In this powerhouse episode of the Page 2 Podcast, SEO strategist Kevin Indig returns to unpack the seismic shifts happening in search—where clicks are declining, AI overviews are taking center stage, and trust beats traditional rankings.
https://page2pod.com - In this episode of the Page 2 Podcast, Jon Clark and Joe DeVita welcome back Kevin Indig for a deep dive into the changing SEO landscape in 2025. From AI overviews reshaping Google search to new content strategies, Kevin shares actionable insights for brands navigating the future of organic growth.
Kevin breaks down his groundbreaking studies on e-commerce SERPs, usability in AI-driven search, and the evolution of trust in online content. Whether you’re a marketer, SEO pro, or brand leader, this episode gives you the frameworks and tactics you need to adapt in an AI-first search world.
Whether you're an SEO, marketer, or brand strategist, this is a must-watch to future-proof your search strategy as we head into 2025.
☕ In This Episode (Key Takeaways):
• Kevin’s coffee ritual and why rituals set the stage for great work
• The growth of Growth Memo and moving into video-first content
• Insights from analyzing 35,000+ queries on SERP features in e-commerce
• How AI overviews are replacing product grids in Google Search
• Why image carousels dominate and what that means for shopping discovery
• Strategies for Black Friday / Cyber Monday campaigns in an AI-driven SERP
• Key findings from Kevin’s LLM & chatbot study (Wikipedia, Reddit, LinkedIn, YouTube as sources)
• Should brands block AI crawlers? The economics of content in 2025
• Content simplicity, structure, and trust signals that matter most now
• Link building’s shifting role — mentions vs. backlinks in the AI search era
• New SEO KPIs: influence, share of voice, and brand trust over clicks
• Predictions on what Google Search will look like in the next 12–24 months
This episode arms you with the insights you need to rethink SEO strategy for the AI era.
✅ Subscribe to the Page 2 Podcast for more expert conversations on SEO, marketing, and growth.
💬 Comment below: How are you adapting your SEO strategy for AI-driven search?
🔗 Links & References from the Episode
• Kevin Indig on Linkedin
• Growth Memo Newsletter
• AI Overviews Usability Study
• Google E-commerce SERP Features 2025 vs. 2024
• LLMs and Content Structure
• Measurement Shifts
If Google's AI overviews are killing your organic clicks, you're not imagining it. But the real shift might be even bigger. Clicks are no longer the point. Kevin Indig is a longtime SEO advisor, strategist, and the creator of the popular growth Memo newsletter. He's worked with brands like Shopify, Meta, and Ramp. But now he's focused on helping companies adapt to a new search landscape where visibility doesn't always lead to traffic and trust Trump's ranking.
In this conversation, we dig into Kevin's latest research on how AI overviews are reshaping user behavior on Google and what that means for brands heading into Q4. We unpack why product grids are disappearing, how Reddit and YouTube are suddenly influencing chatbot results and why content simplicity and structure are now SEO power plays. Plus, Kevin shares what's working and what's not when it comes to LLM optimization AI-first content, and evolving SEO KPIs. As Kevin puts it, search is not just about clicks anymore. It's about answers and influence and making peace with a fractured marketing funnel. Kevin built his reputation optimizing for Google's rules.
Now he's advising brands on whether to block it altogether. That tension between visibility and value runs through this entire episode. If you're trying to navigate this next chapter of search, Kevin's already operating in that future. In this episode, he shows us how. Here we go.
Welcome to another exciting episode of the Page 2 Podcast. I'm Jon Clark and as always, joined by my partner in crime at Moving Traffic Media, Joe DeVita. Today we're very excited to welcome Kevin Indig to the show.
Just as excited to be on.
I think you have the best response so far. And you were a guest back in 2021, episode I think it was 72 or 73. And a lot has changed since then, both in your sort of personal career trajectory but also in the landscape of search. So excited to dig into a lot of that. But before we get into that, the first question, I know you're a fellow uh coffee connoisseur, what beans did you grind this morning?
Philtered Soul by Philz. Fills is a... Okay, I see some reactions here. Probably familiar. I spent a lot of time in the San Francisco Bay Area, fell in love with Philz. And look, I was born and raised in Germany, right? I know German coffee, which is not... It's okay, right? But Italian coffee, French coffee, uh very used to that. And then had my reservations about American coffee, but Philz really is different. And so it's pour-over coffee um I... try to import it to Germany where I live now. But yeah, it's tricky. Anyway, that's my favorite coffee and my morning ritual is to grind the beans, do the pour over, takes a bit of time. But the smell and the taste and everything is really activating my brain for work.
It's very therapeutic, the process of making coffee and those smells. I took the uh easy way out this morning.
Yeah, it's therapeutic and it's a ritual, you know there's this famous coffee ritual by the design team that designed the first iPhone. And they of course had this $30,000, $50,000 Italian coffee machine, but the team came together every morning and really took their time frothing the milk, you know grinding the beans, making the espresso, getting it right. And that set the stage and the tone for good work. And I try to mimic and embody that every day. And it helps.
I love that, love that. The Siphon coffee makers are always very beautiful to watch and very therapeutic as well. But enough about coffee. So growth memo, I think the last post that I read went out to over 21,000 subscribers and I went back and listened to the episode from 2021 and you sort of described yourself as not a writer, which I think I would disagree with, but probably also those subscribers as well. Do you still feel that way? You're putting out a lot of content.
