819 | His Advice Could Help Agencies Survive AI Search | Eli Schwartz

I’m in the first position. I’ve just googled myself and I get no traffic anymore. That entire model of selling visibility, selling rankings, is going out the window. I am very much in favor of this. That’s Eli Schwartz. He’s consulted for Tinder, WordPress, SurveyMonkey, and Mixpanel. He wrote the book on product-led SEO, and for the last decade, he’s been the most uncomfortable voice in the room, telling agencies exactly what they didn’t want to hear. The clients have been demanding first position. The clients have been demanding traffic. The clients have been demanding rankings. So agencies served it. Are there shady agencies that did things to give the clients the thing they wanted? Of course.

In this episode, Eli is going to walk you through what is actually killing agency retainers. No one is going to Google something on the train and find a blog post and decide that they’re now going to rip out their current analytics tool and spend a couple hundred thousand dollars on a new analytics tool because they just read a fantastic blog post. Why traffic being down 40 to 60 percent doesn’t have to mean revenue is down. The $5,000 a month rate blog post five years ago made sense because the margin on that $5,000 was maybe $1,000. Today, the margin could be, you know, $4,900. Someone who just offers content or just offers some very basic level of research that can be gobbled up by AI overviews.

And why everything agencies have been told about showing up in AI search is wrong. eBay completely lost the plot in SEO. If you can’t explain why a user would choose to search for your solution over asking chat, you don’t have an SEO problem. You have a product market fit problem. There’s a lot of agencies in the world have built their business on trying to win the first click around informational type queries and build top of the funnel and traffic. In a world where LLMs and AI overviews answer the how to’s and the what ifs before anyone ever gets to a click. Are we seeing the slow death of the very traffic model that a lot of agencies built their retainers on?

So that I think is a fascinating question. And I don’t want to throw my fellow marketers and fellow agencies on the bus. So I want to really pin it on someone else. Let’s pin it on the clients. The clients have been demanding the first position. The clients have been demanding traffic. The clients have been demanding rankings. So agencies served it. Now, are there shady agencies that created reports and did things to give the clients the thing they wanted? Of course. But I don’t think there ever would have been a product like this to sell if it wasn’t the clients that were looking for this kind of thing. So I think we’re in this transitional world where clients are now asking the questions of is it valuable for me to be in the first position where I’m in the first position?

I’ve just Googled myself and I get no traffic anymore. And the agencies maybe refusing to acknowledge reality will say, well, I’ve done this analysis and it’s awesome for you and you’re making all this money. So maybe they shouldn’t be selling that anymore. But yes, everything has changed. That entire model of selling visibility, selling rankings is going out the window. I am very much in favor of this. And I wrote my book over four years ago, which was focused around building product and product like SEO, building product towards businesses and building products towards revenue because that’s what SEO should have always been. Users are trying to serve a need.

Businesses are trying to sell a need. And search engines are trying to match the two of those. That’s what SEO should have been. But it got convoluted into this thing of rankings and visibility and clicks and impressions and anything that wasn’t necessarily connected to the business. And I think now we’re coming back to this where it always should have been. Well, was the top of funnel traffic ever valuable? In your opinion, what is the biggest illusion that’s about to be popped when it comes to top of funnel or informational content? Oh, that’s popped already. That’s totally gone. I’d say that the best thing that any agency, SEO person, marketer could do, and I’m coming to this reality in a broader sense now, but the best thing that anyone in this position can do is try to do it for yourself.

Try to create a website and do those things and see if you can convert customers. You know, I’ll keep with the medical analogy. There’s so many medical websites that I come across where they have top of funnel content. So say you look like at a dentist website and they have top of funnel content, like what is brushing teeth mean? What is floss?

But someone convinced these dentists that no, your website cannot exist unless you have all the super generic dental content. So I don’t think it has mattered for many years. I do still remember that rush of getting that big traffic pop when you get a top of funnel keyword. Well, it’s the greatest thing ever. I mean, I very rarely started a conversation without pointing him out to the fact that, hey, I just so you know, I do rank number five or six for the word cars. I met a company in the car insurance space. They were a startup. They raised a lot of money and that was their goal. They were going to rank number one for car insurance. And I had to point out to them here in Canada, but I’m sure you’re aware of these brands that progressive and prudential and Geico, those companies are older than Google.

Their websites are older than Google. The only way, and if you Google car insurance, you’re going to find those companies in the first two pages because there’s more than 10 of them. The only way they’ll ever outrank one of those companies is if those companies forget to renew their domain for a long time, not for a day, but for a long time. And they didn’t believe me. They bought lots and lots of links and they ranked on lots and lots of random car insurance keywords. And then they got hit by it. I’ll go update. They’re gone. They’re never coming back from that. But that was an unobtainable goal to begin with.

