In this special Google I/O 2026 recap, Constance Chen shares firsthand insights from inside the event, breaking down what AI Mode, Ask YouTube, Universal Commerce Protocol (UCP), and WebMCP mean for the future of SEO, ecommerce, and digital marketing.
https://page2pod.com - Google I/O 2026 may have marked the biggest transformation in Google Search in over 25 years. In this episode of the Page Two Podcast, Jon Clark, Joe DeVita, and Constance Chen break down the most important announcements from Google I/O and what they mean for marketers, SEO professionals, ecommerce brands, and content creators.
Constance attended Google I/O in person and shares firsthand insights into Google's evolving AI ecosystem, including AI Mode, AI Overviews, Ask YouTube, Universal Commerce Protocol (UCP), Universal Cart, WebMCP, AI agents, and the future of agentic search.
The discussion explores how user behavior is changing, why traditional SEO metrics may become less important, and what businesses should be doing over the next 30 days, 90 days, and beyond to remain visible in AI-powered search experiences.
🤖 In This Episode
• Why Google is positioning AI as the future of Search rather than a search feature
• Key takeaways from Google I/O 2026 and the evolution of AI Mode
• How AI Overviews and AI Mode are merging into a seamless search experience
• What Google's redesigned search bar reveals about the future of search behavior
• How personalization and multimodal search are changing discovery
• What AI Search Agents mean for publishers, marketers, and consumers
• Why Ask YouTube could become a major visibility opportunity for brands and creators
• Understanding Universal Commerce Protocol (UCP) and Universal Cart
• How AI agents may soon complete purchases on behalf of users
• What WebMCP means for websites preparing for the agentic web
• Actionable recommendations for the next 30, 90, and 180 days
• Why video content is becoming increasingly important for AI visibility
• The future of ecommerce, content marketing, and Answer Engine Optimization (AEO)
• Predictions for Google I/O 2027 and the next wave of AI-driven search innovation
Whether you're an SEO professional, ecommerce leader, content marketer, or business owner, this episode provides practical insights into how Google's AI-first future is reshaping visibility, traffic, and online growth.
🔔 Subscribe for more expert discussions on SEO, AI search, digital marketing, and answer engine optimization.
💬 Comment below: Which Google I/O 2026 announcement do you think will have the biggest impact on your business—AI Mode, Ask YouTube, UCP, or WebMCP?
🛠️ Tools & Resources Mentioned
• Google AI Mode
• Google AI Overviews
• Ask YouTube
• Ask Maps
• Universal Commerce Protocol (UCP)
• Universal Cart
• WebMCP
• AI Search Agents
Jon Clark (00:00)
Welcome to the Page Two Podcast, where we uncover the strategies, systems, and tactical decisions that move brands beyond page two and into real visibility across search and answer engines. A few days ago, Google walked out on stage at Shoreline and did something it's been inching toward for three years but never quite said out loud. It stopped treating AI as a feature bolted onto search and started treating search as AI. Liz Reid called it the biggest change to the search box in 25 years. TechCrunch ran the headline: Google search as you know it is over.
And whether you think that's hype or prophecy, the numbers underneath it are hard to argue with. AI mode crossed a billion users in 12 months, and depending on whose data you trust, somewhere between a third and the overwhelming majority of searches now end without anyone clicking through to a website. So that's the question hanging over this whole industry right now. If Google increasingly answers the question itself, and now books, shops, and acts on your behalf, what exactly is left for the rest of us to optimize for? Is this the end of the open web or just the end of a particular way of measuring?
Here's why we're excited about today's episode. Most of the takes you've read this week came from people watching a livestream and refreshing Twitter. Our guest was actually there. Constance Chen, SEO director at Moving Traffic Media, was on the ground at I/O in person. She breaks down the evolution of AI mode and AI overviews, the biggest redesign of the Google search bar in more than 25 years, AI search agents, Ask YouTube, UCP, WebMCP, and the signals Google is sending about how websites will interact with AI agents moving forward.
Jon Clark (01:34)
Welcome to a special episode of the Page Two Podcast. I'm your host, Jon Clark, and I'm joined as always by my partner at Moving Traffic Media, Joe DeVita. Today we're excited to welcome back Constance Chen. She's our SEO director here at Moving Traffic Media. She was on the ground at Google I/O in person, in the keynote, in the demo booth, in the hallways, and part of all those conversations that don't make the press releases. She's going to walk us through some of the key announcements that are real, what felt maybe a little bit rushed, and what actually means for the day-to-day work that we do for every client. So let's get right into it.
Joe DeVita (02:07)
Connie, you're a lifelong resident of California. You live pretty close to the Google headquarters. Tell us what it was like getting into that amphitheater last week.
Constance Chen (02:17)
Yeah, it was, well, one, I think we had some pretty unusual weather because it was super warm. I think that was a common complaint for most of everyone. But getting into the amphitheater is always super chaotic because you have tons of traffic, and then you're just trying to weave in and out of crowds and crowds of people trying to get a good seat, trying to get in the shade before things get super hot. I think just overall, the environment was buzzing as always because people are really excited about some of the new announcements that they're going to make. But what I will say is that there was a very distinct difference in how last year went and how this year went. And I say that because last year there were a lot of those new announcements that blew everyone away. And I'm talking about things like AI mode and Gemini Live and deep research, all of which changed a lot of the search industry.