Yeah, and by the way, congratulations for sticking to the podcast for that long. You know, you mentioned it, but I know firsthand what it's like to follow a project for that long. And I think it speaks to the passion. I think you cannot endure a project like that without really enjoying at least a big part of the process. To answer your question, I still, I obviously write. Maybe I'm a writer to degree.
And I still have my reservations, for example, about terms like content creator. I don't... I create content, but I don't see myself as a content creator. I see myself much more as somebody who reflects on what they see in the field. So I still work with a lot of clients. That's actually like 80 % of my time and money for the matter comes from working with companies directly. And so I see myself more as somebody who shares observations and experiences with others than a writer. And I say that by having the highest respect and admiration of writers and content generation in general. I think it's an art form. I know it's an art form. I don’t see myself as a full-time creator, much more somebody who shares experiences.
Yeah, that makes... It does put it in a different perspective. And I think when you're sharing perspectives or things that you're just observing, that act of writing becomes much easier, right? Because you're not tied to writing something. You're just sort of sharing experience. I love that. You had a goal of moving more into video content. Moving into 2025, right? You do these great wrap-ups of sort of what was successful, what wasn't, what was goals for the coming year. And just following you on LinkedIn, I have definitely seen more and more of that. How are you sort of pacing toward that goal? And I guess, how does that fit into the growth memo, which is more of like a newsletter style?
I would say I'm behind that goal, to be completely honest. I share more video. You know, I had some good traction, I would say a couple of months ago of creating YouTube videos rather than cutting out snippets from podcasts or webinars. And so, I'm behind on my goal. It was originally driven by the realization and observation of how important video has become, right? It's for many different perspectives, but for companies, as a, I would say brand building channel, as a medium for people to consume information. And so that tempted me to get more into video.
I still have that temptation and it's a completely different workflow than what I do right now. And so I've tried a couple of attempts to make my workflow video first. It's not yet exactly where I want it to be, but I'm iterating behind the scenes on different ways to get there. The thing is that I also spoke to a lot of my readers, and a lot of them loved the long form reading. I think that's part of the challenge, right? When you start a newsletter, then obviously you attract people who like to read.
And so if you then switch to video, it's a bit harder because that is against kind of the way that people adopted the product in the first place. I'm playing around with what that looks like. I've done a bunch of different things, you know, from like the videos that I have created with live streams from my premium audience to some of the YouTube content that I've created. And so I still want to get deeper and more into video, while also not abandoning people who like to read long form.
It's sort of interesting. I was taking some of the long form content, dropping it into Notebook LM in preparation for this and just sort of listening to it while I was playing with my daughter doing whatever else. So that's one way you can sort of, at least for me, take that long form and move it into a different medium that you can still digest without having to sit down and sort of absorb it.
But I wanted to maybe transition into one of those longer form pieces. You did a great study both in 2024 and then sort of compared that in 2025 across sort of the e-commerce landscape. And it's probably very timely sort of moving into PQ4, right? So maybe just for the listeners, get an overview of what that study was, and maybe a couple of the key insights that sort of came out of that. I found it really interesting and maybe not surprising just in terms of how much Google has been changing.
Thank you. I had an opportunity to look at 35,000 queries provided by SEO Clarity and focused on the search features that we're seeing for these queries specifically. So rather than just looking at who's dominating for these queries in classical organic search I was curious what do shopping ads, local packs, product grids, overviews, image carousels, etcetera, look like over time? And what I found is that I would probably say there are three bigger realizations. One, interestingly, AI overviews are starting to replace product grids. Now, I want to be cautious.
AI overviews are only showing up for fractions still, right? This is a small trend that is growing rapidly. But you can see that as AI overviews are getting more traction in the search results product grids are losing traction. And it's a very interesting shift, in my opinion, because... Google was very aggressive about showing product grids. When I did that analysis the first time in 2024 I think Google had already shown around 40-ish percent of keywords were showing product grids.
I mean, product grids are essentially a different experience of search compared to classical organic results. It's much more visual grid style way of exploring products. And it seems like a year later, AI is the dominating kind of vehicle for shopping results, is starting to become the dominating vehicle for shopping results. So that was interesting. Two, Google has dialed up the number of image carousels so far that this is not the most prominent search feature. So if you're thinking about how does Google augment the search results the most, it's with images.
And again, it kind of makes sense because it's very, very similar to product grids. The difference might be that when you click on an image you don't get to a product page but apart from that, shopping is visual, right? And you see that also on TikTok, on Instagram, a little bit on YouTube that these platforms are primary discovery drivers in shopping. And then maybe as a third surprise to me at least is that the degree of, or the amount of keywords that are showing a discussions and forums' search feature is decreasing. So we all know that Reddit is one of the largest sites now on the web. It's probably my data.
It's the second largest site on the web, showing up for all sorts of queries. And I noticed that counter to the trend, to the overall trend, in e-commerce the number of Reddit results, at least for the discussions and forums, search is actually going down. So it seems like that in that sense, Google still shows Reddit results, but not anymore. in this specific pack. And so it was interesting to me because my experience observation is that people really do want opinions from others when it comes to shopping meaning from others that are not affiliates or otherwise incentivized. But that seems to be going down for now. And maybe it's just a Google test and they ramp it back up. But that's the data as I see it today.
It's definitely interesting and it maybe forces folks to think differently when it comes to advertising and making Q4 pushes for Black Friday and Cyber Monday those sorts of dates. Do you have any thoughts on how your clients should approach something like that given how the SERP is changing and modifying.
