Research Guides
Compare Search Console Page Impressions With the Right Total
A Search Console chart can report fewer impressions than the sum of a Pages table without either view being wrong. Before calling the difference missing traffic, check whether the chart counts appearances for the property and the table counts appearances for individual URLs.
This worksheet compares counting rules, not the performance of a particular property. The example is fictional and does not require connecting an account or buying a tool.
1. Write down what an impression belongs to
Google's Search Console Performance report documentation explains that chart data is aggregated by property, while table aggregation depends on the selected dimension. In the property example Google describes, two results from the same site for one query count as one property impression. Data grouped by Pages is aggregated by page.
Do not read “impressions” as a complete measurement definition on its own. Before exporting, state whether the question is how often the property appeared or how often particular pages appeared. Use the first answer for property visibility and the second for comparing individual pages.
2. Work through a controlled mismatch
Imagine two eligible search result views for a fictional site. The first view shows distinct pages A and B, and the second shows page A only. Assume each listed result qualifies for the measurement and both URLs belong to the same property.
- Page A: two impressions.
- Page B: one impression.
- Sum of the page rows: three page impressions.
- Property view: two property impressions.
The difference is one in this toy dataset because the first view contributed two page impressions but one property impression. It does not represent one hidden query or a missing report row. In real exports, do not use a totals gap alone to infer how many times two pages appeared together: filters and omitted rows can change the comparison too.
3. Preserve the grouping with the export
The Search Analytics API documentation describes property aggregation and page aggregation by canonical URL. It says that when filtering or grouping by page, the request should use auto rather than forcing property aggregation. The response includes responseAggregationType; keep that actual returned value with the file.
Record the property, dates, search type, filters, dimensions, aggregation and export method. If someone later sends only a spreadsheet labeled “impressions,” they should be able to recover these choices without guessing from the column names.
4. Rule out accidental differences
Match the date window and filters before comparing two exports. Check whether both describe the same search type. Then record row limits, omissions or privacy filtering separately from the aggregation rule. If the files do not share the same conditions, correct that mismatch before interpreting the totals.
Use the page view to investigate a particular URL without demanding that its neighboring rows sum to the property total. Conversely, do not divide a page-row click count by a property-level impression total and label it that page's click-through rate.
5. Put the counting rule in the headline or note
A reproducible note for the fixture is: “Across two eligible search result views, the property appeared twice. Page A appeared twice and page B once, for three page impressions. The page-row sum is not a property total.” One sentence about the aggregation prevents readers from treating those numbers as a contradiction.
If the purpose is a before-and-after comparison, keep the counting unit fixed across both periods. This worksheet establishes what the compared numbers represent; it does not prove a ranking change, diagnose missing data by subtraction or identify what caused visibility to change.