Google Search Console metrics explained

Google Search Console reports your search performance through four core metrics: clicks, impressions, click-through rate (CTR), and position. A click counts a visit from Google. An impression counts an appearance. CTR is clicks divided by impressions. Position is your average ranking. Each one is simple to define and easy to misread, because each hides a limit Google never surfaces in the chart.

This guide gives you Google's own definition of all four, then the caveat behind each: why an impression is not attention, why your clicks and impressions never add up, why average position flatters you, and why the old CTR benchmarks broke in 2026.

TL;DR:

• Clicks and impressions are counts, CTR is clicks ÷ impressions, and position is your average topmost ranking. These are Google's own definitions, all four.
• An impression does not mean anyone read your result. It counts when your link appears, even if the user never scrolls down to it.
• Your query clicks never add up to your page clicks, because Google hides rare queries. Take totals from the page-level view, not the query table.
• Average position is a single blended number standing in for many. It hides the per-query spread, so read position per query, not per page.
• The old "CTR by position" benchmarks are broken in 2026: AI Overviews cut clicks to the top result sharply (Ahrefs measured 58%, Pew found 8% of visits clicked vs 15% without).

What do clicks, impressions, CTR, and position mean in Search Console?

All four are defined by Google itself, and the definitions are short. A click is a visit from Google to your site. An impression is an appearance of your link. CTR is the ratio between the two. Position is where your link ranked, on average, with 1 at the top. The caveats are what the rest of this guide is about.

Here are Google's exact definitions, straight from the Search Console help on impressions, position, and clicks:

• Clicks: "How often someone clicked a link from Google to your site." The cleanest of the four. A click is a real visit, the only hard outcome on the list.
• Impressions: "How often someone saw a link to your site on Google." The word "saw" is doing a lot of work here, as the next section explains.
• Click-through rate (CTR): "The calculation of (clicks ÷ impressions)." A ratio, not a count. It moves when either number moves.
• Position: "A relative ranking of the position of your link on Google, where 1 is the topmost position, 2 is the next position, and so on."

Two of these are direct counts (clicks, impressions). One is arithmetic (CTR). One is an average (position). The trouble starts the moment you treat the average and the impression count as if they were as solid as the click count.

What actually counts as an impression in Search Console?

An impression counts when your link appears on the results page, whether or not the user scrolls it into view. In Google's own words, an impression means a user "has seen (or potentially seen)" your link. So impressions measure exposure, not attention. A page can collect thousands of impressions that nobody actually looked at.

The exact rule, from Google's documentation: "In general, an impression is counted whenever an item appears in the current page of results, whether or not the item is scrolled into view, as long as the user need not click to see more results." The exceptions are carousels, expandable FAQ results, and infinite-scroll feeds like Discover, where the item has to be scrolled or expanded into view to register.

The practical consequence: rising impressions are not proof of rising visibility to humans. A jump in impressions often means you started appearing, far down the page, for a flood of new queries you rank position 40 for. That is why impressions are only readable next to position. Impressions alone tell you Google showed you somewhere, not that it showed you anywhere good.

Why don't your clicks and impressions ever add up?

Because Search Console reports the same traffic two ways and filters one of them. Group by page and you keep every click. Group by query and Google strips out rare searches to protect privacy, so the query totals always come up short. The gap is not small: close to half of all clicks site-wide have no named query attached.

This is the single most common way people misread their own metrics: they add up the rows in the query table and call it a total. That number is always wrong, and always low, by exactly the hidden portion. The complete number lives in the page-level view, not the query view, and the missing searches are the ones Google deliberately drops, covered in anonymized queries in Search Console.

So the rule for these two counts is the same: trust clicks and impressions at the page level, where every click is kept. The query breakdown is a sample of what you rank for, not a tally of how much traffic you earned.

Why is average position misleading?

Because it is a single blended number standing in for many. Search Console takes your topmost position for each query, then averages it across every query you appeared for. A page that ranks position 2 for one search and position 45 for another does not "average to 23" in any way that helps you. The blend hides the spread that actually matters, including a query sitting just off page one that the average buries.