Yes, the thing is that these kind of shopping holidays, I mean, they're stretching out more and more, right? There's a lot of data about how BFCM, Black Friday, Cyber Monday, starts earlier and earlier every year. And also last year and probably this year is also a bit tainted by the economic impact on shopping behavior. It seems that there's some price sensitivity. People are looking more for sales and discounts and promotions. I mean, it kind of makes sense the whole holiday is around promotions but this year and last year, for sure, people were a lot more sensitive to that than in the previous years. And so, my recommendation is to start early with a discount strategy and then try to stretch that over the holidays. So don't discount all your products on day one, but maybe you want to start in... let's say, maybe in October to discount a couple of products. And then maybe every week you have a different category or so or maybe a featured product that is being discounted as under the umbrella of BFCM.
You know, according to the data, promoting that across forums is probably less important now. Having good visuals for that is probably a little bit more important. So, one example is when you have product pages. And you typically have a couple of different product images. Maybe one of these product images, maybe the preferred one, the main one that people see first Maybe that has, like, a little discount, or promotion or BFCM label on it, right? So playing, playing and toying around with that a little bit. And then when it comes to AI overviews, it's just I'm pretty sure that the searches for products plus Black Friday, Cyber Monday or Black Friday or just Cyber Monday you know, these iterations, these BFCM related queries, I'm pretty sure that they're going to start growing at the end of September, maybe early October, maybe even earlier.
And the guidance that I give to my clients is to start monitoring where AI are showing up for these types of queries, what they look like. And if you have the corresponding content for that, and maybe even coverage on other sites, right? So, you definitely want to have at least one landing page for BFCM for every year. Not one that not the same one, but a new one every year. Maybe you can use some of your backlinks to all the ones to link to the new one, etcetera. So that’s some content strategy around BFCM to address AI overviews. Yeah, that's probably kind of my strategy in a nutshell for that.
Got it, I thought that one thing that was sort of interesting about that study was the, like you said the decrease in maybe discussions, forums, maybe people also ask and things like that, which if we take that to one of your other studies around sort of like LLMs and content structure is a little bit in conflict, right? Because one of the recommendations, um not just from you, but from others in terms of how LLMs are interpreting that data, right? like FAQs and question and answer formats and things like that make a lot of sense. So maybe we can jump to that study a little bit. So, you know, you looked across a number of different chatbots.
Maybe take the listeners through some of the key findings across each one of those, that really sort of stood out. There's definitely some similarities across all but, I think some of the nuances were really interesting to dig into.
Absolutely. So what's interesting is that it turns out that the dominant chatbots rely on a very different mix of sources when it comes to how they give their answers. So what it means is chatGTP, for example, relies very heavily on Wikipedia. Perplexity relies a lot more on Reddit. Google also relies a lot more on Reddit. Perplexity emulates Google a lot. very Google heavy. And then, you know, all the other chatbots from Copilot to Gemini, etc. They all have kind of their own mix of sources when it comes to the citations to give answers. Which in return means that if you want to optimize for visibility on these chatbots you might want to set a different emphasis based on which one is most important to you.
I would say most of my clients see the biggest returns from ChatGPT. That seems to be the dominant one. It's far ahead of anyone else. When we look at just the traffic going to ChatGPT which I think strongly kind of correlates probably with users. And then have Gemini growing the fastest. I think they're now in second place and then Claude, Grok, Copilot, Perplexity, and DeepSeek and all the others.
So, most people get the biggest returns from ChatGTP. And so what that means is... influencing Wikipedia is hard. It might be worth a second look just to make sure that as a company you're accurately represented, there's no information gap. If that's the case, I think it's more than a fair pitch to try to correct that. But again, the Wikipedia defenses are incredibly strong, very strong immune system that quickly rejects any attempt to tamper with the pages.
But it's worth a look and a try. After that, Reddit has become a lot more impactful when it comes to ChatGPT. I used in the beginning there was no visibility at all from Reddit. Now, that has jumped up overnight. I have to look at the dates again. I think it was maybe sometime in May this year.
And it could be due to the partnership that OpenAI has with Reddit that it just kind of turned the switch on. Now it seems like over 5%, maybe at this point more than citations come from Reddit actually. So that now becomes a viable optimization platform, but then there's also LinkedIn and YouTube. Both of these are, I think, maybe LinkedIn, not in the Black Hat community but I think in general they're a little bit overlooked as ways to influence AI answers, right? And so things that I'm pushing with my clients is for example, to create YouTube videos for specific prompts that were maybe not visible for And then LinkedIn, uh what's it called? LinkedIn Newsletters or LinkedIn Pulse, I think, is another very popular source with OpenAI.
And of course, I think some of that could be questionable because it is used for maybe some shady tactics but the reality is that these kind of social content platforms have an outsized impact on ChatGPT. And then when comes to software reviews, for example, we have players like G2 when it comes to health, we have Healthline. So every niche, of course, has its predominant citation source but overall, anything that is UGC and content-driven has an outsizing impact.
Yeah, I think it was Lily Ray who showed how easy it was to manipulate with a LinkedIn newsletter pulse or whatever they're called just to say like, who's the best SEO and that sort of thing. The other thing that I saw, which I think aligns really well with this study was someone totally escaping me. But basically, if you monitor perplexities like trending searches, and you can spin up a video really quickly, you'll basically own, you know, that citation.
So yeah, think the folks who want to, I don't know if spam is the right word but manipulate these engines quickly are finding ways to do that relatively easily at this point. I think one of the things that was really interesting from the study was who was not being included, Like CNET and some of these other really large players seem to be mostly related to just blocking the crawlers outright through robots.txt. I wanted to, we've had a couple of conversations with guests around this. Like what are your, you advise a lot of really big brands. Like what are your, you advise a lot of really big brands. What do you, or how do you advise them if they come to you and say We really don't want to give our content away for free. We're proposing just blocking everything. How do you tackle that conversation? And maybe just generally, what are your thoughts around companies that are doing that today?
