Skip to content
Learn

Measuring search with Search Console and analytics

In this lesson

Section titled “In this lesson”
  • Read the four visibility metrics in Search Console: impressions, clicks, CTR and average position.
  • Explain why the rows in Search Console do not add up to the totals, and when to use the API or the BigQuery export.
  • Split traffic into branded and non-branded queries, and check query and topic coverage.
  • Choose the right tool for a question: Search Console, a rank tracker, Google Analytics, Microsoft Clarity or Looker Studio.

Days 01 to 07 taught you how to earn visibility. This lesson teaches you how to measure it. Day 08b will add the new AI measurements: AI citations, prompt tracking and SEO tests.

The measurement stack

Section titled “The measurement stack”

No one tool shows the whole journey. Each tool sees one part of it:

Part of the journey Question Main tool
On the results page Did people see and click my result? Google Search Console
On a fixed list of keywords Where do I rank today, compared to competitors? A rank tracker, for example AccuRanker
On the site What did visitors do after the click? Google Analytics
On one page Where did visitors click, scroll or get stuck? Microsoft Clarity
For the team How do I show all of this in one report? Looker Studio

Two tools matter most. Search Console is the only source of Google’s own data about your results. Google Analytics shows whether search visitors do something useful. Both are free. The other tools add detail, but they depend on these two, or they guess what these two measure.

Google Search Console

Section titled “Google Search Console”

Google Search Console is Google’s free tool for site owners. You prove that you own the site, and then Google shows you data about it. You met parts of it on earlier days: the page indexing report on day 03 and manual actions on day 02.

For measurement, the main report is the Performance report. Google says that it shows “important metrics about how your site performs in Google Search results”. By default it shows the last three months. Search Console keeps data for 16 months.

The report has two parts:

  • Metrics: clicks, impressions, CTR and average position.
  • Dimensions: the way the table groups the data. The dimensions are queries, pages, countries, devices, search appearance and dates.

You can also choose a search type: web, image, video or news. The web type now has two parts. “Web: text-based” is for typed queries. “Web: multimodal” is for searches with an image, for example Google Lens or Circle to Search.

Visibility metrics: impressions, clicks and CTR

Section titled “Visibility metrics: impressions, clicks and CTR”

Visibility metrics are the numbers that tell you how visible a site is in search. Google defines the basic three like this:

  • An impression is counted when a user “has seen (or potentially seen) a link to your site”. A link on the current results page counts, even if the user did not scroll to it. In a carousel, the item must usually scroll into view.
  • A click is a click from a Google result to your site.
  • CTR (click-through rate) is clicks divided by impressions.

Google gives all data to the canonical URL of a page. You learned about canonical URLs on day 03. So if a page has two URLs, Search Console shows one row for it.

AI features count too. A click on a link in an AI Overview or in AI Mode counts as a click. An AI Overview takes one position, and all of its links get that position. Since 31 August 2026, all sites also have separate generative AI reports, with impressions in AI Overviews, AI Mode and Discover. Day 08b covers these reports.

A low CTR is not always a problem. A result can have a low CTR because the answer is already on the results page, or because it is low on the page. Compare a page with itself over time, and compare pages at similar positions. If a page has many impressions, a good position and a low CTR, look at its title and snippet. You learned how to improve them on day 04b.

Section titled “Position and share of search”

Average position is the topmost position of your site in a result, averaged over all impressions. Google gives this example. Query 1 shows your site at positions 2, 4 and 6, so it counts as 2. Query 2 shows it at 3, 5 and 9, so it counts as 3. The average is (2 + 3) / 2 = 2.5.

Google warns that position “can be misleading if you don’t understand the subtleties”. Three points matter most:

  1. Position counts only when there is an impression. A result on page 3 has no position if no one opens page 3.
  2. One element takes one position, even with many links. A knowledge panel can have position 6 but still be very visible.
  3. Your own search shows only your own result. Results change with time, place, device and search history.

Bots can change these numbers too. Several SEO firms reported that in September 2025, Google stopped supporting the &num=100 URL parameter. Rank trackers used it to load 100 results in one request. After the change, many sites saw fewer impressions and a better average position, with no change in clicks. Google did not publish numbers about this, so the size of the effect is not confirmed. The diagram shows the mechanism with example data.

Example data. Move the second slider to add impressions from a rank tracker at position 70. Notice that clicks do not change, but CTR and average position get worse.

Share of search is a different kind of position. It does not measure rank. It measures demand. Les Binet presented it at the IPA EffWorks conference on 14 October 2020. It is the searches for your brand, divided by the searches for all brands in your category. The data comes from Google Trends, which is free and goes back to 2004.

Binet found that share of search correlated with market share in all three categories that he tested: cars, energy and mobile phones. It moved first, so it can predict market share. The lead time can be long, “up to a year for cars”. But Binet also said that it “is not a perfect predictor”: price affects sales, and “not all search activity is positive”. Use it as a signal of brand demand, not as an exact forecast.

