Libraries grow quietly. A few years of calendars, product launches, agency handoffs, and abandoned experiments leave hundreds of URLs with unclear owners. Some still earn; many dilute. An audit exists to restore intentionality. It is not a design refresh, and it is not a keyword tool export with traffic columns.
The output should be a decision for each important URL and a ranked backlog for the next two quarters. If the output is only a slide saying "content hygiene matters," you ran a workshop, not an audit.
Define the job of the audit before you export data
Pick the primary job. Reduce cannibalisation. Recover decaying winners. Free crawl and human attention by pruning. Find evaluation gaps. Most teams need a blend, but one job should lead so scoring does not become mud. Align that job with current strategy priorities so the audit does not resurrect old themes you already abandoned.
- Scope: blog only, or all indexed resources including guides and comparison URLs.
- Timebox: usually three to six weeks for a mid-size library.
- Owners: content, SEO, product marketing, and someone who can approve removals.
- Success: a decided backlog, not a perfect taxonomy.
Build the inventory with fields that force decisions
Export URLs from the CMS and Search Console. Add analytics sessions, conversions if available, referring links if you track them, last updated date, index status, and target query if known. Then add human fields: buying stage, primary role, accuracy risk, and strategic fit (core, adjacent, off-strategy).
| Score | What it captures | High means | Typical action |
|---|---|---|---|
| Demand evidence | Impressions, sessions, rankings | Search still cares | Protect or refresh |
| Business fit | ICP and pipeline relevance | Page supports current offer | Keep in cluster |
| Accuracy risk | Product or claims drift | Likely to create objections | Refresh or unpublish fast |
| Distinct intent | Overlap with other URLs | Unique job | Keep separate |
| Experience depth | First-hand detail vs generic advice | Hard for AI answers to replace | Invest further |
Find decay, cannibalisation, and gaps
Decay shows up as falling impressions and clicks on pages that once held stable positions, often after product changes or SERP feature shifts. Cannibalisation shows up as multiple URLs ranking unstably for the same query with overlapping briefs. Gaps show up when sales asks for comparison pages you never built while the blog is full of adjacent education.
In 2026, also mark pages that are pure commodity explainers with no experience layer. Those pages may still have impressions while earning fewer clicks because of AI answers. Decide whether to deepen them or stop investing. Context sits in what still earns the click.
Decide with clear rules: keep, refresh, consolidate, remove
Write decision rules before emotions enter. Example: refresh if demand evidence and business fit are high but accuracy risk is high. Consolidate if two URLs share intent and neither is a clear winner. Remove if business fit is low, demand is negligible, and accuracy risk is high. Keep pristine pages that still match strategy even if traffic is modest, when sales uses them.
- Sort by business fit first, not by traffic ego.
- Apply decision rules in batches of similar templates.
- Record the redirect target for every consolidation or removal.
- Create refresh briefs only for URLs that will get real SME time.
- Publish the decision log where future marketers can find it.
Redirects and politics
Removals create attachment. Someone wrote the post. Someone linked it in a sales email years ago. Use data and the decision rules. When a URL must die, 301 to the best remaining intent match. Update internal links. Tell sales which URL replaces the old one. Silence after deletion is how broken links linger in decks.
Turn the audit into a two-quarter backlog
Group actions into waves. Wave one: accuracy risks on high-demand pages. Wave two: consolidations that clarify clusters. Wave three: net-new evaluation pages the audit proved missing. Wave four: selective deepening of educational hubs that still support the argument. This backlog should feed production systems the same way new ideas do, including audit and refresh engagements when capacity is short.
Protect capacity for refreshes. Teams that only celebrate new URLs recreate the mess within a year. A healthy ratio depends on library age, but many B2B teams should spend a meaningful share of each quarter on maintenance. Pair this with the operating model in content ops for small B2B teams.
Share a before-and-after map with leadership: how many URLs entered the audit, how many were removed or consolidated, which clusters became clearer, and which net-new gaps were confirmed. Numbers make the work legible. They also prevent the false conclusion that "we need to publish more" when the real finding was that half the library had no strategic job.
If engineering support is limited, sequence technical tasks carefully. Redirects and template updates may need a sprint slot. Do not approve a giant prune list if nobody can implement redirects for two months. Pair editorial decisions with a realistic shipping plan, or the audit becomes a PDF that nobody executes.
Make auditing a quarterly habit
A one-time heroic audit fades. Install a lighter quarterly pass: check top fifty landing pages for decay, review new publish versus prune counts, and re-scan for cannibalisation in priority clusters. After major product launches, run a claims sweep. After messaging changes, update evaluation pages first.
An unaudited library is a liability that compounds. Every quarter you delay decisions, you pay in diluted authority, confused buyers, and SME hours spent on pages that should not exist.
If publishing volume created the mess, read why publishing more stopped working before you refill the calendar. The audit is how you stop funding the habit.