Last year I walked away from the amphitheater thinking, how are they going to top that next year? And this year there's still that lingering question: how are they going to top that? Because it was such a dramatic change. And I know even in the SEO space, there were some rumors and anxiety, a lot of speculation that AI mode was going to become the new default search mode, kind of like what they did with AI overviews, but that it would obliterate clicks completely. And I think there was also some debate whether or not that even made sense for Google's business model. So when the announcements dropped around AI mode and AI overviews this year, I think there was that sense of a collective sigh of relief in a way.
But I would say Google didn't really surprise us with anything too dramatically different from what we've already been experiencing with AI mode and AI overviews. I don't think, though, that the search industry should assume that it's in the clear or that you can kind of stay here and coast for a bit. I think if you're not keeping up with how things are evolving in the product and development spaces, it's going to get really easy to get stuck on the details of fluctuating metrics. And if you stay there, there's a sense that you're going to be so far behind in trying to keep up because the train has already left the station. They're going in a very different direction. And Google and other companies are creating these products that are shaping user behavior, which is also going to shape how the marketing landscape is going to look in the coming years. And what I think the marketing world needs to do is expect more change. And one thing I want to note is that Demis Hassabis, the CEO of DeepMind, was talking on a keynote in one of the villages, and he said that we're only a few years away from AGI. AGI, if you don't know, is artificial general intelligence. And this is the idea that AI can execute tasks the way that humans can and can even do it better. So he said that we're going to see that happen. And I think just in that sense, if that is a general direction that things are moving in, then search marketers will really need to be thinking forward and aligning themselves with how user behavior is inevitably going to change. So I think based on what they shared at I/O this year, I think if there's one thing that I really want marketers to take away, it's that the agentic search era is very much on the horizon and the web ecosystem is changing dramatically underneath our feet.
Everyone used to be on WordPress, and I know we're still seeing changes with that now. People are migrating away from WordPress or going to brands like Webflow and other platforms, but we're also in the day and age where people are starting to use coding agents to build their own websites or customize their own websites. And I think with some of the latest announcements from Google I/O this year, how we build websites, how we market online, and then who we end up marketing to is all going to change quite a bit.
Jon Clark (06:28)
Well, maybe we can talk about some of those behavior changes. I think Sundar Pichai, when he sort of first came out, dropped a bunch of really impressive numbers. AI mode passed over a billion monthly users in the last 12 months. AI overviews hit two and a half billion monthly active users. I thought something else that was pretty interesting was that AI mode queries are, I think, three times longer than traditional search queries. And we've sort of seen this progressive lengthening of the head terms into real conversational types of queries, things that are a little bit maybe harder to put search volume around. Follow-up queries are up 40% month over month. So there's this definite, clear change in behavior that Google is pushing folks towards, certainly with ChatGPT and other LLMs, just the way that they've restructured their search content entry experience. But maybe take us into AI mode and AI overviews specifically. Like, what did you take away as it results to the layouts and the way that those two components are really going to work together for users in the future?
Constance Chen (07:38)
Yeah, I think one report that they shared was just a report they shared about AI mode on some of the data that they've been seeing. They talked about how AI mode is being used to plan and make decisions. People are using it to research products, they're building guides or planning trips and things like that. They're even booking restaurants. And some of the industries where they're shopping or conducting more research around when they're doing their shopping include electronics, apparel, health and beauty, and automotive. And then when you think about how people are booking restaurants, some of those questions that are coming up are things like: is a restaurant kid friendly? Does it have a view? Does it have a bar? So I think a lot of this data, even from that report, should really inform how we're thinking about building content to reach users on AI mode. And when we think about what are some of the new things that are happening this year with AI mode search in general and with AI overviews, the layout has changed in a sense that they've kind of hybridized the two experiences. So instead of AI mode becoming the default search mode, AI overviews will still be there.
Users can still ask these follow-up questions when they're making their search in AI overviews, but they can also go back and forth with AI mode. And it becomes this seamless experience, as Google likes to call it. They really wanted to emphasize that your context stays the same when you search on AI overviews, and then you can continue to explore and learn on that same query, and it starts to become and morph into AI mode where you have this longer conversation and you can dig in deeper. So this is now becoming the default search experience for all users. This is worldwide across all devices. And I think if you think about how this is going to impact websites and search and how people are interacting with search, I think as marketers, it helps you double down on what we already know, which is we need to be building very niche content to reach specific users who are in that research phase. They're starting to ask these follow-up questions. And you can kind of go into this rabbit hole where you can go in very different subtopics and themes underneath this larger query that you have. But I think with the increases in searches using voice and images that they had talked about in their data report, you can't keep relying on text content to be the only way that you're going to be able to reach users. I think in the past we've heavily relied on blogs to reach people, but you also have to now think about visual matches and images and even video, which I think we'll dive into later. But I think the biggest fundamental shift is in how the traditional traffic funnel has mutated. Because I think we used to think of search as more of a linear path where someone can just type in a keyword and then they get that list of 10 blue links and then they click one and they just stay on the website. I think now with AI overviews and how it's morphing into AI mode, it becomes this launching pad where you can have these really engaged back-and-forth conversations and users can just be taken deeper into these rabbit holes without even having to leave Google or go onto different websites. And I think for people who are publishers, you can't just be thinking about ranking for a keyword anymore. You have to be a trusted reference that meets users where they are in this research journey.
