How SEO agencies are adapting to agentic search
- 4 hours ago
- 5 min read
Author: Zoe Ashbridge
The message from Google’s conference, Google I/O, was clear: the age of agentic search has arrived.
CEO Sundar Pichai’s keynote declared Google was “firmly in [the] agentic Gemini era.”
Following that, Liz Read, Google's VP of Search, said, “incredible AI features aren't just in search; Google Search is AI Search—through and through.” Google processes 3.2 quadrillion tokens per month, seven times what it processed two years ago, as Pichai highlighted.
That leaves search specialists and agencies reassessing their strategies for this new age of search. Here’s what's changing, and what to do about it, according to thought leaders in the industry. Ahead, you’ll hear advice and learnings from:
Garrett Sussman, Director of Marketing at iPullRank
Suganthan Mohanadasan, co-founder of Snippet Digital
Simon Schneiders, founder and CEO of Blue Array
What can AI agents do in search?
Agentic search is a new form of search that goes beyond returning a list of links. Unlike traditional search, agentic search can autonomously research, reason, and take action on a user's behalf. It can add items to a cart, trigger back-in-stock alerts, or surface timely information like stock price changes. Agentic search is search that does things, not just finds things.
Research and recommendations
Agents can build personalized, interactive experiences from inside the search results page. In Google's I/O demo, Gemini generated a custom weekend planner by connecting Gmail, Photos, and Calendar, and produced restaurant reservations laid out on Maps in seconds. Not a single website was visited.
If search can deliver a fully bespoke, data-connected experience without a click, websites will be visited less than they are today. (See the Wix State of Websites report for more.)
Monitoring and alerts
Search-based agents can also monitor the web on a user's behalf, around the clock. Rather than a user actively searching for updates, they set a standing instruction, like track biotech stocks with positive cash flow and low debt, and the agent surfaces relevant information only when something meaningful happens. Search comes to you.
At Google I/O, Liz Read shared a finance example:
Booking and transactions
Agents don't just find products—they buy them. Provided a website is set up to allow it, an agent can add items to a cart and complete a purchase without any human involvement. They also bring a layer of consideration to the process; in one Google I/O demo, an agent flagged that a motherboard in a user's cart was incompatible with a processor they'd already chosen, and suggested an alternative. Fewer returns, fewer support tickets.
How SEO agencies are adapting to agentic search
The shift to agentic search changes three things for agencies: how websites need to be built, how success gets measured, and how clients need to be managed. The experts we spoke to are already adapting. Here's what they're doing and why.
Build for agents
For most of SEO's history, agencies have optimized the content layer: the stuff that gets crawled, ranked, summarized, and cited. Agentic search adds an action layer on top of that.
"The action layer is the set of things an agent can actually do on your site. Book a table. Request a quote. Add to basket. Complete a form." - Simon Schneiders, founder and CEO at Blue Array
The problem is that most websites aren't built for the agentic web. Transactional actions sit behind JavaScript, multi-step UIs, and interfaces designed for humans, not agents. AI crawlers like ChatGPT can't render JavaScript, which means anything hidden behind it is effectively invisible to them. Many agencies are now working to optimize websites for the agentic web.
The most credible solution Schneiders has seen is WebMCP, a new browser standard backed by Google and Microsoft that lets a website declare its actions in a structured, discoverable way.
"Think of it as schema for actions rather than schema for content," he says. "An agent won't need to wrestle with your front end to book a meeting room or request a quote. It will call the action directly."
Garrett Sussman, Director of Marketing at iPullRank, adds that agencies need to think carefully before jumping in. "Is it in the best interest of your client? What are the opportunities and risks?"
Some clients may have technical debt or legal hurdles to clear before any agentic readiness work can begin to optimize for new agentic protocols.
For eCommerce clients specifically, Google's Universal Commerce Protocol (UCP) for agentic shopping, is also worth understanding. It creates a common language between agents and merchant backends to power discovery, checkout, and post-purchase flows.
Schneiders' advice is to start small: "Pick one or two high-value flows to pilot now, like the bookings, quote requests, or checkouts where friction costs the most and agent compatibility has the clearest upside."
Measure differently
Traditional SEO reporting had a clear model: rankings, organic traffic, landing page performance, revenue. Agentic search breaks that.
"An AI agent may research a product, compare options, extract information from a website, and complete an action on the user's behalf. Some of that journey may happen without a clearly attributable human session." - Suganthan Mohanadasan, co-founder at Snippet Digital
Mohanadasan says agencies will need to triangulate across several signals GEO KPIs and metrics:
AI referral traffic
Server log analysis to identify agent activity
Citation and brand mention tracking across AI platforms
Assisted conversions
Actions completed through agent-facing interfaces like checkout or booking flows
"None of these signals will tell the full story on their own,” Mohanadasan says. “But when you combine them, you can build a much more realistic picture of how AI visibility contributes to discovery, consideration, and revenue.”
Sussman adds that the content itself needs to work harder, too.
"Agents may evaluate brands based on a specific user's needs, preferences, constraints, and intent. Generic content may have less influence over time." - Garrett Sussman, Director of Marketing at iPullRank
Agencies need to shift from broad campaigns built for a general audience toward content built for narrow, specific use cases, content that gives an agent enough confidence to recommend a brand for a particular persona or situation.
Know your client
Not every client needs to move at the same pace. "The opportunity will look very different by business model, audience, and the role agents may play across discovery, consideration, comparison, and conversion," says Sussman.
For eCommerce, fintech, hospitality, and complex B2B clients, preparing now makes sense. For others, strengthening the basics first—like content quality, technical accessibility, conversion paths—is the better move.
Mohanadasan is equally measured: "The key is education,” he says. “Walk clients through what each signal actually shows and why it matters, and pair the data with concrete wins, no matter how small."
That might be a citation in an AI answer, or the first booking completed through an agent-facing flow. Small proof points matter when clients are being asked to accept a messier, less linear picture of success.
Schneiders brings it back to priorities: "The bulk of effort should still go into AI visibility. That is where the near-term revenue sits." Agentic readiness is worth piloting, but not at the expense of the foundations that determine whether an agent finds your client's site in the first place.
Prepare for agentic search, but don’t forget the basics
Agentic search will reshape how users discover, evaluate, and buy products. But it hasn't replaced traditional search. There are still 3.5 traditional search users for every AI search user. The agencies that will come out ahead are not the ones chasing every new development, nor the ones ignoring it. They're the ones making deliberate, well-scoped bets while keeping the foundations strong.
The practical starting point: educate your clients on what's changing, audit their highest-value commercial flows for agent-readiness, and pilot one flow before scaling anything else. Every move should connect back to a measurable business outcome, because that's still what clients are paying for.