Google calculates position as "the topmost position occupied by a link to your property or page in search results, averaged across all queries in which your property appeared," per its definition. Two honest limits fall out of that one sentence:

• Topmost only. If your page shows up at position 3 and again at position 8 for the same search, Search Console records 3. The number you see is the friendliest one available, not the typical one.
• Blended across queries. One figure represents a page that might rank brilliantly for five queries and terribly for fifty. A page's blended average position can even improve while its most valuable queries get worse, simply because it started ranking for a batch of easy long-tail searches.

The fix is to read position per query, not per page. The page-level average is fine for a rough trend line, but the decision of what to fix lives in the per-query spread that the blend erases.

What is a good CTR, and why is the 2026 benchmark different?

There is no fixed "good" CTR anymore. The old rules of thumb (roughly a quarter to a third of clicks for position 1) came from a results page that no longer exists. In 2026, AI Overviews sit above the organic results and absorb the click. Your honest benchmark is your own CTR trend, segmented by whether an AI Overview shows, not a static table.

The data on the shift is consistent across independent sources. Ahrefs, comparing 300,000 keywords in December 2025, found that the presence of an AI Overview "now correlates with a 58% lower average clickthrough rate" for the top-ranking page, up from a 34.5% reduction they measured in April 2025 (Ahrefs, 2025). Pew Research Center, analyzing the browsing of 900 U.S. adults, found that users "clicked on a traditional search result link in 8% of all visits" when an AI summary appeared, against 15% when one did not (Pew Research Center, 2025). Treat both as correlation, not proof of cause, but the direction is not in dispute.

This is why a position-1 page can hold its ranking, hold its impressions, and still watch its CTR halve. The page did nothing wrong. An AI Overview started capping it. When impressions hold or rise while clicks fall, that is the signal to look at, and a static benchmark table built before 2024 will never show it to you. We watched this exact pattern on a live SaaS, a page holding position 7 at a 0.00% click-through rate, in 30 days of Search Console on a real SaaS. Telling a page you can fix with a better title from one an AI Overview is quietly capping is its own read, and it decides whether a rewrite is worth the afternoon.

How should you read these four metrics together?

Read each metric for what it honestly measures, and they stop fighting each other. Clicks are the only hard outcome. Impressions are exposure, not attention. CTR is a ratio that now bends under AI Overviews. Position is a blend that hides the per-query truth. Used for the right job, all four are useful. Used interchangeably, they mislead.

A short working checklist:

• Totals and trends: read from the page-level view. Never sum the query table into a total.
• Intent: read from the query view, as a sample of what you rank for, never as a traffic count. Watch for a page pulling the wrong search intent.
• Position: read per query, not from the blended page average.
• CTR: compare to your own past, aware of whether an AI Overview is present, not to a generic benchmark.
• History: Search Console keeps only a 16-month window and offers no delta export, so the totals you trust today vanish unless you snapshot them forward.

Reading your Search Console metrics without the headache

Holding four caveats in your head for every page is the tedious part: page view for the count, query view for intent, position per query, CTR against your own history. That discipline is most of what good Search Console reading is, and it is the first thing to slip under a deadline.

That is the work QueryScope is built to hold for you. It pulls your real Search Console data into your IDE, in the terminal instead of the dashboard, reports volume from the page level, reads position per query, and flags the case that matters most: clicks falling while impressions hold. If you want the one-line definition behind any metric, the Search Console glossary covers clicks, impressions, CTR, and average position, each with the caveat that comes with it.

Sources

• Google Search Console Help, What are impressions, position, and clicks? (definitions of all four metrics; impression counting and the "seen or potentially seen" rule; topmost-position averaging).
• Ahrefs, Update: AI Overviews Reduce Clicks by 58% (December 2025 analysis of 300,000 keywords; 34.5% in the April 2025 study).
• Pew Research Center, Do people click on links in Google AI summaries? (July 2025; browsing data from 900 U.S. adults, March 2025).