Knowledge base / For owners
How to read your Search Console, in four numbers and three reports.
It is free, it is your own data, and it is the only honest source for how Google actually sees your site. Twenty minutes a month is enough.
Key takeaways
- What are the four numbers? Clicks, how often someone clicked through to your site. Impressions, how often a link to you was seen. CTR, clicks divided by impressions. Average position, where your link sat on average.
- Which three reports matter monthly? Performance for what you rank and get clicked for, Pages under Indexing for what Google has and has not indexed, and Core Web Vitals for whether real visitors find the site fast enough.
- What is the most common misreading? Treating average position as a rank. It is averaged across every search where you appeared, so a page can show position 8 without ever having been eighth for anything.
- Does Search Console show everything? No, and this one matters. Google withholds rare queries, so the query list never adds up to your total clicks. Trust the page-level data more than the query list for totals.
What do the four numbers mean?
Google defines them plainly in its own performance report documentation, and the definitions are worth taking literally rather than approximately.
| Metric | What it counts | What it tells you |
|---|---|---|
| Clicks | How often someone clicked a link from Google to your site. | The only number that represents a person arriving. |
| Impressions | How often a link to your site was seen in results. | Your reach. Rising impressions with flat clicks means you are appearing for things you do not answer. |
| CTR | Clicks divided by impressions. | Whether your title and description earn the click once you are visible. |
| Average position | The average position of your link, where 1 is topmost. | A trend line, not a rank. Useful compared against itself over time, misleading quoted as a number. |
Which three reports are worth twenty minutes?
Performance. Set the range to the last three months and compare it with the previous period. Look at the Pages tab before the Queries tab, because pages are complete and queries are not. Then open Queries and read them as questions rather than keywords: which ones are people asking, and does a page of yours answer each one in its first paragraph?
Indexing, then Pages. This tells you what Google has actually indexed and what it has refused. Every reason listed there is worth understanding once. A page that is not indexed cannot rank, cannot be quoted, and will not appear in any AI answer, which makes this the highest-value report on the list and the least opened.
Core Web Vitals. Field data from real visitors, split into good, needs improvement and poor. Treat it as pass or fail per URL group. Do not chase the lab score from other tools, which measures a simulated load on a simulated connection and moves every time you run it.
What are the two traps?
Averaging the average. Average position is already an average across every impression. Comparing one month's average to another's tells you something. Quoting it as "we rank 8th" does not, and acting on it leads to optimising pages that were never the problem.
Assuming the query list is complete. It is not, by design. Google withholds rare queries, and the share withheld differs enormously between sites. The practical rule: use queries to understand what people want, use pages and the date view for anything you are going to quote as a number.
A third, smaller trap: the last three days are always incomplete. A Monday morning panic about Friday's numbers is usually a reporting delay.
The one filter worth saving
In Performance, filter queries to those containing a question word: who, what, why, how, when, does, can. That list is the set of questions real people are asking Google and finding you for.
It is the best free source of material for a Master FAQ, because every entry is a question with demonstrated volume and an existing impression count. Take the top twenty, check whether each one is answered in the first paragraph under a matching heading somewhere on your site, and you have both a content plan and an AI-readiness checklist drawn from your own data rather than from a keyword tool.
Search Console questions
Because Google withholds queries that are too rare, for privacy, and the gap can be large. On sites we manage it ranges from roughly a sixth to around half of clicks missing from the query view. Use the Pages tab when you need real totals and the Queries tab to understand intent.
No, and this causes more wrong conclusions than anything else in the tool. It is averaged across every impression, including searches from other countries, on other devices, for queries you did not target. A page showing 8.4 may rank 3 for its main term and 40 for twenty others.
They measure different things. A rank tracker runs one search from one location on one device. Search Console averages every real search where you appeared. Neither is wrong. Quoting one as the other is.
Usually you started appearing for broader searches you do not satisfy, or something above your result started answering the question. Filter to your most important queries and check whether the pattern holds there. If it is only on informational terms, see the AI answer and the map pack.
Twenty minutes a month for most businesses, plus whenever something changes on the site. Daily checking produces noise and panic, because the data is incomplete for the last few days and normal variation looks like a trend at that range.
Not to start. Search Console is the only source that tells you what Google actually did with your site, and it is free. Paid tools add competitor data and tracking from fixed locations, which is useful later and is not a substitute for your own data.
Pair your own data with an outside read of your pages.
Search Console tells you how Google sees your site. The free AI Visibility Score tells you what an answer engine can read on the pages themselves.
Get your AI Visibility Score