How to Run a Content Audit for SEO Recovery

A content audit compares each URL with its historical performance, business value, search intent, and link equity, then assigns one action: keep, update, consolidate, or delete. Its useful output is a prioritized recovery queue with evidence, owners, redirect rules, and a date for reviewing the result, rather than another spreadsheet that nobody acts on.

What does a content audit decide?

A content inventory records what exists. A content audit decides what happens next. The inventory gives you URLs, titles, clicks, impressions, conversions, internal links, and backlinks. The audit turns those inputs into an action and an owner for every page that needs attention.

Anchor the audit to one business goal before opening a crawler. A traffic-recovery audit prioritizes pages that have lost qualified demand. A migration audit protects useful URLs and redirect equity. A cannibalization audit identifies pages competing for the same intent. The same page can receive a different verdict under each goal.

Start with the pages most likely to affect the goal. Auditing every legal page, campaign archive, and utility URL with the same rubric wastes review time and hides the recovery queue inside a giant inventory.

When should you run a content audit?

Audit cadence should follow publishing velocity and risk. A team publishing daily may need a quarterly review of flagged pages. A smaller B2B site can usually run a focused review twice a year, with a full inventory pass annually.

Run an unscheduled audit when a sustained traffic decline, site migration, major positioning change, or visible keyword cannibalization creates a specific decision problem. After a Google core update, wait until the rollout finishes and use the documented core-update assessment process before changing pages based on a few volatile days.

Give substantive updates enough time to produce a measurable result. Rewriting the same page every few weeks destroys the comparison period and leaves the team unable to separate seasonality from the effect of the change.

How do you build a useful content inventory?

Crawl the site, then join search, analytics, backlink, and CMS data by canonical URL. Screaming Frog can combine crawl data with URL-level metrics through its analytics integrations. Add CMS publish and update dates so pages blocked from crawling or missing internal links do not disappear from the audit.

Use one row per canonical URL with these fields:

• URL: The canonical address after redirects resolve.
• Page type: Article, glossary entry, product page, documentation, campaign page, or utility page.
• Primary intent: The reader question or job the page serves.
• Search performance: Clicks, impressions, CTR, and average position for the selected comparison periods.
• Business value: Key events, assisted pipeline, sales use, support use, or another documented purpose.
• Authority: Referring domains, valuable backlinks, and internal links.
• Freshness: Original publish date, last substantive update, and time-sensitive claims.
• Proposed action: Keep, update, consolidate, or delete.
• Owner and review date: Who will complete the action and when the result will be checked.

Exclude pages that have not had enough time to collect signals. The appropriate window depends on crawl frequency, demand, and site authority, but a new page should not be judged against a three-year-old page after only a few weeks.

A strong AI search content strategy also gives each page a defined role. That context prevents the audit from deleting useful pages simply because organic traffic is not their primary job.

How should you score each URL?

Use relative performance before universal traffic cutoffs. Compare the latest complete period with the previous equivalent period and the same period a year earlier when seasonality matters. Google explains how clicks, impressions, CTR, and position behave in the Search Console performance report.

Flag a page for human review when at least two meaningful signals deteriorate. Useful signals include falling impressions, falling clicks, worsening position, lost conversions, declining referring domains, or a mismatch between the page and the current SERP intent. Set the percentage or volume thresholds from your own baseline, then document them before reviewing individual URLs.

Absolute floors still help prioritize large inventories, but they should create candidate queues rather than automatic verdicts. One hundred clicks can be poor for a category hub and excellent for a niche page that supports high-value sales conversations.

Which decay pattern explains the decline?

A decline needs a diagnosis before it gets an action. Four common content-decay patterns produce different fixes:

• Ranking decay: Impressions, clicks, and position decline together. Review quality, competing pages, links, and intent alignment.
• Zero-click capture: Impressions and position hold while clicks fall. The SERP may now answer the query without a visit.
• Intent drift: The SERP favors a different format or reader job. Updating prose alone will not repair a format mismatch.
• Demand decay: Impressions fall while position remains stable. The market is searching less, so rewriting cannot recreate lost demand.

Use current SERP evidence and a repeatable AI keyword research workflow to distinguish intent drift from demand decay. The diagnosis determines whether to update the page, reposition it, consolidate it, or leave it alone.

How should qualitative review change the score?

Run the audit in two passes. Metrics create the candidate queues. A human then reviews accuracy, intent, business value, brand fit, readability, and whether the page still deserves to exist.

This protects low-volume pages that support sales, customer education, or a critical product workflow. It also catches high-traffic pages that remain visible while giving outdated or inaccurate advice.