Branded vs non-branded queries

Section titled “Branded vs non-branded queries”

A branded query contains your brand name, a misspelling of it, or a product that only you sell. A non-branded query is every other query. Google explains why the split matters. Branded queries “typically lead to higher-ranking pages” and higher CTR. Non-branded queries show how new users find you.

So report the two groups separately. If you mix them, a strong brand can hide weak SEO. A brand campaign can raise clicks, but non-branded clicks show whether your content wins new people.

Search Console has a branded queries filter for this. Google announced it on 20 November 2025, and on 11 March 2026 it was available to all eligible sites. It has these properties:

  • An AI-assisted system decides which queries are branded. It finds the brand in all languages, typos, and unique products. It can make mistakes.
  • It works only for top-level properties with enough queries. A subdomain or a folder property does not get it.
  • It has no effect on ranking.

If you cannot use the filter, use a regular expression in the query filter. For example, acme|akme|acme shop matches the brand and one typo.

Query and topic coverage

Section titled “Query and topic coverage”

Query coverage is how many of the relevant queries in your field show your site. Topic coverage asks the same question for whole topics. On day 04 you planned topic clusters. Coverage checks whether the clusters work.

A simple method:

  1. List the topics of your topic clusters.
  2. In the Performance report, filter the queries for each topic.
  3. Count the queries that have impressions, and note their positions.
  4. A topic with few queries is a gap. A topic with many impressions at positions 8 to 20 is an opportunity.

Search Console helps on large sites. In October 2025 Google added Query groups to the Insights report. AI puts queries with the same intent in one group, for example “guacamole dip recipe” and “how to make guacamole dip”. The groups show top, trending-up and trending-down topics. They are only for sites with many queries.

Data limits, the API and the BigQuery export

Section titled “Data limits, the API and the BigQuery export”

The Performance report does not show all data. There are two limits.

The first limit is privacy. Google leaves out anonymized queries: queries that “aren’t issued by more than a few dozen users over a two-to-three month period”. The table never shows them. The chart totals include them, but only when you do not filter by query.

The second limit is size. The table shows at most 1,000 rows. Search Console also keeps only the top rows for a large site, not all rows. Google says that this affects only “a few very large websites”.

The diagram uses Google’s own example. Try the filters.

Query filter

Example data from Google. Choose a query filter, then show the hidden row. Notice that the rows add up to 450 clicks but the chart says 550, and that a filter removes the hidden clicks from the total.

When 1,000 rows are not enough, there are two ways out:

  • The Search Console API lets a program ask for the data. One request returns up to 25,000 rows. The API gives up to 50,000 rows a day for each site and search type. It also has quotas, for example 1,200 queries a minute for each site.
  • The bulk data export sends the data to BigQuery, Google Cloud’s data warehouse, each day. Google launched it on 21 February 2023. It has all the data except anonymized queries, with no daily row limit. Only an owner of the property can set it up. BigQuery is a Google Cloud service, so check its pricing first.

The export has three tables. searchdata_site_impression groups data by property. searchdata_url_impression groups it by URL. ExportLog records which days were saved. Google says that small and medium sites already get all their data in the interface, the API or Looker Studio. The export is mostly for large sites.

Rank tracking and AccuRanker

Section titled “Rank tracking and AccuRanker”

Rank tracking checks where your pages rank for a fixed list of keywords, on a schedule. It answers questions that Search Console cannot answer:

  • Where do you rank for a keyword with no impressions yet?
  • Where do your competitors rank?
  • Which SERP features appear for the keyword?

Rank trackers get their data by sending automated queries to Google. Google’s spam policies call this machine-generated traffic: “scraping results for rank-checking purposes” without permission violates the policies and Google’s Terms of Service. The tools still exist, and the policy is aimed at the tool, not at your site. But know where the data comes from. A tracker also shows one search from one place, not the average of all users. So treat it as a sample, and trust Search Console for real clicks.

AccuRanker is a rank tracker for agencies and large sites. It has been on the market since 2013. On 18 October 2026, its prices started at $224 a month for 2,000 keywords with daily updates. It connects to Search Console and Google Analytics. It now also tracks AI answers, with a product called AccuLLM. Day 08b covers that kind of tracking.

Traffic analysis: Google Analytics

Section titled “Traffic analysis: Google Analytics”

Traffic analysis studies what visitors do after they arrive: where they come from, which pages they see, and whether they act. Search Console stops at the click. Analytics starts there.

Google Analytics (GA4) is Google’s free analytics platform. You add a small JavaScript tag to each page. The tag sends events, such as a page view or a click, to Google. You mark the important events, such as a purchase or a sign-up, as key events.

Three points matter for SEO:

  • Channels. GA4 puts each session in a channel. The Organic Search channel is visits from “non-ad links in organic-search results, including Google’s AI Overviews and AI Mode”.
  • Engagement. An engaged session lasts longer than 10 seconds, has a key event, or has two or more page views. Bounce rate is the share of sessions that are not engaged.
  • The link to Search Console. You can link a Search Console property to one GA4 web data stream. Then GA4 shows queries next to landing pages and their engagement.

