The uncomfortable part of marketing measurement has always been that the numbers we report are the ones we can collect, not the ones that matter. That gap is widening quickly, and this summer produced enough evidence to make it hard to ignore.
Start with the click. SparkToro, using Similarweb's US clickstream panel, found that between January and April this year 68 percent of Google searches ended without a click. In 2024 it was 60 percent. In 2016 it was roughly 45. The detail I found most useful is that AI Mode accounted for only about a third of one percent of searches in that period, so the loss isn't coming from the thing everyone is watching. It's coming from AI Overviews, which appear on more than a fifth of searches and cut clickthrough substantially.
Publishers have felt this first because they had the least buffer. Nieman Lab reported in July that some publications saw search traffic fall more than 40 percent between June 2025 and June 2026, with The Guardian and the BBC as exceptions. Reuters Institute's January survey of 280 media leaders put the global decline in Google referrals at about a third year over year. Those are the sharpest numbers, but the mechanism is the same for any brand that treated organic search as a channel it owned.
What's actually happened is that a layer has been inserted between the question and the answer, and that layer doesn't report back. Someone asks an assistant which agency to call, or which product to buy, and either your brand is in the answer or it isn't. There's no referrer, no session, no path to attribute. The event happened and left no trace in your analytics.
The industry's first serious attempt at a response came on the third of August, when the IAB published a framework for measuring visibility in the AI era. It's organized around what they call the four P's: presence, meaning mention rate, citation rate and share of voice; prominence; portrayal, which includes how often a system says something inaccurate about you; and persuasion. It also draws a line between directional measurement and what they call decision-grade measurement, which is a polite way of saying that most of what's currently sold as AI visibility data isn't good enough to spend money on. The IAB notes more than twenty companies now selling tools in this category, each with its own methodology.
I take the framework seriously mostly because of that last point. Twenty vendors with twenty definitions is how a measurement category dies before it starts. Someone had to name the terms.
There's a related finding worth sitting with. Webflow's AEO maturity index looked at 2,000 websites and found the median company appears in about 16 percent of relevant AI answers but receives a linked citation only about 6 percent of the time. Being in the answer and being credited for it are now separate outcomes, and only one of them shows up in your reporting. If you optimize for the one you can see, you'll systematically underinvest in the one that's actually working.
I don't think the honest response is to find a replacement metric quickly. It's to accept that a real portion of demand is now generated where you cannot observe it, and to rebuild measurement around the things that still hold. Brand search volume. Direct traffic. Whether people arrive already knowing what you do. Asking customers how they found you, which is unfashionable and imperfect and currently more reliable than the dashboard.
The other adjustment is temperamental. A generation of marketers was trained to believe attribution was a solved problem that only needed better tooling. It was never solved. It was legible for about fifteen years, and that turns out to have been a phase rather than a permanent condition. Working without full visibility is the older discipline, and the people who ran brands before the dashboard existed had methods. They're worth reading again.
Sources
- In 2026, less than one third of Google searches still send a click (SparkToro): sparktoro.com
- Measuring Visibility in the AI Era (IAB, August 2026): iab.com
- Search traffic has declined so much that some publishers are considering opting out of Google (Nieman Lab): niemanlab.org
- AI visibility is no longer about referral traffic (Digiday, on the Webflow AEO Maturity Index): digiday.com