So that's less content focused on the what is questions because AI overviews is going to answer that for the users. But it's maybe thinking a little bit further down the funnel where we're thinking about comparative content like brand A versus brand B. Because according to the data and according to some of what Google is talking about, AI mode is being used as a way to help people research, help people make decisions, and help people with productivity. So I think we'll need to really revamp the way we think about building content.
Joe DeVita (11:54)
I think we should stick on AI mode for a little bit longer. This is kind of like a monumental shift in our industry. It was just launched a year ago, right? When you were at I/O last year, they announced it and it created this huge amount of hysteria. But we're a year into it now. The usage has, I don't know if it's quadrupled, but I think I read it has doubled every three months. People have adopted. If you go to Google.com, you get your normal looking search box. Then on the right side, you can click to use AI mode. And a lot of people are doing it. So the old behavior we've all had was you go to Google search, you type what you're looking for, you get your blue links, and you've got to figure out what to do with those blue links. If you don't see what you need in that first page, you probably just search again, like maybe I searched wrong or I used the wrong keywords. With AI mode, you never see those blue links and you never really search again. You just follow up on your first search. It's a huge change in behavior. That's the big change for me. I didn't think it would happen this fast, honestly. As a marketer with a couple of decades in the industry, I kind of assumed behaviors changed slowly over time and you had time to plan for it. Not the case here with AI mode. The product is really so powerful. It's changing behaviors a lot quicker than I anticipated.
Constance Chen (13:19)
Yeah, and I think there is more to come because I think that same conversational mode is going to be rolled out in different product spaces. And we're going to talk about that just a little bit more, but they're bringing it to YouTube, they're going to bring it to a bunch of different things. So I'm excited to see where it goes, but it's also going to change how people interact with websites and with content in general. So it's going to be really interesting to see.
Jon Clark (13:47)
I think the other thing that's probably super interesting to think about is that Google's goal has always been this way, but it's to keep someone on one of their properties long enough that they eventually click an ad and make revenue. And in many ways, this is a sort of even deeper funnel into a conversation with a model that keeps you engaged, keeps you going deeper and deeper until eventually you find the answer to whatever it is you're looking for. And Google's hope is that ends up with a click into something, or we'll get into UCP and all that sort of stuff that they announced as well, but I'm assuming they're going to take a cut of that transaction if that's where the model ends up. So I think that is really incredible to think about from just Google's planning perspective. They were behind in this AI race, supposedly, when everything first launched. And it's really incredible what they've done over the last two years to really not only catch up, but it seems like they're going to only increase their foothold in all of this experience and generate revenue. I think that's the thing they still have to figure out. And maybe why coming out of Google I/O, some people didn't have as much apprehension as maybe last year, because all of this stuff isn't default necessarily just yet. And I really think they do have to figure out that revenue model before it becomes the 100% default. It's coming, but everything comes back to revenue. Let's talk about the search bar. I mean, that I guess will probably be the conduit into AI mode in the future. Liz Reid, so she's the global head and vice president of Google Search, she basically called it the biggest upgrade since it debuted over 25 years ago. So take us into that. What was the big change or changes that were really talked about as far as the search bar?
Constance Chen (15:36)
So a couple of different tweaks, some of which I think are like, okay, that's cool, and others are actually very cool. So I think the new search bar is dynamic. It can expand based on how long your query gets. That's one of the, okay, that's not a big deal. But I think another feature that I think is really interesting is that the search bar now has AI to anticipate your intent. It can also auto-suggest and complete your search. And it'll help you refine it or add to it. So you could be looking, I think this is an example they gave, but you could be looking for yoga classes and then the AI suggestion would also add "looking for yoga classes near me on Tuesday nights around 6:00 to 8:00 PM." So I think it gives you a little bit more of a nuanced search and more specific. I think one feature that I'm really excited about as well is that you're also able to search across different modalities. Now, this has, people have been searching with images for a while, but I think they've kind of expanded this into something different. So this is something that I think is also going to change how people search.
But you can also upload files now, you can upload images, and you can also search using your Chrome tabs. So I think what's really interesting about this feature is not only can you do it with these different types of modalities, but you can also upload multiple things at the same time. So not only can you search with a PDF file, you can also add an image, you can also add your Chrome tab for reference. So that's a really interesting feature that I'm looking forward to playing with.
Joe DeVita (17:21)
Is that not available now to add the Chrome tab to the search box?
Constance Chen (17:25)
It is available now. So I was able to play with it in just the last day. And you'll get different results obviously depending on what is on your Chrome tab, but that is available right now.
Joe DeVita (17:36)
How would you use it?
