How to Run a Content Audit That Ends in Priorities

The GSC report for last month is in. Clicks dropped on two pages, impressions grew on three others, and one page that looked stable has been quietly losing ground for six weeks. Where do you start?

A structured reading of GSC traffic data, built around what the numbers are doing rather than what the site looks like from the outside. The output is one to four priorities you can act on in the next four weeks.

What a Content Audit Is in This System

A content audit, as I use the term, is a structured reading of GSC traffic data that produces a ranked list of priorities for the upcoming period. The input is movement: which pages and query clusters gained or lost ground, and at what scale. The output is a decision about where to focus next.

The process runs in three phases: observe what the data shows, interpret what it means, then define priorities based on the interpretation. Each phase has a distinct output. Running them out of order corrupts both; I’ll explain exactly how as we go.

A crawl-based audit starts with a URL inventory. This one starts with what the traffic data is telling you right now.

What You Need Before You Start

Three things need to be in place.

GSC access to the property, with at least one full comparable period of data. A Looker Studio report built around two consecutive periods of the same length. Comparing last month against the same month a year ago measures seasonality; the audit is reading momentum, and for that the periods have to match.

The report separates two data layers. Pages show where traffic is landing; queries show what people were searching for when it arrived. Each tells you something the other doesn’t, and the analysis uses both.

One more thing: the delta column. A calculated field in the Export Table that shows change in impressions per row rather than the raw number. Without it you’re reading individual data points. With it, you’re reading movement, which is the foundation of everything that follows.

Look Before You Interpret

The first job with the data is to record what moved. Four metrics, in order: impressions, clicks, CTR, average position. Note direction and scale. Causes come later.

Raw clicks surface your biggest pages. The delta surfaces the ones that moved.

A page losing clicks gets flagged as a problem before you notice its impressions are growing. That’s a completely different signal, and the observation step is what keeps it in the picture.

Once the metrics are read, group by topical cluster or page type. Patterns appear at cluster level. A URL-by-URL read at this stage produces 80 data points rather than three readable movements.

Limit each metric trend to three dominant patterns. Opposing movements go in a separate row; so do anomalies: pages with zero data, non-indexed pages, sudden spikes in a small set of URLs.

All of that goes into a movement table:

The movement table records what changed across metrics before any cause is assigned — interpretation is a separate step.

Write the Traffic Evaluation

With the movement table complete, the next step is a traffic evaluation: a short narrative covering what the data showed across the reporting period.

This is the What in the What–Why–What Next structure the full report follows. It describes the picture. The Why — causes and interpretation — comes next; the What Next is the priority list.

Structure: one sentence on overall movement, impressions first. Then one paragraph per dominant pattern, one topical cluster or page group at a time, naming specific pages and queries with the percentage changes in. One paragraph on opposing trends. One on anomalies and known states.

What the evaluation deliberately excludes is anything that explains the numbers or suggests what to do about them. When evaluation and interpretation run together, the output becomes a mixed signal; the reader can’t tell which claims are confirmed and which are guesses.

Interpret the Signals

Interpretation is where observation becomes understanding. Three categories, mutually exclusive.

Certainties are cause-effect relationships confirmed by data or a known change. “Clicks increased 24% following a title update applied mid-May.” If you can verify the connection, it goes here.

Hypotheses are plausible interpretations. A likely cause you haven’t verified. In practice: “this one did it” is a Certainty; “I have a suspect” is a Hypothesis.

Uncertainties are movements with no available explanation. Something visible in the data, nothing that accounts for it yet. (This is the category most teams skip — or quietly convert into Hypotheses.)

Certainties drive this month’s priorities. Hypotheses need one more check before they do. Uncertainties go back into the data.

Keeping the three separate is what makes the output actionable.

What makes interpretation systematic rather than intuitive is the decision table. The same drop in clicks signals something different depending on what impressions, CTR, and average position are doing simultaneously:

Each combination calls for a different response, sometimes the opposite. That’s what the table makes legible.

Define the Priorities

All of the analysis comes down to one to four named priorities. Short by design.

Start with the top ten falling pages by impressions. Check rankings for each, then cross-check the associated query clusters to confirm the trend isn’t an artifact of one unusual week. First priority candidates.

Then the top ten growing pages, same check. Pages growing in both impressions and rankings sit slightly lower in the list. The falling ones take precedence. Cross-reference both lists for topical overlap. A query cluster falling on one page while a topically similar page gains is a cannibalization signal.

Last filter is business relevance. A landing page in the conversion path and a thin informational article from three years ago can show identical movement signals. They don’t get identical priority. The Client Brief is what tells you what this business needs to move and what’s closest to revenue right now.

From Audit to Content Plan

The audit ends with a ranked list. The content plan is where those priorities become defined work.

Each priority becomes a row. The fields that carry over: the primary keyword, assigned from the query clusters identified during observation; the growing keywords GSC showed gaining impressions in the period; the missing keywords (queries accumulating impressions the current content doesn’t address); cannibalization flags from the cross-reference step.

Strategy notes travel too. If the audit flagged that a cluster needs a new supporting article rather than a direct page edit, that note carries into the Strategy Notes column in the content plan.

Nothing in the handoff requires re-analysis. The audit is structured so its output feeds the planning input directly, without a translation step.

The Looker Studio template, the Export Table macro, the content plan structure, and the QA checklists for both are part of the Content Accelerator. If you’re regularly publishing content, definitely check it out.