Pew Research tracked the real browsing behavior of 900 US adults across 68,879 Google searches in March 2025. The finding: when an AI summary appeared above the traditional results, people clicked through to one of those results on 8% of visits. When no summary appeared, they clicked through on 15% of visits. Roughly half the click-through vanishes the moment a summary loads. No ranking has to move for that to happen. This is zero-click search: a result that still ranks well while it quietly stops sending anyone anywhere.
This site already covered one consequence of AI-mediated search: an AI agent gathering quotes decides who’s even in the running before a human ever compares a lineup of options. That’s a story about selection. This is a narrower story about economics. It doesn’t require an operator to lose a single ranking position. It doesn’t require any AI to choose anyone. It only requires the search results page to change shape underneath a ranking that never moved. That shift hits every operator running a ranking-based acquisition strategy. It doesn’t matter how well that strategy performs today.
Zero-Click Search: What Doesn’t Show Up in a Rankings Report
A ranking-based strategy, organic SEO or paid search, rests on one assumption: that ranking well earns a predictable share of clicks. Pew’s numbers say that assumption is quietly breaking. The system underneath the strategy changed. The strategy didn’t. When an operator ranks exactly as well this year as last year, that operator can still watch organic sessions fall by a large fraction, purely because more of the searches that used to send a click now resolve inside a summary instead.
The same study found a second effect that compounds the first. People are also more likely to abandon the search entirely once a summary is on the page: Pew logged session abandonment at 16% on pages without a summary. That climbs to 26% the moment one appears. A visitor who doesn’t click through and doesn’t search again never gives an operator’s marketing a second chance, on that visit or the next.
Pew isn’t the only source finding this. Ahrefs analyzed Google Search Console data across 300,000 keywords from December 2023 to December 2025 and found the same pattern from a different angle: for the page ranked first, the presence of an AI summary now correlates with a 58% lower average click-through rate. A behavioral study and a raw click-log analysis, run independently, land on the same order of magnitude.
Most rank trackers and analytics dashboards don’t separate these two causes. A rank tracker shows the same position it showed last quarter. Traffic is down. The obvious read is that something’s wrong with the SEO, the content, or the offer. The actual cause can be nothing an operator did: the mechanics of the results page changed underneath a ranking that never moved. The dashboard has no field for that.
Pew’s numbers are a snapshot from March 2025. The underlying trend hasn’t paused since. SparkToro’s own analysis of Similarweb clickstream data found that 68% of US Google searches ended without any click at all in the first four months of 2026, up from roughly 60% in 2024. The analysis names AI Overviews as a driver of that acceleration, even though AI-native search interfaces still account for a small slice of total search volume on their own. Whatever share of an operator’s searches already resolves inside a summary, that share is almost certainly higher than it was a year ago, and still climbing.
What an 8% click-through does to a pipeline
Half the click-through sounds abstract. A funnel makes it concrete.
An operator ranks well for a cluster of local moving-related keywords, the kind of terms a well-built local page can hold a top-three position on for years: “movers in [city],” “long-distance moving company [city],” “cost to move a two-bedroom apartment.” That cluster draws roughly 2,000 relevant Google searches a month, a plausible number for a mid-size metro operator whose site has been built out for search. At the old 15% blended click-through rate, that’s 300 organic sessions landing on the site. At a 5% site-to-quote-request rate and a 30% close rate on those requests, the funnel produces roughly 4 to 5 booked jobs a month. At $5,000 average revenue per move, that’s about $22,500 in monthly revenue traceable to that one keyword cluster, at zero incremental cost per lead.
The table below holds every ranking fixed and projects the AI summary’s share of those same 2,000 searches upward, in the direction it’s already moving:
| Share of searches showing an AI summary | Blended click-through | Organic sessions/month | Booked jobs/month | Monthly revenue |
|---|---|---|---|---|
| 0% (2025 baseline) | 15.0% | 300 | ~4.5 | $22,500 |
| 25% | 13.25% | 265 | ~4.0 | $19,875 |
| 50% | 11.5% | 230 | ~3.5 | $17,250 |
| 75% | 9.75% | 195 | ~2.9 | $14,625 |
| 100% | 8.0% | 160 | ~2.4 | $12,000 |
At full exposure, the same 2,000 searches, the same ranking, the same page, produce close to half the revenue: $12,000 instead of $22,500. A $10,500 monthly gap with no single event an operator could point to.
That’s a model built on round, illustrative numbers, not a benchmark for any real business. The two real inputs are Pew’s 8% and 15%. Everything else, search volume, conversion rates, deal size, will vary by operator and should be swapped for real figures to see a real number. What doesn’t vary is the shape of the curve. Revenue erodes continuously as the AI summary’s share of the relevant search volume climbs, long before it reaches 100%. No rankings report is built to catch a decline shaped like that.