Numbers in Analytics and Search Console never match exactly. Search Console counts clicks. Analytics counts sessions, and only when the tag runs and the visitor allows it.

Microsoft Clarity

Section titled “Microsoft Clarity”

Microsoft Clarity is a free tool that shows how people use a page. Microsoft says that it is “a free service forever”, with no traffic limits and no sampling. It has two main views:

  • Heatmaps show where people click and how far they scroll.
  • Session recordings replay the visit of one person, with the mouse moves, clicks and scrolls.

Clarity also flags signs of frustration. A rage click is many fast clicks in one area. A dead click is a click that gets no response. A quick back is a visit to a page and a fast return. Clarity keeps the data for up to 30 days. Some features, such as recordings, need the visitor’s consent to cookies. Use Clarity when a page gets search traffic but no key events. A recording often shows why.

Looker Studio

Section titled “Looker Studio”

Looker Studio is Google’s free tool for dashboards and reports. Its old name was Google Data Studio. Google renamed it on 11 October 2022 and said that it will stay free. A paid version, Looker Studio Pro, adds team and enterprise features.

Looker Studio has connectors for Search Console and Google Analytics. So you can put clicks, positions and key events in one report, and share it with a link. The Search Console connector has the same limits as the API: up to 50,000 rows a day and no anonymized queries.

Key terms

Section titled “Key terms”
  • Impression: one time that a user saw, or could see, a link to your site in Google.
  • CTR (click-through rate): clicks divided by impressions.
  • Average position: the topmost position of your site in a result, averaged over all impressions.
  • Visibility metrics: the numbers that describe how visible a site is in search, such as impressions, clicks, CTR and position.
  • Share of search: the searches for one brand, divided by the searches for all brands in its category.
  • Branded query: a query with your brand name, a misspelling of it, or a product that only you sell.
  • Query coverage: how many of the relevant queries in your field show your site.
  • Anonymized query: a rare query that Search Console leaves out of its tables to protect privacy.
  • Bulk data export: a daily export of Search Console data to BigQuery, without the daily row limit.
  • Rank tracking: checking the rank of a fixed list of keywords on a schedule with a tool.
  • Machine-generated traffic: automated queries to Google, such as scraping for rank checks.
  • Key event: an event in Google Analytics that you mark as important, such as a purchase.
  • Engaged session: a GA4 session that lasts over 10 seconds, has a key event, or has two or more page views.
  • Session recording: a replay of one visit, with mouse moves, clicks and scrolls.

Exercise

Section titled “Exercise”

Exercise · about 15 min

Measure branded demand and visibility

Do part A if you have a site in Search Console. Do part B in all cases. You need only a browser.

Part A: your own site (8 minutes)

  1. Open the Performance report for the last three months. Turn on all four metrics.
  2. Add a query filter with “Custom (regex)”. Write your brand name and one typo, for example acme|akme. Note the clicks.
  3. Change the filter to “Doesn’t match regex”. Note the clicks.
  4. Add the two numbers. Compare the sum with the total without a filter.
  5. Remove the filter. Sort the queries by impressions. Find one query with a position from 8 to 20 and a low CTR.

Part B: share of search (7 minutes)

  1. Open Google Trends.
  2. Compare three brands in one category, for example three phone makers or three supermarkets in your country.
  3. Set the time to “Past 5 years”.
  4. For the last month, add the three values. Divide each brand’s value by the sum. This is its share of search.
  5. Write one sentence: which brand gained share over five years?
Show a possible solution

Your numbers depend on your site and your choice of brands. A typical result:

  • Part A, steps 2 to 4: branded 1,200 clicks and non-branded 2,900 clicks give 4,100. The total without a filter is 4,600. The 500 missing clicks are anonymized queries, which a query filter always removes.
  • Part A, step 5: a query at position 11 with 3,000 impressions and 0.4% CTR is an opportunity. The page is almost on page one. Improve the content for the query, and check the title link.
  • Part B: Google Trends shows normalized interest on a scale from 0 to 100, not search counts. Values of 60, 30 and 10 give shares of 60%, 30% and 10%. If one brand went from 20% to 35% in five years, the trend suggests that its market share can grow too. Binet’s research says that this is a signal, not a sure forecast.

Check yourself

Section titled “Check yourself”
You add up the clicks of all rows in the Queries table. The sum is 3,100, but the chart total is 3,600. What is the most likely cause?
In one month, impressions rose by 40%, clicks did not change, and average position got worse. Which explanation fits best?
A shop gets 70% of its search clicks from branded queries. Non-branded clicks fell by 20% this year, but total clicks rose. What should the SEO report say?
A page gets good search traffic but almost no sign-ups. Which tool helps you see why visitors do not sign up?
When is the Search Console bulk data export to BigQuery worth its setup?

Sources

Section titled “Sources”

Next

Section titled “Next”

Day 08b measures visibility in AI answers: LLM referral traffic, citation rate, prompt tracking tools, and SEO and GEO tests.