Constance Chen (17:37)
So I think when you initially add in your search, there's a little plus sign, kind of like how you can click on it and there's a dropdown. And you'll see that you have the options to checkmark your tabs and you can include which tabs you want to include in search. So I think that's something that you can play around with and see what types of results that you want. But I think that was such an interesting feature that they decided to include.
Joe DeVita (18:04)
So if you've got a bunch of Chrome tabs open for different websites that you've been reading, and then you want to ask a search and say, like, "Take all these websites into consideration when you give me the next answer," that's kind of how you use it? Like, use these sites as a reference that I already have open?
Constance Chen (18:22)
Exactly. And what I've seen with some of my results is that it ends up turning into a conversation with AI mode simply because it generates more of that thought process to understand what's on that page and then helps you be able to generate some sort of summary.
Jon Clark (18:36)
You can really start to see how Google's pulling in the personalization and encouraging users to give them more access. So again, imagine they had access to your calendar and your Gmail, and you did that search for yoga. It could pull from your calendar and say, "Okay, he or she is open at 5:00 PM on a Tuesday." So I'm going to add that to the query. I mean, the personalization future is going to be wild.
Joe DeVita (19:02)
Well, I did play with that personal intelligence feature just yesterday. So you have the option, you go to Google search and you can opt into my calendar, you can opt into my email. And I opted into my calendar and I was like, "When is the first meeting I ever had with this person?" And it was like, boom. It was a lot easier than searching within the calendar. So I think that personal intelligence search tool is going to be pretty powerful.
Constance Chen (19:28)
Yeah, and that was just one theme that we saw at Google I/O. They were trying to really touch across all of their platforms in different ways. We won't be able to talk about it today, but Gemini Spark is one of those products that touch on personalization. It's supposed to be your personal assistant in helping you manage your digital ecosystem.
Joe DeVita (19:53)
Like a hundred dollars a month though, isn't it like really expensive?
Constance Chen (19:55)
Yes, so it's only for subscribers, but you can ask things like, "Hey, go into my email, read these 10 emails, create a summary for me." And you can basically create some agents to help you execute a bunch of different tasks, all from all of your different products that you're using within Google. So I think they're going in that direction.
Joe DeVita (20:22)
Jon, maybe this is the evolution from less of advertising clicks and more of a subscription model. Like maybe that's the slow evolution.
Jon Clark (20:30)
Totally. Yeah, I mean, you can see it baked throughout their presentation. The underlying tone is all these different revenue components. So if they need to give up revenue on the paid side in terms of ads and AI mode or what have you, that subscription revenue might be the replacement for it. Yeah. I was thinking about our conversation with Darren Shaw when you asked him about images with content or text in the image or maybe product placement. And I was thinking more and more about just the future need for every component of your website or your digital footprint to be thought about from an LLM's perspective. Whether it's AI mode or something else where, again, if someone's just grabbing an image and uploading it to the search bar, how can you give more context within the image so that you get a better opportunity to be part of the rest of the conversation, right? Yeah, it's really pretty crazy to think it's not just content anymore for sure. But you mentioned agents. Maybe we jump right into that. AI search agents. Again, you could imagine a situation where you grab a tab from your Chrome browser, drop it into the search bar, and then say, "Hey AI agent, monitor this page for changes" or something like that. But for context, can you talk to us a little bit about what that means for the future? What is it? How do we apply what's new there?
Constance Chen (21:57)
Yeah, so I think this is only available to subscribers as well, those who have Gemini Pro and I think Gemini Ultra. But it's supposed to roll out this summer. And I think what's really interesting about this is with search information agents, you can create and basically customize your own agent to run in the background. It runs in the background 24/7. So if you want to stay updated on a topic, like "Hey, can you keep tabs on this particular stock for me?" or a task that really matters to you, you can have the AI agent keep running in the background and keep tabs on it for you. So you could also have an agent maybe scour the entire web, like go to different websites or blogs, social media, just so you can get updates on either news or something that you want to keep track of, something related to your specific question, or even if it's something that you're trying to get an update on an item, "Hey, tell me if this product is going on sale." You can ask the agent to take action when that happens. So you can have multiple search agents running at the same time to basically work for you, get updates on certain things, and execute tasks so that you can either shop or buy a product when it goes on sale. But you can imagine how this is going to change the way that people are either interacting with websites or interacting with information. I know this is just for paid subscribers, so this isn't going to have a huge impact in search just yet, but the potential is there. I think users can create their own agents just to monitor something for them. And they can close their laptops, walk away, and just let the agents do the work. So I think anything that is top-of-the-funnel and gets top-of-the-funnel traffic because you provide financial market updates or things like that, or your website relies on users coming back multiple times a week in order to get that traffic, I think that habit could be very easily outsourced to an AI agent now, if you think about it.
Jon Clark (24:06)
I thought one thing that jumped out to me as far as this was I can foresee a lot of people using it for price-based monitoring. And just the need for having an updated product feed seems to really rise to the top as it relates to this because I would expect that's where Google will pull a lot of that information. Obviously, the websites themselves as well, but if they have access to those feeds, we know they use them in organic shopping results. So yeah, that one's going to be interesting. This next one I'm super interested in: Ask YouTube. The demos were pretty wild, but I'll let you explain what it is and then we'll get into some thoughts there.
