Search reports are evidence about a defined set of observations. They are not a complete record of every customer's journey or a permanent ranking for the business. Useful decisions depend on understanding the property, filters and measurement limits.
This guide explains how to compare Search Console results, investigate changes and avoid drawing conclusions that the data cannot support.
Record the scope before interpreting results
Confirm that the authorised property represents the website being reviewed. Record the report, date range, search type, page selection, country, device and any query filters. Keep those choices consistent when comparing periods.
Check whether the data is preliminary or sufficiently settled for the decision. Recent values and available reporting features can change. Save the scope and observation date with the report, so another person can understand the comparison later.
Give each metric its proper meaning
Clicks, impressions, click-through rate and average position answer different questions. Click-through rate relates clicks to impressions within the selected scope. Average position summarises observations; it is not one guaranteed ranking seen by every person.
Investigate changes by relevant page groups, devices and markets. A shift in the mix of queries can change an average without every page moving in the same way. Interpret visibility alongside useful visits and enquiries rather than treating position alone as a business outcome.
Respect aggregation and privacy limits
Charts and tables can use different aggregation rules, while visible rows are limited. Some queries are anonymised for privacy. Chart totals can include anonymised queries when no query filter is applied, so summing the visible query rows need not reproduce the headline total.
A query filter changes the available evidence. Do not describe an exported list as every phrase people used, or fill missing values with invented estimates. Explain the report's scope and use the available wording as a useful sample.
Use filters deliberately
Test regular-expression filters on examples that should and should not match. Search Console uses RE2 syntax and partial matching by default, so the pattern's boundaries matter. Save the pattern with the analysis rather than relying on memory.
Where a branded-query filter is available, inspect examples and the treatment of spelling variants. Automated classification can be imperfect and availability is limited. A brand comparison needs an explicit definition before it can guide investment.
Check page and country interpretation
Search Console generally attributes performance to the canonical URL selected by Google, with exceptions for particular reports or features. A duplicate address receiving a visitor can therefore differ from the address used in the report. Review canonical evidence before calling the discrepancy an error.
A country dimension describes where the search originated, not the language of the query or page. For an international website, review country, language, page relationships and device separately. Preserve the same dimensions in any period comparison.
Investigate drops as a sequence of checks
Start with the affected scope: one page, a section, one device group or the whole property. Check public availability, redirects, crawl and indexing evidence, recent content changes and measurement configuration. Record what each observation establishes.
Then consider demand, seasonality, competitors and search-system changes. A traffic decline occurring during an update does not establish its cause. Compare timing and affected pages, and avoid changing several unrelated things before the evidence has been understood.
Keep historical search updates in context
Google's June 2025 core update began on 30 June 2025. The August 2025 spam update began on 26 August and completed on 22 September 2025. These historical events can help label a comparison period, but their dates alone do not explain a particular website's performance.
For a new investigation, establish the relevant event and observation period afresh. Review useful content and technical issues identified on the actual site. Avoid treating a recovery checklist as a promise that one edit will restore a previous position.
Checks for a repeatable search report
- Record property, report, dates and filters.
- Explain the meaning of each metric.
- State aggregation, privacy and row limitations.
- Review canonical attribution and country interpretation.
- Separate confirmed findings from possible explanations.
- Assign an action, owner and follow-up observation.
Our crawl and indexing guide supports technical investigation. The canonical URL guide explains preferred addresses and migration checks.
