Search performance
Where are the changes in your site's search performance?
Clicks, impressions, CTR, and average position must be read together. First determine the time range, then judge changes.
View my search performance Read-only Google connection · $9 / monthTurn a change into a checkable question
Start with one affected page and a reproducible window, not a site-wide ranking explanation.
On a narrow screen, scroll the table sideways to read the verification column.
| What you observe | What to do | What to retain or verify |
|---|---|---|
| Clicks down, impressions steady | Compare the same page/query, country and device; inspect position and the actual result text. | Keep the filters and both equal windows; a different query mix needs a separate explanation. |
| A blank report or failed request | Check permission, filters and response status before considering a content edit. | Record whether the request succeeded and what was returned; missing data remains unknown. |
Work through the distinction
Teaching example: 20/1,000 = 2% and 5/100 = 5%. The combined CTR is 25/1,100 = 2.27%. An unweighted average would change the question.
First, see what you'll get.
This is an interactive sample. Actual results come from sites you authorize.
Four metrics—don't look at just one.
Clicks
How much search traffic was gained
Does not equal orders or all visitsImpressions
How many times it was shown
Does not equal unique usersCTR
The ratio of impressions converted to clicks
There is no universal passing thresholdAverage position
A summary metric of result positions
Not a fixed rank for every keywordBefore using it, understand these details.
How it works, data boundaries, and applicable conditions.
What problem does this page solve
Judge whether the selected site's search performance has changed, rather than deciding to change the site based on a single ranking number. First choose a GSC property you have access to, then fix the country and the complete date window.
What you'll get
Clicks, impressions, CTR, average position, and trends for the current window, plus a summary comparison with the previous equal-length window. Missing prior records or API failures are explained separately; missing dates are not treated as true zeros.
How to confirm the next step
Record the window, country and metrics, then identify a concrete page. Save exact-page joint query/device evidence in SEO operations. For year-over-year seasonality and other native views, verify in GSC; keep scope differences explicit.
Four metrics answer different questions
Clicks tell you how much traffic was gained; impressions show that results were displayed; CTR is the ratio of clicks to impressions; average position must be read together with query, device, and country.
Choose a consistent data window
Currently supports 7, 28, and 90 days and country filtering, reading final Web search data. To reduce the impact of incomplete data, the window avoids the latest three days.
Locate first, then act
Changes in the composition of a country or query can shift the site-wide average. Do not judge that a piece of content has regressed based only on overall ranking.
How to start an analysis
First confirm whether the property covers the URLs you want to check, then choose 28 days and the target country. Record the window end date, and read the summary, pages, queries, and countries separately. After finding an anomaly, return to native GSC to filter specific pages, rather than forcibly merging two separate lists.
A reproducible CTR example
Teaching data: Page A has 20 clicks and 1,000 impressions, CTR is 2%; Page B has 5 clicks and 100 impressions, CTR is 5%. Combined is 25/1,100, about 2.27%, not 3.5%. This is a calculation example, not a user site result.
What changes are worth addressing
Prioritize persistent changes that affect important pages. A worse overall average position may come from newly added low-ranking impressions; changes on the same page in the same market are more suitable for further review. Metrics indicate direction, but without page evidence the cause cannot be determined.
Completeness and failure states
Currently reads limited top rows, not the full query database. Permission errors, API failures, missing data, true zeros, and low samples must be distinguished. When failures occur, troubleshoot the connection first; do not replace real results with simulated data.
Frequently asked questions
Can I see real-time visits?
No. Search Console is search performance data, not real-time visit statistics.
Does increased impressions mean business growth?
Not necessarily. You also need to look at related queries, clicks, page reception, and your own conversion data.
Continue to the next step.
Work through the underlying checks
Start with your own site.
Verify the data and analysis first, then decide your next step.