Constance Chen (24:43)
Yeah, so this is currently available to premium subscribers on YouTube, but I believe it will roll out and be available to everyone later on. So you can expect this to happen. But if you have YouTube Premium, you can go ahead and try it now. I was playing with it over the weekend and it's really incredible. It blew my mind. I think this is one of the top announcements that they made this year that I think is going to dramatically change how people search on YouTube and how people search for video content. So Ask YouTube is very similar to what they've rolled out for Ask Maps. Ask Maps was launched just a couple of months ago. And then it's also very similar to AI mode. So they have combined AI features with their search capabilities within YouTube. You can ask really specific, nuanced questions within YouTube and you'll get back a series of suggested videos that answers your questions. And then you'll even be taken to the exact timestamp within the video of where the answer is. Over the weekend, I was playing with this feature. I asked, "What are the best running shoes under $100 for beginners?" So, what's really interesting about the results they gave me, it basically feels like they've laid out this interactive guide for you. There are very clear recommendations for the shoes that they're recommending. They show the video and then the exact model and name of the shoes. What's also really interesting is that they gave me the suggestions of the running shoes, but a lot of these videos are from smaller content creators. They weren't videos from video companies or large names that I would typically expect. But I think if you're in a rush and you were just wanting a direct answer, they're able to feed that information to you right away. So they might give the exact shoe model, and then once you have your answer, you could walk away. Or if you want to continue exploring, you can watch a video and see the reasonings why that model was recommended to you. So the other part to this is below that, they break out more video options under different categories. So they'll show you, "Here's videos that do budget shoe comparisons," and then beneath that, "Here's a table that compares the shoes that were suggested within the video." And then if you continue scrolling down further, there are sections where they'll show videos that are from expert beginner guides. And then I think, again, just to really highlight what's really interesting about the results that they gave me was that there were such a variety of content creators being featured. It's not always big brand names or even the videos that have the largest number of views. So I think this is going to significantly change visibility for content creators in a lot of different ways. And what I suspect is this could be one of the best features about this new search because we're going back to this place where it's not always going to favor these videos with millions of views already or those who are already ranking at the top, but it opens up the space and gives opportunities to smaller content creators. Or people who may not even be a regular content creator or doing that as a job, they might just be providing answers. So I think as long as you're creating content that provides the information or answers that people are looking for, these are answers to real-world, everyday types of questions or problems for users. It really comes down to content quality and understanding what types of niche topics you can cover in your content for a very specific user who might be looking for that question or query.
Jon Clark (28:30)
Yeah, I thought it was super interesting how they took multiple videos and sort of pieced components of them together to provide a longer-form answer to the query. I thought that was super interesting. And I think we've been preaching to our clients how important video is for a while now. I think this sort of elevates that a little bit more. And just the way that they're going to be included more and more in AI mode. I think you just can't sleep on video at this point.
Constance Chen (28:55)
Yeah, I think YouTube is going to be, I would say, one of the biggest game-changers for smaller brands because it's ground that you can still potentially win if you can create good content and answer questions that people are asking. I think with search, you're trying to build reputation, you're trying to earn trust from Google and users and AI. But with YouTube, I think there's still this opportunity to get in front of users even if you may not necessarily be able to reach them if your content isn't ranking on search as well.
Joe DeVita (29:27)
From a client's perspective, if you want to encourage a client to create more video, hopefully for that exposure their company, their brand, or their product gets exposed in one of these complex searches. The videos that are recommended to you kind of start just at the point in the video where the information is relevant to your query, which it seems to me like the opportunity for a content creator gets a little bit smaller. There's probably a little less advertising revenue share to the small content creator. The opportunity seems a little bit bigger for the brands just to try to create video for exposure.
Constance Chen (30:06)
I think what would help for those who are thinking about maybe expanding their budget for video, it would be to start creating videos that address specific user FAQs. You want to build maybe a series of videos that guide your different audience personas through their specific user journeys. And to do some of this, you really have to go back to the drawing board of doing an analysis and thinking about who they are. Build those highly detailed audience personas where you can think about the different types of questions that they're going to be asking and the different types of answers that they're looking for. And so I think YouTube is an avenue where I think there's a lot of potential. So you'll want to optimize your channel and you'll want to make sure that it's ready for Ask YouTube.
Jon Clark (31:01)
I think it's incredible how quickly you can get to those types of questions now, also. Like if you're recording video calls, if you're transcribing phone calls, all that data now is so readily available for LLMs or other models to basically pick out those key topics. I mean, that's basically your video list right there. Then you have the transcripts from those. You can transform that into a blog post of some sort. Yeah, super interesting. This next one, I think, from Google's point of view is probably the most interesting one. So they've historically, compared to Amazon and other models, been a little bit further away from the actual shopping cart. So UCP allows them to become part of that flow into transactions. Take us through that a little bit. I'm really curious about the reaction in the room as well to this.
Joe DeVita (31:51)
Connie, if you could set us up even better with what is Universal Commerce Protocol. You gave a great presentation on model context protocol a few months ago. I wonder if you could just give a quick 30 seconds of what the new protocol means.
