Is Website Traffic Still a Valid KPI? [2026]

Q: What is zero-click search?

A zero-click search is one that ends without the user clicking through to any website, the answer is delivered on the results page itself, increasingly via AI Overviews and AI answers. SparkToro's 2026 clickstream research found 68% of US Google searches now end this way; of every 1,000 searches, only about 276 reach the open web.

Traffic reports are down almost everywhere, and marketing teams are being judged on a number that may have quietly stopped measuring what it used to. The industry's sharpest minds disagree about what it means, and their fight produces a surprisingly clear answer.

Bottom line up front Traffic reports are down almost everywhere, and teams are being judged on a number that may have stopped measuring what it used to. The data: 68% of US Google searches now end without a click; the first randomized experiment on AI Overviews measured a 39.8% click loss with zero improvement in user satisfaction; and B2B buyers now build vendor shortlists inside AI tools before visiting a single website. The industry's best minds disagree about the prescription, but their fight, read carefully, produces a clear answer: traffic didn't die as a metric. It got demoted, from verdict to symptom. Here's the debate, and the KPI stack that replaces traffic-only reporting.

§ The Monday-morning problem

Somewhere this week, a marketing manager is walking into a review meeting with a traffic chart that points down and no confident answer to the only question that matters: is marketing failing, or is the metric?

The stakes of getting that answer wrong run in both directions. Judge a working strategy by a dying metric and you'll fire the wrong people and fund the wrong fixes. Excuse a failing strategy with "traffic doesn't matter anymore" and you'll bleed for two quarters before anyone notices. So it's worth knowing what the people who study this hardest actually think, because they disagree, informatively.

§ The debate: three positions worth taking seriously

Rand Fishkin The click is dying

SparkToro's 2026 clickstream research found 68% of US Google searches end without any click; of every 1,000 searches, only 276 reach the open web, down 26% in two years. His prescription is "zero-click marketing": influence buyers on the surfaces where they already are (AI answers, social, communities) and stop measuring success purely by visits you'll never get back.

Eli Schwartz Don't panic

Working from 260-billion-session clickstream data, Schwartz's counters are sharp: total search volume is up roughly 26% since ChatGPT launched; users run Google and AI in parallel rather than replacing one with the other; and traffic loss concentrates in informational queries, while commercial-intent clicks are far more resilient. His line: "the data doesn't support extreme measures." Shift toward bottom-funnel and product-led work, yes; burn the traffic dashboard, no.

Cyrus Shepard It's the business model

Shepard studied 400+ sites that survived the 2024–26 declines and found survival wasn't predicted by tactical excellence at all, but by structural traits: strong brand, task completion, proprietary data. His uncomfortable summary: "no amount of tactical excellence can save you if your business model is wrong." Sites whose entire value was answering informational questions are the ones AI answers replace.

Three smart people, three positions. Then the tiebreaker landed.

§ The tiebreaker: the first randomized experiment

Nearly every traffic study until this year was observational, correlation confounded by seasonality and algorithm updates. In June, researchers from ISB and Carnegie Mellon published the first randomized controlled experiment: 1,065 real users, AI Overviews experimentally toggled.

📊 Results: organic clicks fell 39.8% when AI Overviews appeared; zero-click sessions rose 34.5%; and, the finding that matters most, user satisfaction didn't improve.

That last number quietly demolishes the comforting story that lost clicks are just "answered better." Users weren't happier; they just didn't click. The clicks aren't being replaced by satisfaction. They're being absorbed by the interface.

And on the buyer side, the shift is already behavioral, not predicted: Semrush's 2026 study of B2B buyers found 89% expect to rely more on AI for purchase research, with roughly seven in ten using AI tools daily, building shortlists inside the AI conversation before any website visit. By the time your analytics sees a visitor, much of the decision already happened somewhere your dashboard can't see.

