Running a Content Audit: Inventory, Scoring, and the Kill/Keep/Merge Decision

A content library that has grown for a few years without pruning typically carries real dead weight: pages nobody visits, drafts of the same idea written three times by three different people, posts about products that no longer exist, and thin pieces published in an era when volume was the strategy. This is not harmless clutter. Overlapping pages compete with each other for the same queries, outdated pages actively misinform prospects who land on them, and a large low-quality tail can drag on how the whole site is assessed.

There is an internal cost too. Writers duplicate existing work because nobody can see what exists, and every strategy conversation runs on vibes because nobody knows what the library actually contains. An audit converts an unknown liability into a known portfolio, which is the precondition for every other content operations decision: what to refresh, where the gaps are, and what the team should stop producing.

Start with a crawl of the site or an export from the CMS to enumerate every URL, then join in the data that turns a list into an inventory: organic traffic and impressions per page, backlinks, publish and last-updated dates, conversions or assisted pipeline where you can attribute it, and the page's topic and funnel stage. Perfection is not required. An inventory covering the meaningful pages with approximate data beats a stalled project waiting on perfect attribution.

Resist the urge to make judgment calls while inventorying. The inventory phase is mechanical on purpose: enumerate and enrich, nothing else. Teams that evaluate as they collect end up litigating individual pages for weeks and never finishing the sweep. Collect first, judge second, in bulk, with criteria you set before looking at any specific page you feel attached to.

Elaborate scoring rubrics with a dozen weighted factors usually collapse under their own ceremony. Two axes carry most of the decision: does the page perform, meaning it earns traffic, links, or conversions, and is the page still right, meaning accurate, on-strategy, and something you would publish today. Performing and right means keep, and possibly nominate for refresh investment. Performing but wrong is the dangerous quadrant: an outdated page with real traffic is misinforming actual prospects and deserves the fastest fix.

Not performing but right is a distribution question before it is a content question; the piece may deserve internal links and promotion rather than deletion. Not performing and not right is the kill pile. The quadrant model matters because it prevents the single most common audit error, judging pages purely on traffic, which would keep the popular-but-wrong pages and kill the good-but-buried ones, precisely backwards on both counts.

Merging is the highest-value move and the least used. When three mediocre pages cover one topic, consolidating them into a single strong page under the best-performing URL typically outperforms all three, because their authority and internal links stop being split. Redirect the retired URLs to the merged page, carry over anything genuinely valuable from each, and update internal links to point at the survivor. For kills, redirect to the closest relevant page where one exists; a redirect to an unrelated homepage helps nobody.

Ship the changes in batches and watch what happens, rather than executing hundreds of deletions in one dramatic afternoon. Batching makes it possible to notice if a supposedly worthless page was quietly supporting rankings elsewhere, and to reverse course cheaply. Then put a date on the next audit. A library that gets audited on a standing cadence, often annually, never again requires the archaeology that the first audit demanded, because the portfolio never gets more than a year out of known state.