Means	The Action
Keep	Ranks or earns traffic, and adds something the alternatives do not	Leave the content alone. Confirm it links into its cluster both ways, then move on
Merge	Two or more pages split the signal for one query	Combine into one stronger page, redirect the losers to it, reclaim their links
Cut	No traffic, no links, nothing unique, no path to the other three	Redirect to the nearest relevant page, or remove it. Deadweight dilutes the subject
Refresh	Good bones, but stale, thin, or a summary from the start	Rewrite to add what the current top results lack, update the date, republish

Refresh is the verdict people over-use, because it feels like the kind option. Be honest about it. A refresh is a real rewrite that adds new information, not a date change and two fresh sentences. If the page was a summary the day it published, refreshing it means giving it the thing it never had, and if you are not willing to do that, the honest verdict is cut.

The Audit, Step by Step

The process is short once the verdicts are the goal rather than the sheet.

• List every published URL in one place, with the cluster each belongs to or a blank where it belongs to none.
• Add a traffic trend and a last-meaningful-update date to each row, not just a raw pageview count.
• Mark the one thing each page adds that its competitors do not. A blank in this column is the loudest signal in the sheet.
• Flag the thin, the redundant, and the orphaned, because those three account for most of the damage.
• Give every flagged page one verdict and only one, keep, merge, cut, or refresh. No page leaves the audit unmarked.
• Execute the merges and cuts with redirects, so the links and earned authority are reclaimed rather than lost.
• Feed the refresh list and the gaps it exposes into the calendar, so the audit becomes next quarter's plan.

That final step is what separates an audit from a spring clean. The refreshes and the gaps you found are the raw material for the next stretch of the content calendar, which means the loop closes. You are not choosing between maintaining old pages and planning new ones. The audit is how you decide which pages are worth the return they earn, and that decision feeds directly into what you commission next.

Frequently Asked Questions

How often should I run a content audit? Once or twice a year for most sites, or whenever a cluster feels like it has stopped compounding. Auditing monthly is usually procrastination in a spreadsheet. The signal that it is due is drift, overlapping pages, stale dates, and pieces that never linked into anything.

What is the difference between a content audit and a content refresh? The audit is the diagnosis, the refresh is one of its four prescriptions. An audit looks at every page and decides keep, merge, cut, or refresh. A refresh is what you do to the specific pages the audit marked as good bones gone stale.

Should I delete old blog posts? Sometimes, and it helps more than it feels like it should. A page with no traffic, no links, and nothing unique is diluting the subject it sits in. Redirect it to the nearest relevant page so any earned authority is reclaimed rather than thrown away.

What tools do I need to run a content audit? Less than the market implies. Your analytics for the traffic trend, your search console for what each page actually ranks for, and a spreadsheet for the verdicts. The judgment about information gain is human, and no tool assigns the verdict for you.

Does merging two articles hurt my rankings? Done properly it helps, because it stops two pages from splitting one query’s signal. Combine the stronger content into one page, redirect the weaker URLs to it, and update the internal links that pointed at the old ones so nothing is left dangling.

How Postdex Handles the Verdicts

Once the audit hands you a verdict, what it points to is a commission, not a clean-up. A refresh verdict calls for rebuilding a thin or stale page into one that earns its place. A gap, or a summary you cut and still need covered, calls for a new piece to take that slot. The audit sorts the library into these two lists, the refreshes and the gaps, and both are work to hand off rather than tidy.

The commissioning desk takes either one as a brief, a refresh grounded in what the current top results lack, or a new cluster piece written to slot into the structure you already have, with the internal links written in rather than added after. When the gap list is longer than your calendar can absorb, the catalog lists finished articles by the question they answer, each an edition of one, so you can fill a hole this week instead of waiting on a queue.

Run the census in an hour, spend the real time on the verdicts, and act on every one of them. An audit that ends in a color-coded spreadsheet and no redirects was a way of feeling productive without changing anything.