Open Google Search Console for almost any client and the first two numbers you'll see are impressions and average position. People screenshot them. They put them in slide decks. They celebrate when the line goes up.
Here's my honest take: impressions and average position are two of the least useful metrics in the whole tool. Not because they're fake, they're measured correctly, but because on their own they don't tell you whether a single real person found what they needed, or did anything about it.
I still look at them. I just don't trust them alone anymore, and after this article, I hope you won't either.
An impression means your URL showed up somewhere on a results page for some query. That's it. It doesn't mean anyone saw it, scrolled to it, or read a word of it.
You can rack up thousands of impressions from queries you'll never rank well for, sitting at position 38 where nobody scrolls. You can get impressions from queries that have nothing to do with what your page actually offers. A spike in impressions after a Google update can mean you're now showing up for more irrelevant searches, not that anything got better.
Impressions are a signal that Google's index knows you exist for a topic. That's genuinely useful for finding content gaps. It is not evidence of visibility, and it is definitely not evidence of business value.
Averages hide distributions, and average position hides one of the worst distributions you'll find in web analytics.
Say your average position for a page is 4.2. Sounds solid, right? But that number could be built from one keyword sitting at position 1 with zero search volume, and another important keyword sitting at position 12 that nobody in your target audience ever scrolls to reach. Blend them together and you get a tidy 4.2 that describes neither reality.
Position also swings by device, by location, by whether an AI Overview or a shopping carousel pushed everything else down the page that day. None of that nuance survives being flattened into one number.
What actually helps is looking at the distribution: what share of your priority keywords sit in the top 3, top 10, top 20. That's a bucket, not an average, and it's far closer to how real users behave. It's also exactly why a rank-distribution target (like moving 30 to 50% of your priority keywords into the top 10 over roughly two years) is a far more honest goal than chasing a single average number down.
This is the part that actually changes the stakes. Ranking well used to mean getting clicks. That link is breaking.
Ahrefs re-ran roughly 300,000 keywords against December 2025 Search Console data and measured what happens to click-through rate when an AI Overview appears on the results page. The drop is brutal: position 1 loses about 58% of its expected clicks, position 2 loses 51%, position 3 loses 46%, position 5 loses 33%, and even position 10 loses 19%.
Read that again. Position 1, the spot everyone optimizes their whole life for, can lose more than half its clicks once an AI Overview sits above it. Your average position metric has no way to tell you this is happening. It just shows a nice number while your actual traffic quietly erodes.
This is what people mean when they say search is going zero-click. The answer gets synthesized right there on the results page, and a lot of users never click through to anyone at all.
There's a second reason position is losing meaning, and it's more structural. AI-driven search (Google's AI Mode and AI Overviews, and similar patterns in ChatGPT, Gemini, Copilot and Grok) increasingly uses something called query fan-out: it takes one question you typed and quietly breaks it into anywhere from roughly 5 to 20-plus related sub-queries, running them all at once to build one synthesized answer.
So when someone searches "best sneakers for walking," the system might also be checking "best sneakers for men," "sneakers for walking on a trail," "best slip-on sneakers," and more, all behind the scenes. Your content isn't competing for one keyword position anymore. It's competing across a whole cluster of sub-queries you'll never see individually in Search Console.
That's part of why I keep saying your organic SEO strategy and your AI-visibility strategy aren't two separate things. A page that only half-answers one narrow keyword loses on both fronts at once.
None of this means ignore Search Console. It means stop treating impressions and average position as the finish line, and start treating them as one diagnostic layer among several.
Here's roughly how I prioritize things now, from what actually proves growth down to what's just an early-warning signal.
Before you even get to strategy, it's worth checking that your measurement setup isn't quietly lying to you in a different way.
In GA4, open Acquisition > Traffic Acquisition and filter for the Unassigned channel. As a rough rule, anything above 10% is worth investigating, and anywhere near 30% is a clear problem, not a rounding error.
It's also worth building a dedicated channel group for AI referrers (ChatGPT, Gemini, Perplexity, Claude, Grok and so on) so that traffic doesn't get lost inside "Direct" or "Unassigned." Right now that traffic is genuinely small (fractions of a percent for most sites), but small doesn't mean irrelevant, it means you want to see the trend clearly instead of guessing.
Impressions and average position aren't useless. They're just incomplete, and treating them as headline metrics gives you a false sense of progress while your real numbers, actual visitors, actual conversions, quietly tell a different story.
The web is shifting toward zero-click answers and fragmented sub-queries. That makes it more important, not less, to measure what happens after someone (or something) sees your page, rather than just whether it showed up.
If you're rebuilding your own reporting around this, start small: pull rank distribution instead of the average, check your Unassigned traffic, and watch conversion rate alongside traffic growth, not instead of it.
Thanks for reading. If you're wrestling with your own Search Console numbers or want a second pair of eyes on your tracking setup, reach out, I'm always happy to dig through a dashboard with someone.