Originally published on The Searchless Journal
The newest Similarweb report on generative AI search was supposed to infuriate two camps. The AI zealots who believe every marketing dollar should already be chasing chatbot visibility, and the skeptics who always suspected the hype was overblown but lacked the data to prove it. The 38-page report manages to hand both sides a piece of evidence that undercuts their certainty. Having read the whole thing, the more interesting finding is not who is right or wrong. It is that the entire debate is framed incorrectly. AI search has not replaced traditional search. It has stacked a new, fast-growing, unevenly distributed layer on top of a search ecosystem that was already there.
The Number That Challenges the Zealots
Similarweb tracked audience overlap between ChatGPT and Google between March and May 2026. Of ChatGPT's 494 million users, 461 million also use Google in the same window. That is 95% overlap. Almost nobody has left Google for ChatGPT. They have added ChatGPT to a Google habit that has not measurably budged.
Zoom out and the gap becomes starker. Search still pulls 3.3 billion average monthly unique visitors worldwide. AI chatbots, even after growing 57% year over year, sit at 655 million. Search is still roughly five times the size of the entire AI chatbot category combined. If your 2026 budget deck assumes AI search has already eclipsed traditional search, the math in this report disagrees.
Citations tell the same story from a different angle. As of May 2026, only 6.8% of ChatGPT answers in the United States included a link to an external source. That figure is up more than fivefold from roughly 1% a year earlier, which is genuinely fast growth in relative terms. In absolute terms, it means 93 out of every 100 ChatGPT answers still send nobody anywhere. The citation economy is real and expanding, but it is starting from a base so small that declaring it the dominant discovery channel requires ignoring the scale of what still dwarfs it.
Ethan Smith of Graphite makes a sharper observation in the report. Users are folding the prompting habits they learned in ChatGPT back into Google itself. Average query length on Google has been climbing steadily since AI Mode launched. People are not abandoning search boxes. They are typing longer, more conversational sentences into the same search boxes they always used.
The Number That Challenges the Skeptics
Now for the half of the report that should puncture overconfidence in the other direction. Average monthly web visits across generative AI platforms hit 9.5 billion between June 2025 and May 2026, up 70% year over year. App downloads worldwide climbed to 2.7 billion, up 134%. Half of all generative AI users are now 35 or older, compared to 61% under 34 just two years ago. That demographic shift is the clearest signal that this is not a Gen Z novelty running out its trend cycle.
Michael Horrocks of Miro puts it plainly in the report: growth concentrated in younger demographics can fade with trends. Growth spreading into older generations is often what durable, mainstream adoption looks like.
Meta AI's own disclosed numbers confirm from a different angle. Publicly reported monthly active users went from 384 million in September 2024 to 1.2 billion by March 2026, more than tripling in 18 months. That growth happened entirely by riding inside Instagram, Facebook, WhatsApp, and Messenger rather than as a standalone destination anyone had to seek out. The implication for visibility strategy is significant. A meaningful share of AI-assisted discovery now happens inside surfaces that do not look like search at all and that traditional SEO and even GEO frameworks do not cover.
On the monetization side, ChatGPT ad penetration in the United States jumped from 14% of desktop chats in May 2026 to 26% just one month later. Whatever you think about the maturity of AI search as a channel, advertisers do not think it is experimental anymore.
Why the Disconnect Between Citations and Clicks Matters More Than Either Camp Realizes
The most consequential data point in the report comes from Aleyda Solis of Orainti. She highlights that 65% of the URLs ChatGPT cites sit two or three folders deep. These are the pages doing the actual evidentiary work behind an AI answer. But 58.8% of the referral traffic that AI sends back to sites lands on the homepage, not the cited page at all. Cited pages and clicked pages are almost entirely different populations of URLs.
This single data point should reorganize how any team reports AI performance. If you are only tracking whether your deep product or blog content gets cited, you are missing the fact that the humans who actually click through are landing somewhere else entirely and need their own conversion path. The cited page earns the trust of the model. The homepage earns the click from the human. Those are two different objectives served by two different pages, and measuring only one of them means you are optimizing for half the funnel.