What I would say is the goal of SEO is to be a functional acquisition marketing channel. Now, 10 years ago, the way you did that was by ranking number one on that big keyword because you got the most clicks and we all remember those click curves of being number one, you got like 10% and being number two, you got 6% and like, you know, went down to number 10 where you got 0.00 something percent. That was the goal because you got people in and, you know, I did, we made a lot of money off of being number five or six for the word cars. People came in, they looked at pictures, they bought cars, they did all the things we wanted them to do.

That was the goal. That is no longer the goal and that can’t be the goal because you’re not going to make revenue. You’re not going to hit business KPIs like that. So the extension of that is, yes, it does move more further down the funnel from awareness, which is likely top of funnel to now I would say middle of funnel consideration or depends on where you are. It could even be bottom of funnel. But the expectation has to be that if they’re looking for something, they’re educated on that. Now I was working with a client in the mental health space. Their solution for mental health challenges was therapy. Now they, their primary ICP was someone with anxiety. Now someone with anxiety can treat that anxiety in a variety of ways. They can treat it with illegal drugs.

They can treat it with legal drugs. They can treat it with meditation. They can treat it with, I don’t know, retreats. There’s lots of ways they can treat it. The company I was consulting for treated it with therapy. So people were starting on the top of the funnel. If they were looking for anxiety, they were just looking for the definition of anxiety and AI overviews or Gemini or AI mode or chat, which he was going to define it for them. What they needed to do is be in the middle of the funnel. So when someone says, oh, that’s me, I have anxiety, and then they needed to show up and say therapy is a really good solution for your problem. So it kind of really will depend on the kind of business and how you expect to earn revenue from that user, but that’s where you are on the funnel. So you’ll never find like a paid marketer who will brag about, well, I’m very good at my paid marketing job. I spend a million dollars or I’m very good at my paid marketing job. I’m number one in all my keywords or I get this amount of clicks. Their metric is how successful are you? What’s your ROAS? What’s your LTV? What’s your profitability? Like how much are you contributing to this

business? And SEO should be the same thing. So it could be top of funnel, could be middle of funnel, could be bottom of funnel. I think that within our current paradigm of AI overviews, it’s likely going to be middle of funnel because people are more informed, but winning is really contributing to the business. And I think with more aware users, more educated users, if you have the right fit, you’ll be more successful with the right SEO. When an AI overview appears, the click-through rate drops 15 to 20% for a non-branded search. Which real-world industries are feeling the pain right now and which ones aren’t? It’s going to be the kinds of companies that don’t necessarily offer a proprietary solution. Now that proprietary solution could either be content or it could be something physical. So let’s say I consulted for Tinder a couple years ago. Tinder offers something that cannot be solved. Well, I’m going to go so far out of the limb, but for now cannot mostly not be solved by AI. So no matter what you find on AI overviews, it cannot solve what you were looking for if Tinder was in your journey.

So industries that offer something proprietary, either online or offline, and it could be e-commerce, it could be a solution, it could be software, those are likely not going to be disrupted by AI overviews the same way that someone who just offers content or just offers some very basic level of research that can be gobbled up by AI overviews. And I think that’s a differentiator. Going back to the medical space again, if someone searched for a symptom of a disease, the websites that were visible for the symptom of the disease, they aggregated that content from somewhere else. And the reason they were highly visible is because they did SEO things to be highly visible, but not because they were the best solution for it. A website like GoodRx, which has proprietary information about, not on if you have that in Canada, but proprietary information about drug prices at specific stores, because they’ve negotiated those drug prices at specific stores, is in a way better position than any of those other websites.

There’s a similar website, Drugs.com, which just has a glossary of the pharmaceuticals you could buy from a drug store, but that’s just a glossary. They’ve scraped that from somewhere else, but if you are a GoodRx and you’ve created information and you’ve created all sorts of synergies between different medicine because you sold them, that’s totally different. So is GoodRx’s traffic disrupted by AI overviews? Absolutely, but their revenue isn’t necessarily disrupted. And I’ve seen this with many businesses where their traffic is down 40 to 60%, but their organic revenue isn’t. And that’s because they’re still serving the customer just at a different place in the journey. You’ve said something that really challenges the industry, I think. SEO isn’t a marketing tactic, it’s a product question.

So what do you mean by that in the AI context? I think for the longest time, again, we’re blaming clients, we’re not blaming the agencies. For the longest time, it was all about rankings and visibility. In order to achieve that, you did research, you found the keyword, you created the content, and that was your output. Your SEO effort was create the piece of content. So my differentiator was they have 1,100 words of content, I’m going to have 1,600 words of content. They have 42 backlinks, I’m going to have 142 backlinks, but that’s not a meaningful differentiator to the end user. Now, if you think about it from a product context, we’ll use GoodRx as an example. GoodRx built product. They built all those API connectors to figure out the local stores. They built the API connectors to get the up-to-date pricing.