I can very much empathize with that decision. I think the first step is always to do a simple ROI calculation. What are we getting today from AI chatbots, but also from Google? I think that decision is also on the table. And what are we getting? uh What are we giving? What is the cost to create the content? What's the risk profile or degree of risk today versus the next three years if these trends keep trending the way that they are right now. But then also, like, what could we get down the line?
I haven't published this data yet, but according to SimilarWeb, ChatGPT has been growing by over 700 % year over year. So when you compare 25 to 24, you see a 700 % increase in outgoing traffic. I should add that. And so that does not mean that you're traffic will grow by 7x or 8x necessarily. But it just means that these AI chatbots seem to be very interested in sending out more traffic. And again, new model is just out, ChatGPT 5.
We need to kind of keep monitoring that trend. The point that I'm making is you cannot just look at what you're getting now. You have to kind of extrapolate that in the future. Are we maybe getting to a significant degree of AI chatbot traffic that could offset some of the losses that we're seeing from Google Search Because reality is that despite what Google is saying right now I still want to see the site that is not losing traffic from AI overviews, I think every site is. And sure, you can offset that by just pumping out massive volumes of content. So, you don't necessarily lose traffic year over year in absolute terms but in relative terms, you absolutely are like what you would have gotten versus what you get today. You can make that simple analysis. So anyway, coming back to it, before making a decision, Cost benefit analysis. What are we giving? What are we getting? What would it look like in the next three to maybe five years?
And then I think you know, I Think it's more than viable and actually it makes sense for a lot of companies to block Google and AI chatbots. Just yesterday I saw a tweet from WalletHub saying that they're doing that now and yeah, and and so... I predicted that last year that we're going to see lot more companies that are going to exclude Google and ChatGPT and Co. because for them, it just doesn't make sense anymore. And at the front of that trend, we're seeing publishers. Makes perfect sense because they're critically dependent on getting clicks for ad traffic, for affiliate traffic.
Even if you have a subscription model, in most cases, it's not enough to sustain the company, at least in its current form. So publishers are at the forefront. Affiliates have already been hit really hard and decimated in 2023 by the helpful content updates. So if that wouldn't have been the case, then they would probably be staying right next to publishers. And so I think any content heavy model, there are also companies like Chegg or Stack Overflow that are also being absolutely obliterated by AI overviews.
And for these companies, I think it makes sense to say Okay, let's look at other traffic channels, other ways to grow the business and maybe adopt Cloudflare's paper crawl model. So Cloudflare made this really interesting, push into since they govern so much of the web and there's such an intermediary between users and search, etc. And they released this project called paper crawl where essentially you block bots, and you can be specific about which ones from calling and using your content for training data, or for RAG answers, for reasoning. Instead, they either need to ask for permission or they need to pay. Now, this is to be clear, right? Like you're not getting paid yet.
This is still an attempt to build a big enough pole for then the search engines and AI developers to come to the table and say, OK, this is going to be a model. It makes perfect sense to me, right? The economics of the web are deeply in the red for traffic. And so I'm not surprised to see more more companies not take on that deal anymore.
Yeah, I think that's the fundamental challenge of all of this, right? Is that there isn't a big enough pool of users or companies sort of demanding for X thing. Even when you do have a cohort of people, they may not be all asking for the same things. And so you still have this sort of divergence across, I don't know the effort to sort of rein this in, and be more cognizant about where they're getting traffic from. So I think you're exactly right. I thought another thing that was pretty interesting from the study was the simplicity of how the content should be created. You talk a lot about measuring like flash scores and just thinking about like simple sentence structures and things like that. I was curious If you wanted to analyze like how simple your content is at scale. Is that something that, like, a screaming frog, could help us evaluate from a content perspective? Or what sort of tools would you recommend your clients use to sort of evaluate that, maybe just to establish a baseline of where you are today?
Screaming frog is a great option because it allows you to crawl all of your content and then determine like the average flash score or complexity but there lots of tools from the SEMrush and Ahrefs of the world to keyword insights, Surfer, ClearScope... They all give you a sense of how complicated your content is to understand compared to what the other very visible articles are showing. I think that is one good start. I think then also understanding your topic gaps.
And again, there many ways to get there from building your own topic maps to using some of the tools that I mentioned. All of these could help with that. But then there are these kind of... I think it's like it's an evolution of what we've been doing in SEO for a while because the reality is that none of these recommendations are really net new. They're just sticking out more. They're just amplified by AI and AI visibility.
And so there's a lot to be said about the structure of the content, meaning the way you use headings. First of all, you should be using headings if you want to go the extra mile. You obviously want to have a clear hierarchy you want to have these headings be descriptive and related to the content. And I still, all the time, when I just look at headings see that most sites, they use them as a way to style the page or as a design tool rather than a semantic tool... Exactly, right? There's a lot to be said about active voice. There's a lot to be said about enhancements like table of contents, FAQ, key takeaways. I've seen that in my own usability study that we can talk a bit more about kind of the background context of that, but the people deliberately seek out these content elements as a way to skim and parse and anchor and essentially trying to find the answers, right? People don't have a lot of patience anymore. Maybe they never had, I don't know, but when they skim content, when they evaluate it they will look for these content elements so that they can say... Ah! okay, here's my answer now that means that content is interesting now let me read it top to bottom, right?
