Two thirds of Google searches now end without a click. What that means if you run a small business

Open Google Search Console on almost any small business site and look at the last twelve months. There is a good chance you will find the same shape: impressions climbing, clicks flat or sliding. More people are seeing you. Fewer are arriving.

Two thirds of Google searches now end without anyone clicking anything. SparkToro’s 2026 zero-click study puts it at 68%, up from about 60% in 2024. Roughly 230 of every 1,000 US Google searches reach the open web, against about 360 two years earlier. That is US panel data, but the machinery producing it runs on South African results pages too.

What does zero-click actually mean?#

Someone searches. They read the AI Overview, or the map pack, or the snippet lifted off somebody’s page, and their question is answered. They close the tab. Or they do not find what they wanted and they search again with different words.

Both count as zero-click, and together they are now about two thirds of everything. Of the third that does still click, a slice goes to paid ads and another slice goes to Google’s own properties, so the share landing on ordinary websites is smaller again.

Two caveats worth knowing. The panel behind these numbers covers desktop and mobile browsers, not the Google app, and plenty of phone searching happens inside the app where answers are even harder to leave, so the real figure is probably worse than 68%. And SparkToro is upfront that the year-on-year comparison draws on different data panels, which makes the 360 to 230 gap a direction of travel rather than an exact delta. Read every number here that way.

Is AI to blame for this?#

Not really, and this is the part most people get wrong.

Zero-click searches were around 45% back in 2016, when the only AI on a results page was the ranking algorithm. By 2019 they were at roughly half. By 2024, 60%. Google spent a decade learning to answer questions itself: featured snippets, weather cards, currency converters, opening hours, the map pack, the knowledge panel. Every one of those was a click it used to send and then stopped sending.

What AI did was stamp on the accelerator. Roughly eight percentage points in two years is the sharpest move in the series. But if your plan is to wait for the AI thing to blow over, you are waiting out something that started when Zuma was president.

Do these numbers apply in South Africa?#

The percentages are American. I am not going to pretend otherwise, and be sceptical of anyone who quotes them at you as though they were measured in Cape Town.

What does carry across is the mechanism. AI Overviews show up on South African results. So does the map pack, and for a local business that is the one that matters most. Search “plumber Durbanville” on your phone and count how far down the page you get before there is any reason at all to visit a website. Three businesses, star ratings, a call button, directions, opening hours. The click did not evaporate. It went to a Google Business Profile, and if yours is thin or neglected, it went to your competitor’s.

So for most of the small businesses I work with, zero-click is less about AI eating the internet and more about a specific, winnable surface that sits above the search results. Claiming and filling that profile properly is usually the highest-return hour anyone spends on their search presence.

Which searches still send a click?#

The ones closest to a transaction, which is the good news buried in all this.

SparkToro reckons branded searches are about 44% of Google’s volume, and those still click through reliably. Someone typing your company name is going to your site. Nothing in the results page can satisfy that intent on its own.

The other pattern in their data: sites that let a person finish a task hold on to their clicks. Book a table, get a quote, check whether you have the part in stock, see the price. Pages that merely explain a topic and then point somewhere else are exactly what a summary replaces.

Put those together and the shape of the loss is clear. What is dying is the explainer article at the top of the funnel, the one written to catch a search and hand the reader a call to action at the bottom. What is holding up is the part of your site where someone does business with you. If most of your traffic was the first kind, this hurts. If it was the second kind, you may barely feel it.

Where is the searching happening, if not Google?#

Still Google, mostly. Let us be honest about that before anyone reallocates a budget.

SparkToro and Datos looked at 41 sites where people run searches and found Google took 73.7% of US desktop searching in the last quarter of 2025. The other quarter is spread across Amazon, YouTube, Reddit, TikTok, marketplaces, app stores. And the LLM tools everyone is worried about are real but small: on SparkToro’s 2024 comparison Google was handling something like 373 times ChatGPT’s query volume.

The useful read is not “Google is finished”. It is that roughly one buyer in four starts somewhere you may not have thought of as a search engine at all. If you sell physical product, some of your searching happens on Takealot, not Google. If you sell to trades, a chunk happens inside Facebook groups. Worth knowing which one is yours before you spend anything.

So what do you actually do about it?#

Four things, roughly in the order I would do them.

Change what you count. Sessions were always a proxy. If impressions are up and enquiries are steady, nothing is broken, and a dashboard that only tracks traffic will tell you the opposite. Count enquiries, quotes, calls, bookings. Those are the numbers the business runs on.

Make your pages finish the job. If a page explains something and then asks the reader to go elsewhere to act, an AI summary does the explaining part better than you and the trip elsewhere never happens. Put the price, the availability, the booking form, the actual answer on the page.

Own the surface that replaced your click. For a local business that means the Google Business Profile: claimed, categorised properly, real photos, hours that are right, reviews arriving steadily. For everyone it means being quotable enough that AI answers name you, which I have written up separately in the GEO post.

Build something the platform cannot switch off. An email list, a WhatsApp list, a handful of customers who refer you. Unglamorous, and it is the only audience nobody else controls. Every business I know that shrugged off the last two years of search upheaval had one.

What I would not do#

I would not fire your SEO and go all in on AI optimisation. The channel sending most of your enquiries is still Google, and the one everyone is panicking about is still a small fraction of it. Do both.

I would not chase a bigger audience when the problem is a vaguer one. Traffic that was never going to buy has stopped arriving. Replacing it with more traffic that will never buy is just paying twice for the same nothing.

I would not write another article whose only job was to rank for a question. If it would not be worth publishing when nobody searched for it, it is not worth publishing now.

And I would not rebuild your website over this. The search results page changed. Your site is probably fine, or fine enough that the problem is somewhere cheaper to fix. If you want to know where to look instead, I listed the six problems I find on almost every small business site, and a redesign fixes none of them.

If you want to know which of those applies to you, send me the domain. I will look at what you actually rank for, where the clicks went, and tell you whether it is worth doing anything about. Half the time the honest answer is that your numbers are fine and the industry is having a panic attack.