Why the research phase is more exposed than the booking phase
A 50% AI-summary exposure share isn’t a distant hypothetical for the kind of searches that happen early in a move. Pew’s own breakdown shows exactly which searches are most likely to trigger a summary in the first place: 53% of searches with ten words or more did, against just 8% of one- or two-word searches. Question-format searches triggered a summary 60% of the time: “how much does it cost to move a two-bedroom apartment across the state,” “what’s included in a full-service move,” “when should I book movers for a summer relocation.”
That’s the shape of query a person doing real research for a move types. It isn’t the shape of query a person doing a quick local lookup uses. A move isn’t decided in one search. It involves comparison shopping, cost estimation, and timing questions across a string of longer, more specific queries over several sessions, long before anyone searches “movers near me” and picks up a phone. Every one of those upstream research queries sits well above the 18% average AI-summary rate Pew found across all searches, and closer to the 50-60% range reserved for exactly the long, question-shaped searches that dominate early-stage research.
An operator selling something bought once and compared carefully depends more on organic search than a business selling something searched and bought in a single session. That dependence runs headfirst into the query types AI summaries already favor most. Staying invisible during research while still winning the booking leaves a thinner margin than a business’s current call volume might suggest.
“Our near-me searches still convert”
The objection that near-me search traffic is holding steady is fair, and it deserves a real answer rather than a dismissal. The final query in a moving decision tends to be short and transactional, something like “movers near me” or a branded search for a company already on a shortlist. That’s exactly the query shape Pew found least likely to trigger a summary at all. An operator watching that specific query hold steady isn’t wrong about what they’re seeing.
What that read misses is everything upstream of it. The AI-summary drop doesn’t happen at the final, branded, decision-stage search. It happens at the research-stage searches that used to be how a business earned its way onto a shortlist in the first place: the cost questions, the comparison questions, the “how does this work” questions that build awareness of a business a customer had never heard of before that search. An operator tracking only the last-click query sees stable traffic on the one query that was never exposed, while losing volume on the queries that used to introduce new customers to the business at all.
The businesses most protected from this are the ones a customer already knows to search for by name: built through referrals, past customers, or a demand partner sending work under an established brand. The businesses most exposed are the ones still counting on a stranger’s first, generic research query to discover them. Real analytics can check this directly: is traffic to branded and near-me queries holding steady while traffic to broader, informational pages quietly falls? Reporting that doesn’t separate the two makes that split invisible either way.
What changes instead
No fix restores the old click-through rate. That number belongs to how people use search now, not to anything an individual operator’s website, content, or SEO spend controls. The AI summary hasn’t outranked anyone. It has inserted itself as the new layer between a query and a decision, and it keeps the relationship with the searcher for itself. Publishing better content doesn’t make Google stop summarizing it. Ranking first doesn’t help when the position above rank one absorbs the click.
What changes is where new demand has to come from to make up the difference. An operator’s Direct Demand Ratio becomes a more load-bearing number than it used to be: the share of revenue coming from repeat customers, referrals, and channels a business owns outright, rather than search traffic it has to win fresh every time. A referral system, a direct relationship with relocation-adjacent partners, or a demand partner not solely dependent on the same ranking mechanics can produce bookings without asking a stranger’s search query to survive an AI summary first.
None of that replaces organic search overnight, and it shouldn’t try to. It’s a hedge against a channel that’s quietly losing yield industry-wide, for reasons no operator caused and none can fix by working harder at the same strategy.
A drop with no line item
If an operator’s organic traffic fell by close to half over the next year, with no ranking ever moving, would the monthly report even show why? Or would it just read as a marketing problem to go fix?
Most reporting built for a ranking-based world has no field for “the results page changed shape.” It has a field for position, a field for traffic, a field for conversion rate. None of those isolate an AI summary’s effect on click-through from an actual SEO problem, so the two get diagnosed the same way: something’s wrong with the content, the site, or the offer. The real answer might be that nothing is wrong at all. The ground underneath a working strategy just moved.
Mapmakers have always chosen what a traveler could see before the traveler ever left home. A medieval cartographer who left an unmapped kingdom blank, or drew a rival city smaller than its true size, shaped which routes people took and which ones they never considered, and almost nobody who used the map ever questioned why the choices were made that way. An AI summary is doing the same selective mapmaking to a page full of search results, just faster and less visibly than ink ever was.
Movaros builds demand that doesn’t depend on the click surviving.
Building on shared infrastructure means referrals and repeat business keep producing revenue even as organic click-through keeps falling.
