How to read clicks, impressions and CTR in Search Console

Use Search Console to understand which searches showed your pages and generated clicks. Compare similar groups, confirm the destination page and connect data to customer questions. Impressions, CTR and average position describe search; alone, they do not say how many customers bought.

By Lyntev. Updated on .
About 5 minutes to read.

From data to the next decisionChoose the question, define the calculation and act on the result.

Translation published on . Original Portuguese guide; source revision .

By the end, you can:

  • Read metrics with the correct denominator
  • Separate brand queries, pages and devices
  • Choose an intervention and record its result without false attribution

Start with a specific commercial question

Choose a decision the report can help clarify: does a service page gain visibility for relevant questions? Do customers find the pricing page when searching for costs? Has a new page started appearing? Avoid opening the dashboard just to find a number that rose. The question determines filters and the period you need to compare.

Record the property’s scope. A domain with several subdomains may combine products, documentation and the institutional website; its figures do not automatically belong to the page you want to improve. To study an area, filter the relevant page or prefix and confirm the address before drawing conclusions. Also keep search type consistent: Web, images and video may represent different journeys.

Read every metric with its unit

Clicks count actions in search results; impressions describe appearances under the report’s rules. CTR relates the two: clicks divided by impressions. Average position aggregates observations and does not represent a place reserved for your business. For detailed definitions and special situations, consult the report’s own documentation instead of assuming every format is counted alike.

Do not add positions or take a simple average of percentages for differently sized groups. If a group has few impressions, a small click change greatly affects the percentage. A high CTR for the company name can coexist with little new demand. Usefulness comes from combining query, page and intent.

Read every metric with its unit
MetricHelps you askDoes not demonstrate alone
ImpressionsAre we being shown for this topic?That everyone read the page
ClicksHow many visits came from the result?Enquiries received or sales
CTRWhat share of impressions generated a click?Quality of the whole support service
Average positionHow did presentation of this set change?The same position for every user

Compare equivalent periods and segments

Choose periods of comparable duration and check events changing demand: holidays, seasonality, campaigns, news or a changed offer. A holiday-related business should not automatically interpret a September decline against August as a technical failure. With enough history, comparing the same season in the previous year can add context.

Separate country and device when that answers the question. A page can work well on desktop and have a hard-to-read call to action on mobile. Avoid splitting data until groups contain only two or three observations. Keep filters used so you can repeat the analysis and distinguish a real change from changed report configuration.

Connect the search with the page appearing

Open a page’s query set and see whether it matches what that page promises to resolve. Then reverse the process: choose a query and observe associated pages. If Google shows a guide where you wanted a commercial page, this may reveal informational intent, unclear content or a page split needing review. Do not conclude two pages are a problem merely because they appear for related terms.

Some queries are hidden for privacy, and aggregations may have particularities. The visible table is therefore not a complete inventory of every possible search. Record what you can observe and keep customer questions as another source. A silent row does not prove nobody searches for that subject.

Make a small calculation before changing the title

Fictitious example: a workshop page receives one thousand impressions and thirty clicks in a period. CTR is 30 divided by 1000, or 3%. In another period, it receives two thousand impressions and forty clicks, corresponding to 2%. There are more clicks despite the lower percentage. Interpretation changes if the new impressions came from broader queries.

The team examines those queries before concluding the title got worse. It discovers the page started appearing for a task it does not perform and clarifies scope. These figures are only educational; they are not a benchmark for good sector CTR. The next test should preserve alignment between the search, result text and information actually delivered.

Connect the pattern to a verifiable intervention

If there are relevant impressions and few clicks, review title, description and offer alignment while considering position and competing result formats. If there are clicks but the page does not clarify conditions, work on content and contact options. If indexing is absent, start with technical diagnosis. Each pattern needs a hypothesis, not the same action applied to every page.

Make a clearly bounded change: explain an exclusion, add the process or improve a title’s alignment. Do not promise a price or service in the result that the page does not offer. Keep the previous version and date to relate the change to later observations without attributing every external event’s effect to it.

Track enquiries outside the search report

Search Console does not confirm a form reached the team or a contact became a sale. Use proportionate commercial records and journey tests to follow those stages. If using website statistics, respect consent and understand coverage may be partial. Metrics need names showing exactly what was counted.

A useful review may conclude the priority is not more traffic but fixing a form or clarifying the offer. Also record inconclusive results when volume is low. Not turning weak signals into victories avoids hasty changes and helps invest in the obstacle actually preventing customers from proceeding.

Keep an analysis another person can repeat

The worksheet should include question, filters, period, data and hypothesis. Include a report link or screenshot only when you can share it with authorised people; do not send account access. If exporting data, limit purpose and recipients to the analysis work.

Schedule a review suited to enquiry and search frequency. A week may reveal an obvious error yet be short for assessing an editorial change on a low-traffic website. The aim is to choose the next step with context and make explicit what the data still cannot establish.

Worksheet for your project

Analysis question:
Property and pages covered:
Search type, country and device:
Periods and relevant events:
Brand queries separated from others:
Clicks, impressions and calculated CTR:
Page shown for the chosen intent:
Hypothesis and alternative explanation:
Intervention, version and date:
How received enquiries are verified:
Review date and result, including inconclusive:

Sources and criteria

Content prepared with AI assistance and research using technical sources, with checks of the relevant examples and calculations. The references support the guide’s decisions; examples identified as fictitious let you try out the method.

Wellinton Oliveira

Written by

CEO of Lyntev

Technology with context, original analysis and sources you can consult.