Google expands AI search, shifting value from clicks to answers

The panic is the wrong story

The loudest read on Google’s latest Search changes is that SEO got blown up overnight. That is the shallow take. The better read is more uncomfortable for agencies: Google is not abandoning the web, it is reorganizing how value flows through search, and the loss lands first in measurement, traffic assumptions, and content economics.

What changed is the front door. Search now starts more often with AI-generated help, richer inputs, and more mediated answers, while the classic blue-link result is pushed farther from the default experience. The web still matters, but the path from query to click is less direct, which means the old habit of treating raw sessions as the main scorecard is now a liability.

What Google actually changed

The biggest product move is the expanded Search box. It now accepts images, files, videos, Chrome tabs, and text, which makes search feel less like a keyword box and more like a command layer for whatever the user has on screen or saved locally. That is a meaningful shift for discovery strategy because it broadens the kinds of prompts Google can interpret and answer without forcing a traditional web search first.

Google also made Gemini 3.5 Flash the default AI model globally in AI Mode. On top of that, Google said AI Mode had passed one billion monthly users and that queries were doubling quarterly. Those are not cosmetic numbers. They signal that the answer layer is already operating at scale, and it is getting more embedded in the core search experience every quarter.

There is more coming. Information agents are slated for this summer to monitor the web for updates such as apartment listings or product changes, and generative UI features plus mini apps are on the roadmap for U.S. subscribers. That points to a future where Google is not just matching users to pages, but handling follow-up work inside the search interface itself.

The web is still in the loop, but not in the same way

This is where the panic narrative gets sloppy. Traditional results still exist, including through the Web tab, and Google has said AI Mode is not the default experience in Search. So no, the web is not vanishing, and blue links are not dead.

But they are losing their privilege. If the first screen is increasingly an AI answer, a multimodal prompt, or an agent-assisted result, then the click comes later, if it comes at all. That changes the job of search marketing from “win the page one click” to “win a place in the system that produces the answer.”

That distinction matters because agencies have spent years optimizing for outcomes that Google is now diluting. A ranking alone is no longer enough if the user gets what they need inside the interface before they ever reach the site.

Where the real agency risk sits

The economic risk is not that every click disappears. It is that the wrong clicks disappear first. Low-intent visits, quick-bounce informational queries, and casual comparison traffic are the most vulnerable when the answer is summarized upfront or handled by an agent. Even if your content still powers the response layer, your site may get less direct traffic from it.

That creates a measurement trap. Teams that still judge search performance mainly by sessions and top-of-funnel traffic will think demand fell off a cliff, when what really changed is the distribution of value. The answer may still be driving authority and relevance, but not always a visit. If you do not separate visibility from visits, you will misread the channel and overreact in the wrong places.

The smartest agencies should audit three things immediately:

• SERP visibility across query types, not just branded rankings
• Click loss by intent type, especially informational versus commercial queries
• Downstream conversion efficiency, so the traffic that does come through is worth more

That audit tells you whether you are losing awareness, losing clicks, or simply losing junk traffic that never converted well in the first place. Those are very different problems.

How service offerings need to shift after I/O

This is where strategy has to catch up. The old package of keyword research, content production, and reporting on organic sessions is too narrow for a search product that now blends multimodal input, AI summaries, and agent behavior. Agencies need to sell visibility in a broader sense: source eligibility, answer inclusion, and business value from the users who actually arrive.

That means content plans have to be built with source trust in mind. Pages need clearer authorship, stronger proof, and tighter topical authority, because the value is no longer just “can this rank,” but “can this be trusted as input to the answer layer.” The winner will not be the site with the most words. It will be the site Google is most comfortable pulling from when it needs to answer quickly and credibly.

It also means conversion work matters more. If fewer people click but the ones who do are deeper in intent, then landing pages, lead capture, pricing clarity, and product positioning have to do more heavy lifting. Agencies that can connect AI-era search visibility to revenue will be useful. Agencies still selling traffic volume as the main prize will look dated fast.

A practical read for growth teams

The right response is not to declare SEO dead or to pretend nothing changed. Google has kept the web in the system, but it has moved attention toward AI-first interactions and away from the old default of search results as a list of links. That is a structural change, and it affects how agencies should budget, report, and prioritize work.

Treat this as an audit moment, not a eulogy. If your reporting cannot show where clicks are leaking by intent, whether your pages are still eligible to be sources, and how much revenue the surviving clicks produce, then your search program is built for a version of Google that no longer exists. The brands that adapt fastest will not be the ones chasing more traffic at any cost. They will be the ones measuring visibility, trust, and conversion as one system.