§ The verdict: traffic was demoted, not deleted

Put the three positions and the experiment together and the synthesis is surprisingly clean:

Schwartz is right that commercial-intent traffic still matters. Track it, defend it, attribute revenue to it.
Fishkin is right that the top of your funnel has left your website, and needs to be measured where it now lives.
Shepard is right that this is a business-model question: if all your value was informational visits, no KPI change saves you.
The experiment proves the clicks aren't coming back and aren't being "quality-filtered". The interface simply keeps them.

Traffic didn't die. It stopped being the verdict on marketing and became one symptom among several.

A business measuring only visits in 2026 is reading yesterday's scoreboard: accurately, and irrelevantly.

§ The replacement KPI stack

What to measure instead, five metrics, all practical today:

Citation frequency on your prompt sethow often AI engines name you on your buyers' 10–25 real questions, measured with repeated runs against your own baseline (the honest method).
Branded search volumepeople who saw you in an answer and later searched your name. The classic proxy for invisible influence, newly relevant.
Direct-traffic trendsame logic: influenced buyers skip the funnel and type your URL.
The intake tallyone question added to every form and call: "How did you hear about us?" Count the answers that say ChatGPT, Gemini, or "an AI." Crude, honest, and connects visibility to revenue better than any dashboard.
Pipeline-source mixthe share of closed business that began with an AI-era signal (branded search, direct, AI-referred) versus classic organic clicks. That ratio is your personal version of this whole debate, updated monthly.

Keep from the old world: traffic and conversion on commercial-intent pages. Retire: celebrating informational-page traffic as a success metric. That's the traffic AI absorbed first, and it isn't coming back.

See where you stand before the next review meeting

FirePencil is built for the demoted-traffic era: its agent tracks your citation frequency across ChatGPT, Gemini, Perplexity and AI Overviews the honest way, repeated runs, trendlines, receipts, then executes the work that moves those numbers, weekly and owner-approved, with your SEO foundations maintained in the same system.

Run my free AEO audit → Try the free llms.txt generator

§ Frequently asked questions

Is website traffic still a valid KPI in 2026?

Yes for commercial-intent pages, no as the single verdict on marketing. Traffic was demoted, not deleted: 68% of searches end without a click and a randomized study measured a 39.8% click loss from AI Overviews, so the top of the funnel has left the website. Commercial-intent clicks are more resilient and still worth tracking; measuring only visits means reading yesterday's scoreboard.

How much do AI Overviews reduce clicks?

The first randomized controlled experiment (1,065 users, ISB and Carnegie Mellon, 2026) measured organic clicks falling 39.8% when AI Overviews appeared and zero-click sessions rising 34.5%, with no improvement in user satisfaction, undercutting the idea that lost clicks were "answered better."

What is zero-click search?

A search that ends without a click to any website, the answer is delivered on the results page, increasingly via AI Overviews. SparkToro's 2026 clickstream research found 68% of US Google searches now end this way; of every 1,000 searches, only about 276 reach the open web.

What should replace website traffic as a KPI?

A five-metric stack: citation frequency on your buyers' real prompt set, branded search volume, direct-traffic trend, an intake tally ("how did you hear about us?" counting AI answers), and pipeline-source mix. Keep traffic and conversion on commercial-intent pages; retire celebrating informational-page traffic.

Do B2B buyers still visit websites before choosing a vendor?

Increasingly they build the shortlist first inside AI tools and visit later, if at all. Semrush's 2026 study found 89% of B2B buyers expect to rely more on AI for purchase research, with ~70% using AI tools daily. By the time analytics sees a visitor, much of the decision already happened.

FirePencil.AI is an autonomous AEO agent. This article summarizes third-party research and public commentary (SparkToro, Semrush, and the cited ISB/Carnegie Mellon experiment) for general information; figures are as reported by those sources. It is not a guarantee of any specific result. Third-party names (ChatGPT, Gemini, Perplexity, Claude, Google AI Overviews) are trademarks of their respective owners; use is descriptive.

Is website traffic still a valid KPI?