Zero click search: how to measure marketing when nobody clicks

For twenty years, a search worked like a handshake. You published something useful, Google ranked it, and a person clicked through to read it. The click was the proof. It landed in your analytics, you counted it, and everyone agreed on what visibility meant. In 2026 that handshake has quietly come apart. Google now answers a large share of searches on the results page itself, most visibly through AI Overviews, and the reader gets what they came for without ever visiting you. This is zero click search, and it breaks the single number most businesses still trust: the click.

We are going to make an unusual argument for a marketing agency, and back it with our own numbers rather than a tidy client case study. Those numbers are not flattering. That is the point. If you can see honestly what impressions do and do not mean now, you can stop measuring the wrong thing, and start measuring what actually predicts new business.

What zero click search looks like in our own data

Here is our Google Search Console property for the 28 days ending 6 July 2026. We are showing it whole, because a rounded-off version would let us off the hook. Across the period we recorded 46,072 impressions and 49 clicks. Split by whether the query contained our name, the picture sharpens.

Read the two rows against each other. Non-branded search put us in front of people 41,523 times and produced six clicks, a click-through rate of 0.014%. Branded search, where someone typed our name, showed 31 times and produced seven clicks at 22.6%. Almost nobody who does not already know us clicks. Almost everybody who does, does. The impressions are real exposure. They are just not visits.

The trend line says the same thing louder. Against the prior 28 days, our impressions rose 86% while clicks rose only 53%, and average position stayed flat near 60. We became more visible and less clicked at the same time. If you are watching only clicks, that growth is invisible to you. If you are watching only impressions, you would think something wonderful happened. Neither number, on its own, is telling you the truth.

An honest word about these figures

We will not pretend the split is cleaner than it is. Google withholds rare queries from Search Console to protect user privacy, so the branded and non-branded breakdown does not cover everything. In our case it accounts for about 90% of impressions, but only 13 of the 49 clicks. That means the impression story is solid: 46,000 impressions and a rock-bottom non-branded click-through rate are conclusions we stand behind. The click split is directional. It points clearly in one direction, that branded intent converts and cold discovery does not, but we would be overclaiming if we treated 13 of 49 clicks as the whole picture. We are telling you this because an agency that hides the gaps in its own data has no business advising you on yours.

Why impressions stopped meaning visibility

None of this is peculiar to us. It is the search page changing shape underneath everyone. Forbes reported in May 2026 that AI Overviews were appearing in around 58% of Google searches, with click-through rates falling sharply where they show (Forbes). Treat that 58% as a direction rather than a fixed constant, because figures like it vary a lot by methodology, query set and measurement date. The direction, though, is not in doubt.

TIME made the same point from the reader's side, describing Google's AI overhaul as reshaping how people find information, and warning what that does to the publishers and businesses who depended on the click (TIME). There is a counter-current worth knowing: in May 2026 Google said it would put more links to websites back into AI Overviews (Ars Technica). That is genuinely good news, but notice it sharpens the argument rather than reversing it. If the answer carries citations, then being one of the cited sources is the whole game, and a citation without a click is still a win you now have to measure deliberately.

So here is the plain conclusion. An impression used to imply a fair chance of a visit. Today it mostly means your link was rendered on a page whose answer had already been written for the reader. Impressions are a weak exposure signal now, not a measure of visibility, and certainly not a measure of interest. That is uncomfortable, because impressions are the number that has been climbing on most dashboards. Climbing impressions can hide a business that is being read and quietly forgotten.

What to measure instead

If the click is no longer a reliable proxy for value, you need measures that survive a click-free journey. Four of them do the job. None is perfect alone, which is the point: you read them together, the way a doctor reads several vitals rather than one.

1. Share of voice in AI answers

This is the one that most directly replaces rankings. Write down the questions your customers actually ask, then prompt the major answer engines, ChatGPT, Google AI Mode, Perplexity and Copilot, on that fixed question set on a schedule. Record whether your business is named, and who is named instead of you. Do it monthly and you have a real trend: a citation rate that rises or falls, per engine, per question. It is manual and a little tedious, and it is the closest thing to a ranking report the AI-answer era gives you. Our AEO services are built around exactly this measurement, and our explainer on what an AEO agency does covers the work behind it.

2. Branded search volume

This is the truest proxy for the awareness those uncredited impressions create. When you are cited in an answer without a click, the effect shows up later as someone typing your name into Google. Our own branded row proves the mechanism: 22.6% of branded impressions clicked, against 0.014% of cold ones. So watch branded search volume in Search Console over quarters. If your AI visibility is doing its job, the tens of thousands of anonymous impressions should slowly feed a rising line of people searching for you by name. That line, not raw impressions, is the awareness scoreboard.

3. Referral traffic from AI assistants

When a click-free answer does send someone your way, it usually arrives as a referral from an AI assistant rather than from a Google results page. In GA4, watch for referrals from ChatGPT, Perplexity and Copilot, and check the newer AI Assistant channel grouping. We can vouch for it directly: a real Perplexity referral has landed under GA4's AI Assistant channel on our own property. The volumes are small today, which is precisely why you want to catch the trend early rather than wait for it to be obvious. Tag the links you control with UTMs so nothing slips through unattributed.

4. Conversions and assisted conversions, not sessions

Finally, stop grading yourself on sessions. In a zero click world the visitor who does arrive is often further along, having already read the answer, so a smaller number of visits can carry a larger share of intent. Measure enquiries, calls, form fills and sales, and pay attention to assisted conversions, where a channel touched the journey without being the final click. A drop in sessions with steady conversions is not a problem. It is the new normal, correctly measured.

A warning: your GA4 is probably undercounting conversions

This short section may be the most valuable thing on the page, because the mistake it describes is genuinely common and quietly expensive. Most GA4 properties undercount conversions for two reasons, and both are settings, not code.

First, lead events often fire perfectly but are never marked as Key Events in the GA4 Admin. The event is being collected; it just never shows up in your conversion reports, so every enquiry it represents is missing from the number your board looks at. Second, internal traffic from your own team is never filtered out, so your staff testing forms and browsing the site inflate sessions and drag down every conversion rate you calculate. Both are fixed in the GA4 Admin UI in minutes, once you know to look. Go and check yours today. If your conversions look too low or your rates look strange, this is the first place to look, before you blame a channel or an agency. Getting the meter right is the whole foundation of our analytics work, because every other decision rests on trusting the number.

Where we fit, and where you can start free

We are an Australian AI-focused agency, and measuring visibility in a zero click world is central to how we work rather than a bolt-on. That is why we were willing to open our own Search Console here: the honest read on our numbers is the same one we would give you on yours. If you would rather start without talking to anyone, our free tools let you check how you appear before you spend a cent, and there is more in this vein across the blog. When you are ready to fix the measurement itself, that is what our analytics service is for.