The Zero-Click Panic Is a Publisher Problem. You're Not a Publisher.

A zero-click search is a Google search that ends without a click to any website, ad or Google property, often because the results page already showed the answer. In June 2026 Rand Fishkin of SparkToro put the share at 68% of US Google searches. It is the kind of number that ends up in pitch decks telling every business owner to panic. The loss behind it is uneven. It falls hardest on publishers, the sites whose product is the content itself. If you sell a service, a product or an appointment, your problem is narrower and cheaper to fix.

How many Google searches end without a click?

On June 8, 2026, Rand Fishkin of SparkToro published an analysis of Similarweb clickstream data covering US Google searches from January through April 2026. His finding, in his own words:

"In the first four months of 2026, a whopping 68.01% of Google searches ended without a click."

Rand Fishkin, SparkToro, June 2026

For context, that figure was 60.45% in 2024 and roughly 45% a decade ago. Searches that end in any click at all, whether to a website, an ad, or one of Google's own properties, fell almost ten percentage points compared to 2024. That is a 23% relative decline in two years. Ahrefs data cited in the same study shows AI Overviews now appear on more than 20% of searches and cut clickthrough rates by nearly 60% when they show up.

Ignoring it costs you real money. If your business plan assumes search traffic grows the way it did in 2019, you are budgeting against a pool of clicks that shrinks every quarter. Publishers are already living this. Press Gazette reported that publisher traffic from Google fell by roughly a third in 2025. A Search Engine Land survey found news publishers forecasting search referral losses of 43% within three years, with a fifth of them bracing for losses above 75%.

Those losses and forecasts all come from publishers. The damage also varies by category among them. Blogging has stopped working in finance, healthcare, legal and consumer tech and still works in other categories. A business that sells something has a different relationship with the click. In 17 years in search, one result I point to is a London ADHD clinic I booked solid for three straight months, until it had to hire more specialists and outsource the overflow. An answer box cannot give anyone an appointment with a clinician.

Do zero-click searches hurt a business that sells a service?

Less than the headline suggests. The mistake starts when an owner sees the 68% statistic and assumes it applies equally to every website. From there the owner concludes search is dying for everyone and either cuts SEO entirely or panic-buys whatever AI search package the agency is selling this quarter.

Both reactions misread the data. Zero-click loss concentrates in informational content, the "what is" and "how to" articles that AI Overviews and chatbots now answer directly on the results page. Google no longer needs to send a user to a blog post to answer a definition question. It does still need to send someone to a locksmith, a dentist, a software checkout page, or a booking calendar, because Google cannot unlock a car or clean teeth.

Fishkin himself is blunt about what this means for the standard playbook:

"Much as it pains me to say, there's not much point (nor any hope) of fighting back by simply getting better at SEO."

Rand Fishkin, SparkToro, June 2026

In the same passage he writes that your SEO "still matters as much or more than ever before, it just won't earn you traffic the way it once did."

Rankings and visibility still count, sometimes more than before, because AI answers draw heavily from what already ranks. Informational rankings just no longer pay out in visits. In the same study, Fishkin names the categories where SEO still produces clicks and customers: branded searches, local businesses and high-intent transactional queries.

Google's AI Mode, the conversational search experience most people blame for the decline, accounted for just 0.34% of US searches in the January to April window. Most of the zero-click growth comes from older Google features, AI Overviews and answer boxes resolving queries on the page. The informational click was collapsing before the conversational AI wave fully arrived. Google did say at I/O 2026 that AI Mode passed one billion monthly users, with usage more than doubling every quarter. The squeeze gets tighter from here. The businesses still winning clicks today win them because of what their websites let a visitor do.

Which sites still win the clicks?

On April 9, 2026, Cyrus Shepard, a search practitioner best known for his years at Moz, published an analysis of more than 400 websites tracked over twelve months. He separated the sites that gained Google traffic from the sites that lost it. His summary of the project:

"Google decimated traffic to legions of websites. But traffic to other websites rose. What makes the winners different? I analyzed 400+ sites and the answers became obvious. Google rewards a different kind of site now..."

Cyrus Shepard, Zyppy Signal, 2026

His study found five features that correlated with traffic growth. In plain English, the site sells a real product or service and lets a visitor complete a task (book, buy, calculate, apply). It owns proprietary assets like original data or tools. It stays tightly focused on one topic area, and it has a recognizable brand.

Sites with four or five of those features came out ahead roughly 68% to 70% of the time. Sites with none of them won just 13.5% of the time. Among the winners, 70% sell their own product or service, against 34% of the losers. A full 83% of winning sites let visitors complete a task on the site, against 50% of losers. And 92.9% of winners own a hard-to-replicate asset such as original research, a proprietary tool or real photography of their actual work, against 57% of losers.

Put Shepard's findings next to Fishkin's and the picture sharpens. The 68% zero-click figure describes the fate of commodity informational content. The remaining 32% of clicks flow disproportionately to sites where clicking accomplishes something an AI answer cannot.

