By Michael Semer
What you’ll learn:
- Zero-click search has gone mainstream: 58–68% of Google searches now end without a click, and AI Overview queries resolve without one 83% of the time — 93% inside AI Mode.
- AI referral traffic is growing 130–150% year over year, but it rewards citation, not ranking — showing up in an answer now matters more than showing up on page one.
- Healthy B2B funnels still convert only 1–3% of leads to closed-won, so chasing volume without fixing yield just moves more water through the same leaky pipeline
For about twenty years, “get more leads” was a perfectly good marketing strategy, in the same sense that “get more calories” is a perfectly good nutrition strategy. Technically true. Directionally catastrophic if it’s the whole plan.
That plan is now expiring, and it’s not because marketers got smarter. Believe me. It’s because the machinery underneath it broke.
The funnel that fed you traffic doesn’t work like it did
The blue links you built your career as a content marketer on are quietly being replaced by a paragraph Google writes for the user and hopes they don’t scroll past.
Ahrefs’ December 2025 analysis found that pages ranking first now see 58% fewer clicks when an AI Overview appears above them — up from a 34.5% hit measured just eight months earlier. Seer Interactive clocked an even steeper drop for AI Overview queries, with organic click-through falling from 1.76% to 0.61%. Some of this rebounded slightly in early 2026, but the direction hasn’t reversed, and the swings just make the numbers harder to plan around.
Zero-click search — the searcher who gets an answer and never visits anyone’s website — is no longer an edge case. Fresh data puts overall zero-click rates in the high 50s to mid-60s percent, and searches that trigger an AI Overview resolve without a click 83% of the time; inside Google’s AI Mode that climbs to 93%. Mobile is worse than desktop, which makes sense. Informational queries are worse than transactional. The traffic you used to convert into a form-fill is being consumed before it ever reaches your form.
Meanwhile? The traffic that does still leave a trail is migrating to a different set of gatekeepers. AI-driven referral traffic is growing 130–150% year over year. ChatGPT alone is up roughly 27% year over year in referrals, and Perplexity referrals grew around 180% even as its share of the overall AI-traffic pie shrank against faster-growing rivals. None of that traffic cares whether you rank #1 on a results page that fewer people are scrolling through in the first place.
So the top of the funnel, the part “more leads” was measuring, has been rerouted through a system that answers questions instead of sending traffic, and a second system that occasionally cites you by name if a language model happens to like your sourcing. Optimizing the old funnel harder just means optimizing a pipe with a smaller hole in the bucket at the end.
The math was always uglier than the dashboard admitted
Here’s the part “more leads” conveniently skipped over: most leads were never going to become customers. Healthy B2B funnels convert somewhere between 1% and 3% of inbound leads to closed-won; even top-performing teams top out around 5–8%. Stack that against a stage-by-stage funnel where lead-to-MQL runs 20–25%, MQL-to-SQL another 12–18%, and closed-won off a qualified opportunity lands around 6–9%, and a “more leads” strategy is really a strategy of feeding a leaky pipe more water and hoping the leak stays proportional.
It rarely does. A single weak stage (bad lead scoring, a slow SDR response, a form that collects the wrong intent signal) sneakily halves your yield no matter how much top-of-funnel volume you throw at it. Though a lot of brands do, indeed, throw a lot at TOF, and it’s often the wrong kind of content to drive engagement.
Marketing teams kept reporting lead count because lead count was legible, chartable, and flattering in a QBR. Pipeline value, sales-cycle length, and win rate by source were harder to isolate and easier to ignore. So the industry optimized the metric that was easy to move, not the one that paid the bills.
A prediction about the next twelve months
Here’s a claim you can hold us to: over the next twelve months, the B2B companies still reporting raw lead volume as a headline marketing KPI will show flat-to-declining pipeline value even where lead counts stay flat or rise.
The mechanism isn’t mysterious — AI-mediated discovery pre-filters intent before a prospect ever fills out a form, so the leads that do arrive are already more qualified and, in aggregate, fewer.
Volume and value are decoupling. Track both for a year on your own funnel and you’ll either prove us right or hand us a case study in why we’re wrong. Either outcome is more useful than another slide about MQL count.
It’s not that heroic of a prediction, because I’ve already seen it happen in various forms in the past few years. Even pre-AI.
What replaces it
The strategy that survives isn’t “more leads.” It’s earning a citation in the three or four places a buyer now forms an opinion before they ever type your company name into a search box: the AI Overview, the ChatGPT or Perplexity answer, and whatever’s left of page one.
That means structuring content so a model can quote it accurately, building the kind of source authority that gets you cited rather than paraphrased into oblivion, and measuring share of answer the way you used to measure share of voice.
None of that is more comfortable than counting form-fills. But it is, at least, a strategy for staying viable and visible on the internet that you now have to reckon with.