So it's this kind of like pyramid approach where you want to give the answer right away. So people just know, okay, my answer is here. it's worth for me spending time here. And then they might take time to read the rest of it. And all these content enhancements and structure and stuff also help significantly with that. And it helps bots who operate on a much simpler level to understand how this content relates to a larger topic. AI chatbots seem to be no different when it comes to that than search engines.
You've written a lot about these guiding principles for web design. You just mentioned a few of them. The ease of navigation. There's a lot of great examples like FAQs and summaries and highlights and stuff that you've gone into really great depth on. And I appreciate it.
Trust symbols is another piece that I think like you consider that a part of your website design. That's a guiding principle now product pages, anything. The one thing I heard you say that I hadn't really heard is the importance of page speed being less important than we all thought. I think John and I produce a lot of, you know technical recommendations and a big part of our tech audits is typically how to improve page speed. It seems like it has just naturally become less of a priority. We always find more important things to address first, but we never lose sight of page speed. It seems like a new idea for me a little bit, it being less important.
Yeah, you're right, Joe. Absolutely. It's a good observation. The way that I think about all of these recommendations and criteria is that it's very contextual. So most of what we talked about applies to blog content, right? But when we talk about maybe product or category pages, the focus can shift a little bit. And so in my experience with page speed and core web vitals I just group all of that together is that it's often, it gets a lot of attention because it's easy to measure, right? We get Search Console data, we can run tests and stuff. And so a lot of focus goes into that because we can see when the needle goes up or down. And I've had clients in, for example, the insurance space that have spent weeks on improving core web vitals with nothing to show for. And then I've had clients in e-commerce that have spent weeks improving core web vitals with a lot to show for.
And so for me, the takeaway here is page speed, core web vitals, etc, matters the most when we're talking about transactional pages, product, category, e-commerce stores, marketplaces, even right consumer marketplaces. That's when people have a transactional intent, page speed matters the most. And it scales specifically for e-commerce. It scales less when it comes to something like uh SaaS, right? So because you could argue, Oh, a SaaS landing page also has transaction intent, and that is true, but their page speed matters much less.
When you are shopping, on the other hand, in e-commerce, and there's a lot of competition and kind of low patience, right? If the page takes too long to load, you might just go to another store and buy a competitor product. And so that's where it's really critical. And I also see search engines reward that. And I'm not sure if search engines reward that because they measure page speed and the core vitals specifically or because they see that users prefer that page in the store because it loads faster. Now, there's one caveat to me, and that is time to first byte. Time to first byte or server response time, I think, is universally important for any site out there because it's so important that not just search engines, especially AI chatbots can crawl your site quickly.
And not the page speed number that impacts scroll rate or speed, the most is time to first byte. How long does it actually take the crawler to get the, I would say, raw HTML from your server. And so that is something that I'm keeping tabs on for all of my clients.
Interesting. Yeah, we've you know, in core web vitals first came out, it was like, oh my God, we have to audit everything. I think now we're starting to understand that it isn't so binary, right? Like it does matter based on who the competition is. So, one way that we've approached is looking at it on a template level comparing a handful of searches against who we're competing with on page one. And if we see like really big discrepancies then we'll focus on the page template itself, not universally looking at the site. So that's helped us be a little bit smarter about advising clients on spending a ton of time here where you may not see, you know, the, output of it, like you mentioned.
Smart.
You mentioned the UX study. Maybe we can transition to that. I think that, being one of the first studies I thought I'd got a lot of great attention and there was a lot of great analysis and effort behind it. I think that, probably the most interesting thing to me was that there are still about I think, I think it was 80 % of users are still clicking after seeing the AIO. I don't know, was there anything else that was super insightful that you didn't expect to see from that study? Maybe just take us through some of those things.
Man, so much. I would say it has significantly changed my view of SEO and maybe organic growth in general. For context, in March of this year together with a partner, named Eric Manbusker, he has experience with these types of usability studies, he has done them for Backlinko and others. We conducted the first usability study of AI overviews. I was very unsatisfied with... look, I got a lot of access to aggregate data, so maybe it's a bit unfair for me to say, but I had this, like, big picture view. The SimilarWeb analysis of like seven billion queries, like, you know, to understand what's going on in search results. Are people still clicking or not and zero clicks, all that kind of stuff. Incredibly helpful, but there was still a gap for me.
And so, luckily I met Eric on a webinar and we caught up and Eric was like, hey, we should do something together. We riffed and then we came to this idea of this usability study. And what we did is we monitored and tracked the user behavior of 70 participants out of the US across different age ranges, across different devices in solving eight tasks. We gave them eight tasks like find a tax accountant in your area, find this e-commerce product, find this health information.
And a lot of these but not all of those tasks were geared specifically for people to find AI overviews because we wanted to see how they engage with them without bias in them. So we didn't want to say, hey, if there's an AI overview, just engage with that. We wanted to just give them the task and then see how they naturally engage with it. And we got a ton of material out of it. I think it's an incredibly robust study. We had six people working on this. It cost a lot of money. It took a lot of time. But the insights, I think, are very profound.
And so the first one, you already mentioned it is that even though people click a lot less and this is also validated by a lot of different research. For example, the study by Pew Research recently, which was also incredibly robust showing that 1 % of users click on a source or citation in AI overviews.
That is just such a tiny number.
Yeah. It's nothing. It’s just nothing. And it can explain also why these 1 % click at all. But yeah, it just goes to show that Google Search is going away from clicks, right? I hope that's pretty clear for everyone listening, right? Especially when you look at or think about AI mode. But search is not about clicks anymore, right? It's much more about answers and influence.
