Content audits tend to produce one of two outcomes: a spreadsheet that nobody acts on, or a consolidation programme that deletes the wrong things. The first happens because the audit doesn't produce prioritised decisions, only data. The second happens because the decision framework is wrong - usually traffic-based, which treats low-traffic pages as low-value pages. In B2B, that's often backwards.
Here is the decision tree I use. It won't produce a perfectly optimised site in one pass. Nothing does. What it will do is give you a consistent, defensible framework for making decisions at scale, and a prioritisation logic that ensures you work on the highest-leverage pages first.
Content audit frameworks that start with traffic are building on shaky ground. Traffic is a lagging indicator of past decisions and past search demand: it tells you what has happened, not what should happen. Before you pull any data, answer these three questions for your specific situation:
- What does the site need to do in the next 12 months? A site that needs to capture new category demand has different consolidation priorities than a site that needs to convert more of its existing traffic. Know the goal before you build the framework.
- Which pages are commercially important regardless of traffic? In B2B SaaS, a product page for a feature that is actively being sold to enterprise accounts is commercially critical even if it has 200 visits a month. Never delete or consolidate commercially critical pages based on traffic alone.
- Which pages support the sales process directly? Many B2B teams don't know this. Ask sales. There are almost always three to five pages that get shared in deal cycles that don't appear important in the analytics.
With those answers documented, the data analysis can begin. Without them, you're making decisions that look rational in a spreadsheet and turn out to be expensive in practice.
For each page or cluster of pages, work through these questions in order. Stop at the first decisive answer.
Question 1: Is this page commercially critical or sales-supporting?
If yes: Keep. Improve if needed but do not consolidate or remove regardless of traffic. These pages are not a content problem; they're a conversion optimisation problem.
If no: Continue to Question 2.
Question 2: Does this page currently rank for any query with meaningful search volume?
If yes, and the ranking is relevant: Keep. A page that ranks is doing work. Improving it is almost always better than removing it. The question is whether the traffic it earns converts appropriately. If not, that's a post-click problem, not a content problem.
If yes, but for irrelevant queries: Update or consolidate. A page ranking for terms you no longer want to be associated with can be redirected to a page that better reflects your current positioning, once a better candidate exists.
If no: Continue to Question 3.
Question 3: Is there a better, live page on the same topic?
If yes: Merge. Redirect the weaker page to the stronger one. Be specific with the redirect: the destination should be the page most likely to satisfy the same intent, not just the homepage.
If no, but there could be: Kill and redirect to the closest relevant page. Don't create a new consolidation target just to give the redirect somewhere to go. That defers the problem rather than solving it.
If no, and the topic is worth covering: Kill the thin version and plan a proper replacement. A thin page on a topic you want to own is often worse than no page.
A thin page on a topic you want to own is often worse than no page, because it occupies the ranking position with content that doesn't deserve it and suppresses the chance to produce something that does.
Question 4: Does this page have any backlinks or external citations?
If yes, meaningful links: Keep or merge carefully. Backlinks are worth preserving. If you're consolidating, use a 301 redirect and the link equity transfers. Check the linking context: sometimes the page was linked because it covered a specific angle that the consolidation target doesn't cover. Make sure the destination page earns the link.
If no: Kill or consolidate without concern for link preservation. A page with no external signals and no ranking is a pure overhead cost.
Question 5: Is the content accurate, current, and non-embarrassing?
If no to any of these: Fix before keeping. An inaccurate page that ranks is worse than no page - it creates a bad first impression for buyers who will never come back. Update or kill.
Content audits have always been primarily about search performance. In 2026, that means accounting for AI search as well as traditional organic. A page that ranks in position 3 for a traditional query but never appears in ChatGPT or Perplexity responses for the same query is doing partial work.
Adding the AI citation check
When running a content audit today, I add a lightweight AI citation check: for the top 20-30 most important pages by traffic or commercial value, test whether those pages' topics produce citations to the site in the main answer engines. Run the same query in ChatGPT, Perplexity, and Google AI Overviews. Note whether your brand appears and whether your content is cited.
The key question is whether any gap is a content quality issue or a source-worthiness issue. The answer determines whether the right intervention is rewriting the page or building off-site presence - these are completely different programmes.
The merge vs. replace decision in AI search
Traditional audit logic says: if you have two thin pages on the same topic, merge them. In AI search, the calculus is more nuanced. A merged page that's genuinely comprehensive and authored may perform better in AI citations than two thin ones. But if neither page has the named authorship, specific claims, and source-worthiness signals that drive citations, merging them produces one medium-thin page instead of two thin ones - the citation problem is unchanged.
The content portfolio decisions that follow (particularly the choice between updating an existing page vs. creating a new, more citable version) are covered in the GEO roadmap I'd build if starting today.