There is an analogy here that clarifies the dynamic. Twentieth-century advertisers proved that billboard and radio spend worked by measuring lift in store visits, not by counting who glanced at a billboard. The mechanism has changed. The discipline of measuring downstream behavior instead of surface impressions has not.
Three Strategy Shifts to Make Now
The practical takeaway is not to pick a side in the replacement debate. It is to treat the search stack as a stack, measure each layer separately, and allocate budget based on where actual human behavior lands.
First, split your reporting into two separate metrics. Track citation rate and citation folder depth as one key performance indicator that measures whether AI trusts your content enough to use it as evidence. Track referral landing pages and downstream conversion as a completely separate KPI that measures what happens once a human clicks through. Conflating the two in a single dashboard is how brands miss both problems at once. A high citation rate with low referral traffic means your deep content is trusted but your homepage is not converting the spillover. A low citation rate with high referral traffic means your brand is visible enough to generate clicks but not authoritative enough to be cited as a source. These are different problems requiring different fixes.
Second, stop treating AI visibility as a single category. Similarweb's brand visibility index shows how category-specific the landscape already is. In beauty, CeraVe leads with an index of 100 while NYX Cosmetics sits at 19 in the same category. Kevin Indig argues in the report that share of voice is the metric that matters because it is a relative comparison in a stochastic system, not an absolute score. Pull your own category's leaderboard before assuming you are winning or losing. A brand can be dominant in AI visibility for one query cluster and invisible for another within the same vertical, and aggregated scores will hide both extremes.
Third, measure your share of the combined stack, not just the AI layer. If search pulls 3.3 billion monthly users and AI chatbots pull 655 million, your share of visibility across both layers combined is what determines your actual reach. A brand that dominates traditional search but is invisible in AI citations is still reaching the majority of its potential audience through the search layer. A brand that is strong in AI citations but weak in traditional search is reaching a growing but still much smaller pool. Neither position is inherently wrong. Both need to be understood in the context of total addressable discovery before budget decisions are made.
The Real Story Is About Layering, Not Substitution
The debate between AI zealots and skeptics is structured as a zero-sum question. Will AI search replace Google? The data says no. The data also says the gap is closing faster than skeptics expected, and the behavioral patterns of users who adopt AI search are shifting in ways that will compound over time. People who use ChatGPT are not stopping their Google searches. They are adding a second discovery step, running different query types through each channel, and gradually allocating more of their research and evaluation work to the AI layer.
That behavioral layering is what makes the stack metaphor more useful than the replacement metaphor. You do not choose between optimizing for Google and optimizing for AI. You optimize for the sequence of queries a real user runs across both surfaces during a single research session. That sequence might start with a Google search, move to a ChatGPT prompt for synthesis, return to Google for a specific transactional query, and end with a click on an AI-provided citation that happens to land on your homepage.
Every step in that sequence is a measurement opportunity and an optimization target. Treating them as separate channels with independent KPIs is the only way to see where you are losing people between layers. Treating them as a single funnel where AI visibility is just another form of SEO will leave blind spots exactly where the layering happens.
What to Track Starting This Quarter
Build a combined visibility dashboard with three panels. Panel one: traditional search performance, including rankings, impressions, clicks, and conversion by query cluster. Panel two: AI citation performance, including citation rate, citation depth, brand mention frequency, and model-by-model comparison across ChatGPT, Perplexity, Gemini, and Claude. Panel three: cross-layer behavior, including referral traffic from AI sources landing on your site, the specific landing pages receiving that traffic, and conversion rates for AI-referred sessions compared to search-referred sessions.
The third panel is where most teams have no data at all. It is also where the highest-leverage insights live. Knowing that ChatGPT cites your deep guide pages but sends traffic to your homepage tells you exactly where to focus conversion optimization. Knowing that Perplexity users convert at twice the rate of Google users tells you where to invest in citation depth. Knowing that your category's AI visibility is concentrated in three competitor brands tells you whether the gap is closeable or structural.
The teams that win the next two years are not the ones who correctly predicted whether AI would replace search. They are the ones who measured the stack, found their gaps, and optimized for the actual path a user takes across both layers. The data has been available for months. Most teams have not built the dashboard yet. That is the opening.
Top comments (0)