But anyway, we saw that the clicks that are still going out, or let me put it this way. Clicks are a very noisy signal. A lot of clicks happen and we saw this in this study. put in the write-up of the study, I uploaded some clips of showing exactly that like how actual users behave and how they annotate and how they think. And so a lot of clicks are noisy because people click around to skim and scavenge. And a lot of clicks are not targeted and intentional. They're very impulsive. And people are searching and going back and forth and this and that.
What we found is that a lot of the noise and these impulse clicks are going away when AI overviews are present because people get answers to the questions right away. And as a result, the clicks that they still do are much more intentional. And they click for a couple of reasons. One is to validate information from AI. I'll come back to that in a second.
And two, they click to take action. And so the revelation for me from the study is that there's this thing called a final click or what we called a final click which is basically the last click that people perform when they complete their user journey. And what we found that you addressed is that 80 % of these clicks are going to organic results, not to AI overviews. And it's aligned with this study from Pew Research because... People don't really click on citations, but they still click on organic results, much, much less than before AI overviews. But they do so to take action, like transact, find the information they're looking for, find the tax account, and etc, all this stuff that I mentioned before. So there's still a lot of value in organic results, but we cannot measure that value with clicks anymore, at least not to the same degree that we did before.
Two, and I already mentioned this, people are validating AI results by going to platforms and communities like Reddit, YouTube, niche forums, LinkedIn, etc. So there is this new interesting user behavior that consists of trust and relevance. So it seems like before AI overviews, people predominantly judged the search results by relevance. Does it answer my question? Now, people first look at the... domain, the citation, the brand, and they ask themselves, do I know this? Do I know this and trust this? And when they do, then they apply the relevance check. So then they ask themselves, uh can this answer my question? And as a result, it means that, first of all, we need to be focusing much more on these other platforms because that's where people gain and build trust into brands and results. But second, also means that ideally we have a preconceived notion in the minds of our users before they search.
What does it mean? They should ideally have heard of us or know us and even better in a positive sentiment. And so some of that can come from people having visited our site previously, but you get into this chicken egg problem, right? Where it's like, you need trust for people to click on your site and you want to build trust when people are on your site. So where do you start? Right.
I think a lot of that goes into advertising and maybe collaborations and other ways to get in front of users. But that is becoming so much more important because if people don't trust and have heard of you, then you're mostly invisible, right? Unless we're talking about like utility utility, like they're just trying to find the cheapest product then these filters fall away much more. But a lot of other cases, they very much apply.
And the last thing I'll say is there's also a big age difference. I found that older users are much less receptive to AI answers. They're much more likely to skip them all together. Whereas younger users are very receptive and they don't look at the organic search results nearly as much as older cohorts. So there is that age part as well.
Can I, I wanna try to paraphrase, maybe summarize one of the big points that I heard. I've heard you say this in different ways with different people and you've written about it a little bit differently too. Just for that older cohort that doesn't trust the AI overview as much as the younger cohort, it seemed like your usability study showed that older cohort would scroll down to organic results to verify, maybe not clicking in to the organic results, but at least using that first set of organic rankings to build some trust in what they saw in the AI overview. And for me, that is a... That's an indication that rankings still matter. We've been talking a little bit about rankings mattering less and measuring search SEO success with clicks matters less, but it still feels like you want to rank well organically to build trust.
Yes, that is absolutely right. You want to rank organically to build trust. And you also want to rank organically so that Google pulls you into the AI overviews. And yes, people click a lot less. And at the same time, Google still plays around with the format a lot. They start to show a lot more links, fragments or parts of the answer. They highlight brands in the answer. So, you are correct. Organic ranks still matter.
What is changing and becoming more important also is this whole kind of idea of covering many user intents which directly relates or alludes to query fan out. So another big topic right now is query fan out, not just in the AI mode but also in ChatGTP and Perplexity all use query fan out technology. And in a nutshell, it's about these AI is trying to cover and include different user intents. Rankings are important and also clustering them by topic and by user intent filling these gaps even more important as well.
I want to dig into that in a second, but you mentioned this study really having an impact on sort of how you think about SEO. Is that was that purely from a traffic driving perspective, or were there actual tactics or I don't know, methodologies that you maybe decided weren't as important as what you saw in the past?
It’s a good question. I definitely got a new found love again for these, like content features because we saw so many people who you know they were looking for information they quickly go to a site like US News and they have this box with key takeaways and like oh yeah that's what I'm looking for right, so that kind of motion came up a lot trust and authority came up a lot and so for me it for example put into perspective the Google Quality Rater guidelines which also say that trust is the most important factor when it comes to EAT. just basically, like this study helped me understand what EAT is in general. It's just, it's really a framework for how users evaluate search results. And that's why Google wants to double kind of verify that with humans, or basically verify their search results with humans. But it was like, what's less important, it's hard to say, you know, I mean, backlinks that's always kind of a big question like how important are they still and I haven't come to a definitive answer especially now in this new AI search world.Speed was not really a factor in those studies so it's hard for me to pinpoint something that has become less important. I think it just re-ranked my set of things to pay attention to.
Yeah, you mentioned link building. How do you talk to clients about link building in today's world? Like, certainly you work with some brands that... I remember when Joe and I were at Razorfish, was always fun to work with those brands because you didn't always need to worry about link building.
