# How to analyze Google Search Console data: five questions that find the work

> A repeatable way to analyze Search Console: what changed, where the demand is, what underperforms, where pages compete and what is missing, with the filter for each.

Source: https://anotherseoguru.com/en/resources/search-console-analysis

**Analyzing Google Search Console means turning the Performance report into a short list of changes worth making.** Most people stop at the chart. A useful analysis asks five questions in order: what changed, where the demand is, what underperforms for its position, where your own pages compete, and what people search for that you have no good page for. Each question maps to one filter or comparison in Search Console, and each answer ends in a concrete fix.

## The four metrics, read correctly

| Metric | What it really measures | Common misreading |
| --- | --- | --- |
| Impressions | How often a result of yours was shown, whether or not it was scrolled into view | “People saw us 50,000 times” |
| Clicks | Clicks from Google Search to your site | Expecting it to match sessions in Analytics |
| CTR | Clicks ÷ impressions | Comparing CTR across positions |
| Average position | Your top position per search, averaged and weighted by impressions | Treating it as a rank on one results page |

Two consequences matter for analysis. First, impressions and clicks move for reasons other than your rankings: demand, seasonality and results-page layout. Second, totals at the top of the report include anonymized queries that the query table hides, so the rows never add up to the total.

## 1. What changed, and why?

Turn on **Compare** in the date filter: the last 28 days against the previous 28, or the last 3 months against the same 3 months a year ago if your business is seasonal. On the Pages tab, sort by click difference. A page that lost clicks lost them for one of three reasons, and the columns tell you which.

| Pattern | Most likely cause | Next step |
| --- | --- | --- |
| Position fell | Rankings slipped | Check what now outranks you; refresh the page |
| Impressions fell, position held | Demand fell | Check last year's same period before touching the page |
| CTR fell, position held | Something above you takes the click | Look at the live results page; rewrite the snippet |

If the whole site dropped at once, follow the [organic traffic drop diagnosis](https://anotherseoguru.com/en/resources/organic-traffic-drop) instead.

## 2. Where is the demand you already have?

Sort Queries by impressions and remove brand queries. The top of that list is your real non-brand demand. Queries ranking between 4 and 20 are the cheapest clicks you can win, because Google already finds your page relevant. [SEO quick wins](https://anotherseoguru.com/en/resources/seo-quick-wins) explains how to push them.

## 3. What underperforms for its position?

Compare each top-10 query's CTR with the CTR its position usually earns. A query far below the curve is losing clicks to a better snippet or a results-page feature. The curve is in [CTR by position](https://anotherseoguru.com/en/resources/ctr-by-position) and the fixes are in [high impressions, low CTR](https://anotherseoguru.com/en/resources/high-impressions-low-ctr).

## 4. Where do your own pages compete?

Click an important query and open the Pages tab. More than one URL with meaningful impressions means Google is unsure which page answers it. That split usually costs positions. [Keyword cannibalization](https://anotherseoguru.com/en/resources/keyword-cannibalization) shows how to confirm it and which of four fixes to use.

## 5. What is missing?

Look for queries with impressions where the ranking page is only loosely related: your pricing page ranking 30th for a how-to question, or the homepage ranking for a specific use case. That is demand Google already associates with you, without a page built for it. Those are your first new articles.

## Filters that make the analysis faster

- **Exclude brand:** Query → Custom (regex) → Doesn't match regex. Patterns are in [branded vs non-branded keywords](https://anotherseoguru.com/en/resources/branded-vs-non-branded-keywords).
- **Questions only:** a regex such as `^(how|what|why|when|which|can|does|is)\b` finds informational queries.
- **One section:** Page → URLs containing `/blog/` to analyze content separately from product pages.
- **Mobile versus desktop:** the Devices tab, compared, shows whether a drop is device-specific.

> **Note:** The Search Console interface shows at most 1,000 rows per table. For larger sites, use the Search Console API, the Bulk Data Export to BigQuery, or a tool that pages through the API. AnotherSEOGuru App loads up to 100,000 rows per period.

## Make it a routine

The five questions take about an hour once you know the filters. Doing them weekly beats doing them perfectly once a quarter. Our [one-hour weekly Search Console loop](https://anotherseoguru.com/en/blog/gsc-weekly-ops-playbook) is a ready-made agenda, and the [SEO forecasting](https://anotherseoguru.com/en/resources/seo-forecasting) method turns what you find into a ranked plan.

## In this guide

- [High impressions, low CTR](https://anotherseoguru.com/en/resources/high-impressions-low-ctr): Why pages are seen but not clicked, and the fixes in order.
- [Branded vs non-branded keywords](https://anotherseoguru.com/en/resources/branded-vs-non-branded-keywords): How to split them in Search Console with regex.
- [Keyword cannibalization](https://anotherseoguru.com/en/resources/keyword-cannibalization): Confirm it in Search Console and pick the right fix.

**How often should I analyze Search Console?**

A short weekly check for changes and a deeper monthly pass through all five questions works for most sites. Daily checks only make sense during a migration, a launch or a Google core update.

**Why don't Search Console clicks match Google Analytics sessions?**

They count different things. Search Console counts clicks on Google results; Analytics counts sessions that it could record, which depends on consent, tracking and how sessions are attributed. Expect the trends to agree, not the numbers.

**What date range should I use?**

Three months for analysis of current rankings, with a comparison period. Up to 16 months when you need to see seasonality, which is the most Search Console keeps.

**Is average position reliable?**

It is reliable as a trend and for comparing queries, not as an exact rank. It averages your top position across every search where you appeared, so one query can rank differently by country, device and day.