How do you choose keep, update, consolidate, or delete?

Assign one verdict after the quantitative and qualitative reviews are complete.

• Keep: The page serves its intended job, remains accurate, and earns meaningful traffic, links, conversions, or operational value. Record the next review date instead of changing it for freshness alone.
• Update: Demand and intent still exist, but the page has a repairable weakness. Common fixes include a clearer answer, current evidence, stronger examples, better internal links, or a format aligned with the present SERP.
• Consolidate: Multiple pages serve the same intent or split links and rankings. Merge the strongest unique material into one surviving URL, then redirect the weaker URLs.
• Delete: The page is redundant, inaccurate beyond repair, tied to an expired program, or trivial to every documented audience. It also has no meaningful links, conversions, or historical value worth preserving.

Deletion should be the smallest queue. Google recommends removing content only when it cannot be improved and cautions against changing dates or deleting pages merely to make a site appear fresh in its current ranking-systems guidance.

Write the reason beside the verdict. “Update because impressions are stable, CTR fell, and the title no longer matches the current SERP” is auditable. “Needs work” is not.

How does AI search change the audit?

Add an AI-citability check to pages that answer questions or explain categories. This is a separate quality signal, not a replacement for traffic and conversion data.

A large analysis of Google AI Mode citations found that cited passages often answered early and stood alone. Use that finding as a practical review prompt rather than a universal scoring formula:

• Does the section answer its heading immediately?
• Can the answer stand alone without the preceding paragraph?
• Does the page distinguish evidence from opinion?
• Are key claims supported by current sources?

A page can retain strong organic traffic while remaining difficult for answer engines to extract. Put those pages in the update queue when AI visibility matters to the page’s job. The guide to optimizing content for AI search provides the deeper rewrite workflow.

How do you update underperforming content?

Use the diagnosis to choose the smallest change likely to solve the problem. A low CTR with stable top-page visibility calls for a title and snippet review. Ranking decay may require better evidence, fuller intent coverage, and stronger internal links. Intent drift may require a new structure or a different page type.

Review these elements in order:

• The title and description against the current SERP.
• The opening answer and whether it fulfills the query quickly.
• Time-sensitive claims, examples, screenshots, and statistics.
• Missing sections that matter to the current reader intent.
• Internal links into and out of the page.
• Structured data that must still match the visible content.
• The CTA and whether its destination still exists.

Do not update both pages in a cannibalizing pair. Consolidate them around the stronger URL so the site has one clear answer for the intent.

How do you consolidate or delete pages safely?

Check backlinks, internal links, conversions, and historical purpose before removing a URL. When a closely relevant replacement exists, redirect the retired page to it. When no replacement exists, return a 404 or 410. Google documents both outcomes in its crawl-error guidance.

Never redirect an unrelated page to the homepage or a broad category page. Google can treat irrelevant redirects as soft 404s, which hides the deletion rather than preserving value.

For consolidation, select the surviving URL by intent fit, qualified traffic, backlink quality, and business value. Merge useful sections from the weaker pages, redirect them to the survivor, and update internal links to point directly to the final URL. Verify the redirects and indexation after release.

Which tools belong in the workflow?

Each tool should answer a defined audit question:

• Screaming Frog: Which URLs exist, which are indexable, and where do internal links, canonicals, redirects, or near-duplicates need attention?
• Google Search Console: Which pages are losing clicks, impressions, CTR, or position, and for which queries?
• GA4: Which landing pages produce engaged sessions and key events?
• Ahrefs or another backlink index: Which pages hold external link equity that must be protected?
• CMS export: Which pages are new, stale, unpublished, orphaned, or blocked from the crawl?

Do not let tool availability define the verdict. The audit needs a business-value field even when that evidence comes from sales notes or support workflows rather than an analytics platform.

How do you maintain the audit as a recurring system?

Turn the spreadsheet into an operating queue. Every flagged URL needs an owner, approved verdict, completion date, and next review date. Track the result after implementation so future thresholds reflect what actually recovered traffic or conversions.

Run focused reviews on flagged pages between full audits. Large sites can also benefit from removing duplicate and low-value crawl paths because Google identifies URL-set quality as a controllable crawl factor. Small sites should prioritize content value and demand over crawl-budget work.

Teams scaling output through programmatic SEO should add template-level monitoring. When a shared template loses performance or accuracy, review the whole page family before editing hundreds of URLs individually.

For a weekly operating lesson on content systems and AI search visibility, subscribe to The Messy Middle newsletter. It gives you practical criteria to keep the audit rubric current as search surfaces and citation behavior change.