They were just so strong and you could do something very simple and see a great impact. But you also work with a lot of companies who are just sort of starting, right? So they don't necessarily have that domain authority and link building must come up I would imagine, in your conversations when you're advising. So how do you talk to them about link building is link buying ever an option? Is it a PR push? Yeah, just curious about that.
It's still a big conversation point, because you're absolutely right. There are some companies that still need links. For example, on the local search level, that is still very much a thing. And there's a whole machinery behind it that still works like maybe 10 years ago or 20 years ago, where you buy links from the publishers or affiliate sites, and they do have an impact. But then, as you also said, there are these other large brands where link building is basically PR, and it's much more about being mentioned and the sentiment and the perception you create rather than hard backlinks. For me, I'm still investigating this much, much deeper, but it does seem like I mean, from the early data that I pulled, backlinks are much less important.
The volume of backlinks doesn't matter. The quality is still something that I'm circling around for AI search. And then it's all very contextual, right? I think you can see this in a lot of websites that were historically very strong when it comes to links I have the feeling that Google just devalued backlinks as a signal significantly. And the value of the content, the engagement, all that kind of stuff has taken a lot of that place. So the backlink conversations with my clients are a lot more around, again, the unique data PR, like how do we create narratives, support them with data, and then amplify them in a way that gets us in front of the audience and creates like a really good... the perception that we want. Yeah, there's still some questions.
And at the same time though, I think a lot of the focus switches from backlinks to mentions. And that is a slightly different game, but also a completely different game, right? You still have, like a lot of companies still do outreach for mentions. Not sure how successful that is, but big brands, they think much more about stuff like hey, should we give a site like NerdWallet or Forbes or others, should we give them a couple of 10,000s of dollars so that we get an advertorial or that we get featured by them and then we get a backlink and a mention or not? Those are much more the dimensions where, again, local players are like should I spend 500 bucks on getting 50 links across these like... local publishers, it's just very different dimensions. And again, the impact is also very different.
Well, the trade-off is using the NerdWallet example, right? If I could pay an amount and be visible like they've done all the work to rank for some of these really competitive terms if I can just pay and be there tomorrow, versus having to build out a plan, creating the content, right? Like this massive effort behind the scenes, especially for a smaller brand. You know, if you do that calculation, oftentimes it makes more sense to just be there. We worked with a bank who had a really strong, checking bonus offer and for what it would have taken them to rank for, you know the nerd wallet content would have been monumental. But they got into the affiliate program and they saw tons and tons of success from that for, you know, basically a fixed fee. So for them, it made a lot of sense.
I want to talk, we've talked a lot about shifting traffic and, you know, measurement. You had this great post on LinkedIn around like the new KPIs for AI driven search. Things like measuring how influential an impression would be, share a voice, those sorts of things. Brand recall and trust, which we've talked a lot about. How do you, again, in your sort of advisor role, how do you help your sort of client take these changing KPIs and sort of train the board or train the CEO on like how to think differently about what SEO's value is or what organic traffic's value is? I think this is the conversation Joe and I are starting to tease a little bit, and some of the reception is like, it makes total sense. Others are like, well, you know, I still want to know where I rank for X keyword. So I'd love to get a little bit of a sense on how you're, how you're thinking about that.
It's a very timely question because I have that conversation with almost all of my clients. The goal is to drive more towards leading indicators like ranks, visibility, share of voice then lagging indicators like impressions and revenue impact. But it's very challenging. It's very, very challenging because... It's a big pitch and leadership loses a lot of control because clicks were such a good intermediary with this classic search funnel. You had the search, click, convert. Makes perfect sense.
Now it's like search, read a lot, be influenced, and at some point come to the home page and sign up. It's a very disjointed broken funnel and accepting that is hard for leadership. It's a slow transition. And there are more leading indicators that I'm developing with my clients as a result to give a sense of control, right?. And look like, you know, there can also be... There's also a bigger focus on input. What are we actually doing to drive the needle, remove the needle? Are we meeting our targets in that sense? And then down the line, what's the downstream impact? What's the output? And also in lot of cases, it takes like a test or transition period because the reality is that things can still change very quickly. Don't forget, it's been about two and a half, a bit more, maybe two and three quarters years since ChatGPT 3 came out and changed a lot.
And then it's been about a year, maybe a bit more since AI overviews launched and the big impact really just came this year. So all of this is very fresh and new. And it's very hard to then say, OK, ah yesterday we saw this change. Today we're developing this new model of metrics. And tomorrow we're going to just stick to that.
Things are going to change again. And that creates lot of uncertainty and mistrust from leadership. We're doing things like, again, measuring some of these leading and lagging indicators. We're trying to get more qualitative data from brand surveys, from user, market research, user interviews, all these kind of things. And then of course the big question is always like what comes out on the other end like what's the bottom line and that's the anchor point for leadership.
That's what gives them more security, and shows them that as an as a team. We're not we don't want to move away from company impact, right? We're just changing the way that we're leading up to that. So again, challenging conversation takes out a pitching in a lot more metrics that we're now measuring to give that sense of Pseudo control I would say.
I think the role of the marketing leadership in an organization has to change too. You need someone who's willing to take a little bit of risk right now willing to experiment, push a team to experiment, but also maybe spend a little less time in the day to day of leading and lagging metrics. Like maybe you only really need to review every three months or every six months with your team because the day to day is changing so quickly for us right now as a leader of an organization. It make a little sense to step back right now and just try to push your team to experiment and as long as there's still this correlation to your marketing activity with that hero goal of a sale or a lead, as long as that correlation is still moving in the right direction, do you back off a little bit right now in this time of transition?