Constance Chen (32:08)
Yeah, so Universal Cart and UCP, I think these are going to have huge implications for Google shopping and e-commerce sites. This is rolling out in the summer for Search and Gemini, and then they're going to roll out to Gmail and YouTube. But UCP is this open standard protocol created by Google. And basically what this is, it allows AI agents and apps to discover products and then they can start a cart for you. They can complete transactions with merchants, all within this AI interface. So instead of forcing a human to go onto a website, click around, and purchase the product, basically what UCP allows is an AI assistant to go into a store on behalf of the user. They can read the inventory, they can create the shopping cart, then they can buy the item on behalf of the user. So keep in mind, when they came out with Universal Cart, this is Google's way of introducing AI and agentic features into shopping.
Joe DeVita (33:13)
Am I confusing commerce protocol with universal cart? UCP and the same thing?
Constance Chen (33:17)
No, these are specifically different things. UCP is the protocol and then Universal Cart is the product that Google is rolling out.
Joe DeVita (33:26)
Okay, that wasn't clear to me. I wonder if other people are confused by that too.
Constance Chen (33:31)
Yeah, so UCP is the protocol that you could do this with. And it's just a protocol that allows AI agents to start shopping on behalf of customers and users.
Joe DeVita (33:43)
So the Universal Commerce Protocol is something that a developer or a retailer is going to use. The Universal Cart is like the consumer product.
Constance Chen (33:53)
Yes. The Universal Cart is a little bit different. This is Google's way of introducing AI and agentic features into shopping. With Universal Cart, users can basically start adding items into their universal cart using Google Search. They can do this while they're chatting with Gemini, they can do this while they're watching YouTube, or when they're reading Gmail. So it's Google's way of allowing you to shop from different merchants and vendors while you're on the Google platform. So for shopping, you'll basically have this way of automatically tracking deals or price drops, you can understand price history, you can get back-in-stock alerts. So basically what you can do is, since they've already rolled out this feature with these major partner brands like Walmart, Target, Nike, and Sephora, if you are using Universal Cart, it basically allows you to shop from those multiple places at the same time. So let's say I have this running checklist of things I want to buy. I want to buy some lipsticks from Sephora, I want to buy contact solution from Walmart, and then I want to pick up some running shoes from Nike. I can add all of those things within my cart, and it's one cart, and it's all within the Google platform. And then I can use Google Pay or Google Wallet to check out for all of those items with just one button or one click. So in that sense, UCP is the protocol that allows this Universal Cart to operate. And I think what's really interesting about this feature as well, other than monitoring price drops and things like that, it gives you a lot of those alerts. When you're shopping, like let's say I'm shopping for electronics, your cart will also let you know if you have compatibility issues. So the example that was given was if you're building a custom PC, it could alert you if you're buying an incompatible component and then suggest alternatives to you while you're shopping. It's really interesting. And I think for people who are using Google Wallet or Google Pay, the Universal Cart will automatically scan or apply your payment perks or your loyalty points to whatever purchase that you're making. So all of these automation features are where they're making this shopping experience much more seamless and easy for the user. It's almost like now the checkout lane is following the user. They don't have to go to multiple stores or multiple websites to get something. They just need to be shopping at Google.
Jon Clark (36:29)
And again, you can imagine the personalization layer on top of that. Once they start getting more and more of this data related to your purchase history, the ad personalization in the future, you can see how impactful that's going to be.
Joe DeVita (36:42)
Connie, if you were going to give a client, one of our e-commerce clients, some general advice on how to be thinking about commerce protocol and the Universal Cart just for the next 30 days, give them something to think about, maybe some actions they could take or conversations they need to have with their leadership teams.
Constance Chen (37:00)
Yeah, I think if you're running an e-commerce store, you need to pay attention to what's happening right now because this is huge. Google is essentially moving users away from your website to shop. So I think this is entering this era where you're having distributed commerce rather than people coming onto your website to make those purchases or transactions. So I think right now Google has a waitlist where you can sign up to adopt UCP and you can enable agentic actions in order for you to show up on AI mode or Google Search and Gemini. I think all of this is a step in helping your website become more agent-ready. So I think these are all things you want to do. You want to sign up to get your website to adopt UCP and then also start thinking about something we'll talk about next, which is WebMCP. You'll want to do all of this earlier than your competitors because this will help you get ahead of the game. So I think anytime they're rolling out new features like this, you'll want to pay attention and see what you can do to start implementation.
Jon Clark (38:06)
Yeah, let's talk about WebMCP. I think this one's interesting as well. It's really pushing development to be very considerate of how AI and LLM crawlers access websites in general. So take us through this a little bit. What were the big takeaways from that conversation?
