AI search is
quietly becoming the default.
Six months ago, AI search was still seen as "error-prone training wheels." A year later it has become the main battleground of Google Search itself. Here's what Google's own announcements and TechCrunch's new market data show about the pace pushed by Gemini-powered AI Overviews.
Six months ago: "training wheels."
Today: the default.
Data reported by TechCrunch confirms AI search has become the main battleground.
Six months ago, AI search was still viewed as "error-prone training wheels" that couldn't be fully trusted. For most users, search still meant scanning ten blue links and clicking through to compare them — even Google's own AI Overview felt like a bonus box that happened to appear, not the main event.
That premise has flipped over the past year. According to data from market intelligence firm Similarweb, reported by TechCrunch on July 27, 2026, the share of Google searches showing an AI Overview has climbed from 15% a year ago to 43% today. Monthly visits to the conversational AI Mode have also more than doubled, from 126 million in June 2025 to 279 million in May 2026. TechCrunch described the shift as a move "from an era when Google was the entry point to the web, to one in which Google itself is the destination."
| 6–12 months ago | July 2026, now |
|---|---|
| AI Overview share: 15% | AI Overview share: 43% |
| AI Mode visits: 126M/mo | AI Mode visits: 279M/mo |
| Ten blue links as the arena | Being cited in AI answers is the arena |
| Seen as "error-prone training wheels" | Gemini 3 is the default response model |
A year of growth, in numbers
Cross-referencing Similarweb's tracking with Google's own announcements shows just how fast this moved.
One driver behind this growth: on January 27, 2026, Google announced it was switching the default AI Overviews response model to Gemini 3. The official Google blog said the upgrade lets AI Overviews return "a best-in-class AI response right on the search results page." It's hardly a coincidence that the model built by Google DeepMind was followed, within six months, by AI Overview's share nearly tripling.
Search is turning from a place
you choose from, into a placeyou finish reading.
Who it hits, and how
Marketers, business leaders, and product teams each feel this differently.
For marketers
Designing purely for "rank higher, win the click" now hits a ceiling. You need to build for citation inside AI answers (GEO/AEO) alongside — not instead of — your existing SEO program.
For business leaders
Acquisition plans built on assumed search traffic and cost-per-click need a rethink. Track "citations inside AI answers" as a new metric alongside CPA and conversion rate.
For product / PM
Rewrite explainer pages and FAQs into clear definitions, comparison tables, and structured data that AI systems can lift easily — that raises the odds of being cited. Track it via Search Console's generative AI report and similar tools.
What to do next
"Wait and see" is no longer an option. Start with the highest-priority moves.
Start measuring AI citations
Set up tracking for how often your content is cited inside AI answers, via Search Console's generative AI report or third-party tools. Most companies still can't see this at all.
Rewrite content to be citation-friendly
Lead with the conclusion, use clear definitions, comparison tables, and scannable structure. Conclusions buried under long preambles rarely get picked up by AI answers.
Diversify away from search-only acquisition
Build up direct channels — email, social, community — in parallel, so traffic isn't entirely at the mercy of whether an AI Overview appears.
This isn't all upside
First caveat: zero-click growth isn't a straight line. Data firm Datos found the U.S. zero-click rate actually dipped from 24.5% in December 2025 to 22.4% in March 2026, and SparkToro's analysis suggests the decline may be leveling off rather than accelerating indefinitely.
Second, Google's own search share is still above 90%. Most of this shift is a change happening inside Google — from link results to AI answers — rather than a mass migration to rival AI search products like Perplexity or ChatGPT's search feature, which remains comparatively limited so far.
Third, click-through rates on searches that show an AI Overview still tend to run lower than on plain results, and the impact varies a lot by industry and query type. Treating "just do AEO" as a guaranteed fix oversimplifies things — the metrics need ongoing monitoring, not a one-time playbook change.