It would make sense because again, you have a slower time to impact. And so therefore it theoretically doesn't make sense as much to check in as often. However, at the same time, in times of uncertainty, leaders want to be a lot more involved, right? They want to make sure that the team figures out a plan and once they have trust in that plan, then they can step back.
And so as a leader, I would do the same. I would be very, very involved right now. And I think a healthy balance is to say Ok, let's all kind of be in the weeds until we have a model that works and then let's take a step back. And so, it's obviously very uncomfortable, right, for lot of teams when the leader or the manager is like very deeply involved that can feel like high pressure. or the manager is like very deeply involved that can feel like high pressure. And at the same time, again, I get why that's happening.
Best case scenario is you take maybe a four to six week sprint to hash out a new model or at least a plan for the next six months. I think 12 months planning is very hard right now, but maybe for the next six months. And you define, first of all, like how to measure, how to think about this new world and what experiments you want to do in these six months. And once you see that the team is gaining traction, right? They're active on Reddit, they're creating content, the refreshing content, they're shipping scheme or whatnot. Once you see that moving the needle, that's the time then to step back and say, okay, this is working, this is in a good place. Now I don't need to be on top of it as much. Now is the time to give the team some breathing room so they have the autonomy and the space to experiment and really bring this playbook over the finish line. But I think initially it does make sense for leaders to be involved. Also to reduce the time to decisions, to maybe get some of the necessary tooling on board basically unblock the team, and then step back.
That's really good advice. I know we're bumping up against time, so maybe we'll jump to a couple of quick rapid fire round questions if you have another five minutes with us. OK, perfect. Let's start with... If you, speaking of content, if you had a choice of investing in maybe topic consolidation a new expert source or resource, or some sort of interactive explainer, which one would you focus on?
Interactive explainer. In my experience, interactives like tool generators, quizzes, all that stuff is well received, drives conversion rates, can help with ranks, can have a positive conversion impact on lot of blog content and other page types. So I would go with interactives.
What was the hardest part about moving from a situation where you have a lot of operational control, think like Shopify versus your more influence driven role today as an advisor? What was the hardest part about that transition?
For me, it was the ego attachment to the privilege that comes with being an in-house operator with a large team. You know, that was a big learning curve for me. I was very proud of being at that point and having that leverage and having large teams and scope. And I will say, I couldn't be happier today. A lot of pressure and responsibility comes with that. It’s incredibly energy draining and obviously realize that impact. And, you know, importance is not related to how many people you manage or the scope. Right? There is much, better ways to measure that. So I would say that was the biggest initial challenge for me to overcome. And I'm in an incredibly happy space today.
That's amazing. I would guess as being a dad, having more control over your schedule is also a huge part of that. I definitely found that to be true on my side.
Massive, massive. For me, it's so much more about leaning into the things that I thoroughly enjoy and the things that I'm strong in. That overlap is the sweet spot for me. And then, you know, scheduling, yes, you're right. You have more agency over your schedule. And if you're a chronic overcommitter like myself, then that needs to be taken carefully.
Yeah, definitely. Let's see. I know we only have another couple of minutes. What? You have a long list of accomplishments. Is there one that you're most proud of? Maybe one that you don't often talk about?
Let's see, there's a couple. Last year I worked with Meta very closely. That was certainly amazing. Looking into some of the largest sites on the web and what moves needle there was incredible. I worked with the Ramp team very closely from 23 to 24. That was... incredibly rewarding that company is just, you know, like a rocket and they're doing amazing things. But you know, there's so many things that I'm proud of. There's also some like no name for... no name is probably unfair, but like less known or popular company that we were able to achieve tremendous results because they were just, you know, open to advice. There were implementing things very quickly and very receptive to recommendations and that was incredibly rewarding as well. So there's a lot that comes to mind and it's easy just pick out the kind of brands that are front and center. But yeah, man, I have to write you a long letter about this.
Well, I'd love to read it. But you're right. I think being receptive to the recommendations, I think, is a huge part of what we do. We work with some clients who get recommendations or ask for them and then don't take action. That's really frustrating. So it's always fun to work with someone who's willing and open to testing things. All right, last one. What's your favorite exercise at the gym? It's a good question. I will probably say squats because they have outsized returns, but they can also be incredibly challenging. I think going under a heavy squat is nerve-wracking and... can make you really nervous and at the same time, you know, when you stand that thing up then it's an awesome feeling. So I'll go with squats.
Very good, very good. Yeah, there's a sense of accomplishment of pushing a heavy weight through a squat for sure. All right, before we let you go, we'd like to ask a prediction question, something I'm sure you're thinking a lot about. But if you were to go to google.com in 12 months from now, what would you expect that experience to be?
It's funny you asked that because if you were asking me about 24 months I would say it's probably starting with something like AI mode or web guides. 12 months, I'm not sure we're ready yet. So I think in 12 months, we'll see the same kind of experience today. I would not be surprised if we're seeing more AI first experiences in the browser meaning, you know, today, you can already go to Gemini directly when you type add Gemini in your browser. And I think that Google will highlight that experience even more because they'd want they need to win this AI battle and they're going to put that more front and center. But the core search experience, besides again, AI overviews, AI mode, web guides doesn't change as much. But then in 24 to 36 months, unless something goes wrong. I have a strong feeling we're to start on AI mode.
Got it. Love it. Kevin Indig, everyone. Thanks for joining us on another exciting episode of the Page 2 Podcast. And remember to please remember to subscribe, rate, and review if you enjoyed the show. Thanks so much.
This was fun. Thanks a lot.