Constance Chen (38:23)
Yeah, so when I went to go see this talk, I knew the agentic web is really here. Because WebMCP is basically Google's way of ushering in the agentic web era. They are now creating and setting a proposed web standard. It's designed to be this bridge between websites and AI agents. So if you think about what you are doing now or where Google is going, you are now getting websites ready for AI assistants to come and visit your website and be able to take actions for other users. And previously, I think websites have only been built for human interactions or human eyeballs. And now we're getting into the stage where Google is starting to define what exactly needs to be in place in order to serve and interact with AI agents. So I think one example they gave was, let's say you're a hotel booking website. WebMCP is now this protocol that allows websites to explicitly tell the AI agents: here is what this form does, here's how to fill it out, here's how to book a hotel room. And what's really amazing about WebMCP is that developers who are building these websites can now give AI agents tools that can help them convert. So most of the time when AI agents come to your website, especially without WebMCP being implemented or any tools being implemented, when AI agents are trying to interact with websites, it's not always successful because agents can't always click or interact with certain elements. They have to scan the website. It takes a lot of resources and tokens to be able to do this. They have to figure out which elements are what, what's a button, what function it performs. But WebMCP shortcuts all of that. It removes a lot of that extra work. So it helps websites provide more structured data and context and information so that these agents, when they come onto your website, they can execute these tasks a lot faster and a lot better. They have up-to-date and real-time information. So for example, if you're that hotel booking website again, the agents that are visiting your website are not trying to book a room that doesn't have any availability, but instead those rooms are filtered out. So let's say if you're a developer working on that website, you can structure the information using JSON schema. And this is to provide information about the hotel details or the pricing and the amenities that are available. And it's supposed to structure that information in a very legible and helpful way to the AI agents. So I think you can build specific tools that AI agents can use and call on that will make that experience easier. So going back to that example, if that AI agent is coming to your hotel website to book a hotel room because they got asked to book a hotel room that has a skyline view and it should have a jacuzzi, with WebMCP, a developer can basically build a specific tool that filters out search results by amenities. So the AI agent can use this tool, and then they can find exactly what they're looking for and then just book the room. So I think this not only assists with conversions, but it also gives AI agents more tools to work with so that they can perform actions on your website.
Jon Clark (41:51)
This one's super interesting because I think the downstream effects are unique. So using that hotel as an example, a lot of those booking engines live off-site. So the ownership in some ways falls to that third party, not the website owner themselves. Same thing for, we work with a handful of restaurants, a lot of their reservations live off on Resy or something else. And so it'll be interesting to see who adopts this first and if they get sort of a leg up. Or do website owners decide to take that on as a development task and build it so that they know it is agentic-friendly? I think that's a really interesting space to monitor as we move forward here.
Joe DeVita (42:35)
I was just going to ask maybe a question to both of you guys as we talk about our own AEO practice optimizing for more of that traffic. The clients who adopt WebMCP faster, maybe they get a leg up on more discovery from agentic traffic.
Jon Clark (42:53)
Well, I think all of this is really forcing the conversation back to where the revenue is. And if more and more revenue is being generated by these models in the future, yeah, I think it has to be a consideration. And over time the default, right?
Constance Chen (43:09)
Yeah. And they gave us some different examples, not just with hotel and travel. They also showed us maybe a pizza store. You could do this with a pizza store and just being able to create these very specific tools for different companies or use cases. But I think this is a huge development because I think it's going to be a part of how search is going to become more agentic in the future. And it's such a big fundamental piece of infrastructure in terms of the shift from traditional search to agentic search. And I think what's really going to start to see is agents execute tasks on behalf of users. And especially if your site relies heavily on conversions, I think you're just going to want to consider what types of tools to create for these AI agents on your website, especially in accordance with reaching the users who may be looking for something specific.
Jon Clark (44:06)
Yeah, I think the reason I find this so interesting is that we're talking about more transactions: ordering a pizza, booking a room, but even down to lead forms. We're working with a client right now who embeds Pardot forms in an iframe, which is not friendly for really anything. And so the future, do they choose to do that or do they build that lead form directly within their site where they can make sure that it's executable by WebMCP, for example? Yeah, it's just a lot of interesting conversations to be had in the future just around the tech stack for a lot of these companies and what it means for these third parties that have traditionally been the lead form generator, right? Like Pardot, HubSpot, Salesforce, all these guys. I imagine Salesforce will figure it out. I know they're doing a lot with AI and everything around MCP, but yeah, it's going to be interesting to watch.
Let's talk about I guess maybe some things that clients should be thinking about. I think we've touched on a couple of them. One thing that really I think makes a ton of sense, especially as it relates to MCP and UCP, is really identifying maybe the top 20 SKUs and making sure that those are LLM friendly. I think we're learning a lot around what sort of defines whether or not an LLM reads the rest of the page. So thinking about those key things that need to be toward the top. Anything around authorship, right? That expertise component. Original data is always great. So just again, rethinking what those product pages look like, like what's actually important there. What do we assume that the models already know? What do we want them to know? And making sure that all of that's very relevant on the page. And I think that translates down to the shopping feed, which we've talked about a little bit already. Just making sure that those are updated. We've seen some success with our clients where we make a strategic focus on building out those description fields and product attributes.
Joe DeVita (46:00)
Maybe we could talk about everything you learned in the last week. There's another huge dump of information we need to sift through, but how do we think about advice we want to give for the next 30 days versus the next 90 days versus maybe the next six months just to try to help people get themselves organized? We can talk about it in general or maybe we can talk about it e-commerce versus lead gen. What do you guys think?
