SEO
How to Read a Google Search Console CSV Export
An export is a set of clues, not an action list. The useful work is connecting a query to the page that appeared, checking the reporting window and deciding whether the pattern is strong enough to investigate.
Published October 8, 2026 · 7 minute read
Know what each column describes
Clicks count visits from a search result. Impressions count appearances in Search Console’s reporting context. Click-through rate (CTR) is clicks divided by impressions. Average position is the average position of your topmost result, grouped by the selected report dimensions and period; it is not a fixed rank that every searcher sees.
These measures answer different questions. Impressions show visibility in the chosen report. Clicks show recorded visits from those results. CTR gives a ratio between the two. Position helps add context, but should not be read alone as a promise about what a person sees in a live search.
Start with the page and query dimensions
A Pages table shows which URLs received the reported activity. A Queries table shows the searches that Search Console makes available. Neither table by itself necessarily explains the relationship you need. Open the report in Search Console, apply the same date range and filters, then inspect the page/query relationship for a row you want to understand.
Some queries are anonymized and omitted from query data. Exported rows can also be limited by the selected report and export process. Therefore, totals across exported dimensions may not match the property totals shown in the interface.
Read patterns before rewriting snippets
Suppose a URL has impressions but few clicks. That is a reason to inspect which queries generated the impressions, not proof that the title is weak. If the page appears for a broad or mismatched set of queries, changing the snippet may not solve the underlying issue. Check the actual result, the page’s purpose and whether its content answers the task.
Likewise, a query with no clicks and a low average position is not automatically a title problem. First identify the associated page and compare its subject, coverage and internal links with the search task. A query in a stronger position with impressions but no clicks may deserve a closer result-preview check, but small samples and short periods can be noisy.
Use comparable periods carefully
When comparing exports, use matching report types, search type, filters, country, device and date windows. A comparison is less useful if one export includes a different mix of traffic or a different length of time. Look at both the absolute counts and the direction of change; a percentage change can look dramatic when it begins from a very small number.
- Keep Pages and Queries separate until you have checked their relationship.
- Compare like-for-like dates and filters.
- Prioritize repeated patterns over one isolated row.
- Open the page and live result before proposing a content change.
- Write down the evidence and the hypothesis separately.
Use a CSV review to narrow the next investigation
The Search Console interface remains the source for checking reports, filters and page/query details. A local export review can be useful when you want to scan several report tables, compare snapshots and identify rows worth opening. It should reduce the time spent finding a candidate, not replace the inspection that establishes what the data means.
The Search Console CSV Analyzer reads supported Search Console export ZIPs or CSV tables locally in the browser and surfaces evidence-based review prompts. It also creates a Markdown brief for a follow-up analysis. The current product page describes the included standalone HTML tool and read-me; check it for current compatibility and file details.
Sources
Google’s documentation describes the Performance report metrics and notes that anonymized queries are omitted from query data. Check the current documentation when interpreting changes to the interface or export format.