How to measure search visibility without overreading the numbers
An article's average search position may improve, but that alone does not tell you whether more readers reached the page. A page can appear more often while receiving few clicks. Search visibility rarely fits into one number.
The Google Search Console performance report gives authorised site owners impressions, clicks, click-through rate and average position. Read them together with the page, query, country and period clearly identified.
Why separate impressions from clicks?
An impression concerns an appearance in search results; the counting rules vary by result type. A click records a user's move from a Google result to the site. Click-through rate, or CTR, is clicks divided by impressions.
In an illustrative example, 20 clicks from 1,000 impressions produce a 2% CTR. Five clicks from 100 impressions produce 5%, even though the second period brings fewer clicks. These are hypothetical numbers, not RadioTEDU data. They show why the rate and its underlying counts belong together.
Is average position what every user sees?
No. Average position summarizes positions across relevant impressions; it does not mean everyone saw the same result in the same place. Google's measurement documentation explains rules for different result types.
When comparing a page before and after a change, use comparable time windows and consistent filters where possible. Seasonality and demand can also affect results. A difference by itself does not prove the edit caused it.
What does Google's generative AI report show?
The generative AI performance report helps properties with sufficient data inspect link impressions in supported Google Search features including AI Overviews and AI Mode. An impression is not a click, and the report does not measure another provider's answers. A missing report can have several explanations, including insufficient impressions.
Listen while exploring this story
Tune into campus radio with non-stop independent music and culture.
Google says these impressions are also included in the general Search performance report's web search type. Adding both totals as separate audiences can therefore mislead. Report dates use Pacific time, which matters for daily comparisons in Türkiye. Some unavailable values can export as zero, so preserve the original interface state. Google also marks preliminary recent data with a dotted line; it may change as collection continues, so an early movement should not be treated as a settled result.
Imagine one generated answer displays links to two different RadioTEDU pages. Google may count one property-level impression in the chart, while a page-grouped table can show two separate page rows. The report documentation explains that chart and table totals can differ because of aggregation. This is a hypothetical example, not measured RadioTEDU performance.
The documented report dimensions are page, country, date and device; a query or answer-text dimension is not listed. Because most page data is assigned to a canonical URL, the absence of a separate row for a redirected or alternate address does not necessarily mean the content received no impressions. Do not infer which prompt produced a particular sentence from this report. Google also excludes Search Labs experiments from it. A separate observation using an exact prompt answers a different question and is not a substitute for the report.
Does a drop always come from the content?
When a chart falls unexpectedly, note the filters and periods before explaining the change. The same Google report documentation describes separate filters for text-based and multimodal web searches. A difference between two periods with different filters should not be attributed to an edit without further evidence.
The measurement system can also have a temporary issue. Google’s data anomalies log says a logging error reduced reported impressions in the Search generative AI report for 13–17 August 2026. Its 21 August update says the missing data was restored. This does not establish that RadioTEDU experienced a drop; it shows why a team comparing those dates should check both known issues and whether they were corrected.
Observing other AI answers
For another service, record the exact prompt, language, date, known country and session conditions, visible sources and number of repetitions. A small sample does not represent every user's experience or establish market share.
No RadioTEDU performance figures are asserted without access to its Search Console property. Measurement helps explain which content reaches which readers within the limits of the evidence.