Jon Clark (46:25)
You're the one.
Joe DeVita (46:25)
Yeah, all right. Well, let's talk about it in general, just for the diverse set of clients that we work with. What's the advice we want to give them: next 30 days, next 90 days, next six months?
Constance Chen (46:37)
I think from a fundamental standpoint, the first thing that they'll probably need to address is, if you haven't already, you need to make a decision about robots.txt and AI crawlers. If you're blocking AI crawlers at all, it's just going to make you invisible for the future. You're not going to appear in any way. So I don't think you should be implementing that unless you are a massive publisher with immense proprietary content. And in that case, you just need to figure out that balancing act. But I think working with your internal teams to have a very strategic decision about where you stand on AI crawlers and then moving forward, how are you going to appear in AI visibility? And I think in that sense, you need to understand what AI already perceives or thinks about your brand. So running an AI brand audit, thinking about how Gemini views your business today. Setting up manual trackers if you need to or using tools, and then maybe take about 10 to 20 of your highest intent buyer questions, query them, and then track them across different LLMs each week. So I think using that, you'll want to also think about how you can calculate your AI share of voice. But all of these things are just setting a fundamental baseline of where you're appearing in AI visibility and what your stance on AI crawlers is for your company, and then moving forward from there.
Joe DeVita (48:12)
So get the baseline in order in the next 30 days if you haven't done that already. Connie, you've given some really good advice around the next 30 days. Get that baseline in order. But what about the next 90 days? You've got to take on a little bit more than just getting an understanding of the baseline.
Jon Clark (48:26)
Well, I think one thing to be thinking about is on the measurement side. So there will be, probably already should be, a conversation with executives around what the future of data looks like. Right, traffic is decreasing. I think referrals are going to become more and more from LLMs, and what that means for site metrics is going to be impactful. We're already seeing it. So I think starting to have those conversations around resetting the measurement baseline, thinking about different metrics like AI citation share, assisted revenue from zero-click queries, schema validation scores, right? All those different things that maybe we're not talking about enough today are just going to be more and more impactful as we move forward. So really resetting how we're reporting and getting everyone comfortable with that. Little baby steps in terms of getting terminology and awareness, I think is going to be super important.
I mean, one thing maybe to add is just content. Again, I think a lot of folks are trying to figure out how we approach content in the future. Connie's touched on it a number of times, but thinking about what your content is actually made up of. It's not traditional text on a page. I think it has to be video, it has to be images, and really thinking about how all those things work together in an ecosystem. So restructuring content plans, maybe even topics as it relates to query fan outs and things like that, I think are going to be super important as well.
Joe DeVita (49:55)
Then what about the six-month plan where you have some time to really invest in something big for your company? What are the things you're thinking about for the next six months, maybe even 12? What are the big bets you make as a company to prepare for this?
Jon Clark (50:11)
I mean, here I think you have to start thinking about WebMCP, modern web architecture, if that's what you want to call it. And really deciding what needs to change on your website today to allow it to be more discoverable and more usable. I think that's the key, more usable by these models to execute transactions, to understand context and information. I don't know that you can necessarily re-architecture a site in six months, but I think you really have to start planning and thinking about it within the next 12 months for sure, as a lot of this protocol becomes more standard. And again, Joe, I think you talked about it at the top of the show, just how quickly adoption is happening. If we expect that for these shopping protocols and WebMCP, starting to plan for that sooner than later is definitely going to be useful.
Constance Chen (51:03)
And then I think if your team has the budget allocated for content creation with video, expanding into YouTube. And if you already have existing really popular blog traffic or blog content that receives a lot of traffic, taking those existing blogs, turning them into video, maybe creating a series of different videos or content that really targets that user base or audience that was responding to the blog content. I would take a look at and do an analysis of where your user base is going now, because I think six or nine months from now, we're going to start to see more changes in user behavior, I would imagine. And so I think it's taking a look at where your user is going and then trying to meet them in that space, trying to create content to meet them where they are. And that might be Ask YouTube.
Jon Clark (51:55)
Definitely agree. Maybe a wrap-up question. Based on your experience in 2026, what would maybe surprise you or not surprise you for Google I/O 2027? What do you expect to hear from them then?
Constance Chen (52:07)
Yeah, so I would not be surprised if we started to see more AI agents shopping and purchasing for users. I would expect to see that reflected in metrics soon. So I think that's something that people can look forward to. But I also expect to see websites fully adopt WebMCP in the coming year or at least move towards that direction. And I'm expecting that Google will start to build more in that direction where they're going to start equipping developers and website builders with these tools to be able to interact with AI agents. And I would not be shocked if people continue to say that SEO is dead next year.
Jon Clark (52:51)
It wouldn't be a new year without someone mentioning that, right? Yeah, I mean, I think it's going to be super interesting to stay on top of the Sephoras and vendors of the world who are part of this initial test and really see what the responses and outcomes are. This has been great. This has been a fun recap. It's nice to have the three of us on an episode for once. So thanks again for joining us on the Page Two Podcast. If you enjoyed the show, please remember to subscribe, rate, and review. We'll see you next time. Bye-bye.