When AI does send a visitor, the visit tends to be worth more. Similarweb clickstream data from April and May 2026 indicates that visitors referred by ChatGPT convert at around 7.1%, just behind paid search at 7.8% and far above typical organic averages. Those figures come from clickstream modeling and your own analytics may show different numbers. The direction is consistent across multiple datasets: the visitor who clicks through from an AI answer arrives pre-sold. Fishkin made the same point about the broader shift, noting that traffic can fall precipitously even as revenue rises. Fewer visits, better visits.

What should I fix first if my search clicks are falling?

These fixes move money and attention. You will not write a line of code for any of them. Run them in this order.

Step 1: Audit your site for task completion. Open your website on your phone and try to actually buy, book, request a quote, or schedule something in under two minutes. If a visitor cannot complete the core task your business exists for without calling you during office hours, fix that before spending another dollar on content. Hand this one to a web developer. Ask for online booking, instant quote forms, transparent pricing pages, or a working checkout, whichever fits your model. Shepard's data put task completion at an 83% feature rate among winning sites. It is the cheapest structural advantage on the list.

Step 2: Build or surface one proprietary asset. Most businesses already hold something no competitor can copy. It might be real project photos with real outcomes, a decade of pricing data, an internal calculator your estimators use or results from your own customer base. Pick one and publish it properly. A roofing company publishing its actual average replacement costs by suburb, updated quarterly, owns something an AI cannot generate and other sites will cite. The instruction to your team is simple: find the data we already have that customers always ask about and put it on the site with our name on it.

Step 3: Reallocate the generic blog budget. Look at your last twelve months of content invoices. Any post that answers a generic question your customers could ask ChatGPT ("how often should gutters be cleaned") is losing its clicks. Ahrefs found a nearly 60% clickthrough drop when AI Overviews appear. Move that budget toward pages tied to buying decisions: service pages for specific jobs in specific places, comparison pages, pricing explainers, case write-ups with numbers. These are the branded, local and high-intent categories Fishkin identifies as still paying off.

Step 4: Tighten your topical focus. Shepard found tight topical focus correlated with growth while sprawling, cover-everything sites lost ground. If your plumbing site has a recipe section because someone once said more content equals more traffic, prune it. Then ask your agency: "What percentage of our pages are directly about what we sell, and what is the plan for the rest?" A confident agency answers with numbers.

Step 5: Demand evidence before buying any AI search package. In June 2026 Google published guidance warning businesses to think critically about third-party SEO tools and services. It noted that no vendor has access to Google's internal ranking data and that using a tool does not guarantee results. If a vendor pitches you AI visibility services, ask for one named client, the before-and-after numbers and the mechanism. Vague answers mean the 68% statistic is making the sale for them.

How do I judge search when fewer searches end in a click?

Raw traffic is the wrong grade for a business that sells something. The number is structurally rigged to decline for almost everyone now. An agency that reports traffic and leaves out revenue is hiding the ball, intentionally or not.

Grade the work on conversions from organic and AI-referred visitors (form fills, bookings, purchases, calls), on branded search volume and on revenue per organic visit. If traffic falls 20% while revenue holds, you shed the visits that were never going to buy. For the influence your reports never record, five cheap instruments close the attribution gap.

Rankings for generic informational keywords, total impressions and content volume make poor grades. All three can climb while the business wins no new customers.

I run my own site the same way. Over the 90 days to September 1, johntalaguit.com earned 1,230 impressions and 17 clicks in Google Search Console. I do not grade it on impressions or clicks. I built its measurement layer myself in GA4 through Google Tag Manager. The event I judge the site by is book_call_click, a click on my booking link. Scroll depth and read completion are tracked as separate events and kept off the scoreboard.

How long before the fixes show up?

Task completion fixes (Step 1) show up in conversion rates within 60 to 90 days because they affect every visitor immediately. A proprietary asset typically needs six months to a year to accumulate citations and links. Topical pruning and budget reallocation generally show effects across two or three quarters. Anyone promising AI search dominance in 30 days is selling the panic.

The 68% number is real and likely to get worse. Make your website the place where the job finishes and back it with a content strategy built around what you sell.

Zero-click is one layer of the bigger picture, which I cover in my guide to AI search visibility.

Frequently Asked Questions

My traffic is down but sales are steady. Should I be worried?

Probably not, and you may be looking at the new normal working as intended. The clicks disappearing first are informational visits that rarely converted anyway, while AI answers increasingly pre-qualify the people who do click through. Check revenue per organic visit over the past year. If it is rising while traffic falls, your site is shedding low-value visits and keeping the customers. Worry only if conversions and branded searches are falling alongside traffic.

Should I stop paying for SEO since most searches never click?

No, but you should change what you pay for. AI answers are built largely from sources that already rank well. Visibility work still decides whether you appear in those answers at all. What deserves cutting is generic blog content aimed at questions a chatbot now answers on the spot. Shift that budget to service pages, pricing transparency, local pages, and making your site a place where a visitor can complete a purchase or booking.

How do I know if my business is in the group that still wins clicks?

Run a simple test against the five features from the Zyppy Signal study of 400+ sites. Do you sell a real product or service? Can a visitor complete a task on your site without calling? Do you own original data, tools, or photos competitors cannot copy? Is your site focused on one topic? Do people search your name? Sites with four or five of these won around 70% of the time. Each yes moves you further from the danger zone.