So the page might have the exact same information about the pharmaceutical product that they’re trying to connect you to, but it’s dynamic. It’s a product. It’s not just a piece of content that was written either by a professional writer or AI or just someone that you found an upward. It’s different. It’s something that took time. Now, a lot of times when I work with companies and I recommend they create a product, the biggest pushback I’ll get is, but that’s expensive. That will take us a long time. And that’s good pushback, but the longer it takes and the more effort you put into it, the bigger moat you have to the point that when someone else sees that, they can’t cross that moat. One of my favorite examples of this is TripAdvisor. So TripAdvisor was born in a world very, very long time ago when they were competing against travel blogs. So a travel blogger, and we’ve all read these websites, or now we visit their Instagram pages, they would go to a hotel and they would write 2,000 words on what it’s like to stay in the Waldofer story in Manhattan. And TripAdvisor created a website that was architected to have a review for the Waldofer story of Manhattan, just like they

have for the Motel 6 of Manhattan. The exact same SEO effort went into it. They built the structure, the UGC was slotted in, but again, from their SEO effort, it was the exact same. That’s a product. Today, when you search hotel reviews, you’re still going to see TripAdvisor living off that moat, even to the point that Expedia and Kayak and Hotwire and every single of those sites has tried to encroach on that, but TripAdvisor is still dominant globally in all the languages they serve for that product they created. Is the old content factory and model publishing 8 to 10 blog posts a month targeting a specific set of keywords officially dead?

Oh, absolutely. I had this eye-opening moment, so there are many things that led to me writing product-led SEO and feeling like this is something that I should focus on. And one of those was I consulted for Mixpanel, which is an analytics platform, and they were doing that. They were writing the blog posts and I assisted them in writing those blog posts and coming up with the keywords and writing really good blog posts and using all the keywords and all the SEO best practices as effectively as possible, but they didn’t convert. And Mixpanel is an analytics tool. We were able to see why they didn’t convert. And the reason they didn’t convert is because Mixpanel is very expensive and very, very sticky to a company. So no one is going to google something on the train and find a blog post and decide that they’re now going to rip out their current analytics tool and spend a couple hundred thousand dollars on a new analytics tool because they just read a fantastic blog post. What they will do is it will plant a seed in their mind and many, many months later, they will do something and they might meet a salesperson. So the entire

attribution model is broken. So I think for B2B, those blog posts, maybe they did something, maybe they didn’t, but it’s hard to know. So if you’re an agency then, right now, you’re selling a service $5,000 a month to write eight blog posts a month. What is the replacement service to still provide that $5,000 worth of value and get real value to your customers? The $5,000 a month rate blog posts five years ago made sense because the margin on that $5,000 was maybe $1,000 because you had to pay $4,000 to the writer. Today, the margin could be you know, $4,900 because you’re just, you have a subscription to chat to BT and then you pay an editor on Fiverr or Upwork and then they send it off to the client, which is something I’ve seen over and over agencies doing. So agencies are burning themselves by trying to make as much margin as possible. So I think the right answer and I don’t know that most agencies are equipped to do this. The right answer is to do a proper intake, understand search market fit, understand what is best for the client and to say you’re missing these product pages. There have been many

big businesses I’ve worked with where we got some of our best lift from brand keywords because their brand keyword content didn’t exist. I consulted for WordPress and WordPress is the word WordPress is by far the largest word in that entire development space. And WordPress.com, which is the for profit side of WordPress was missing many of the things that people searched for WordPress plus WordPress plus themes, WordPress plus plugins, they didn’t focus on those things. So we got some of our biggest wins by focusing on those branded keywords and creating revenue out of the brand keywords. It doesn’t feel as sexy because it’s branded versus non branded, but if the goal is revenue, it’s the same thing. So I think agencies should do that intake and say, what are people looking for? How will you earn revenue from SEO? What do we need to build?

Is it blog posts? Is it product pages? Is it a new product? And really again, I’m not sure the most agencies can do this, but that’s what they should do. If you were running a 10 person agency and it had been slow to evolve today, how would you structure a retainer and design your services now? Anything that you can just very clearly turn into a prompt. So if all you’re doing is creating content, create 10 pieces of content a month, you go into a cloud and make me 10 pieces of content a month for an automotive shop, cannot serve. So anything you’ve done can be distilled into a prompt, probably won’t survive. Let’s take a mid-sized local business, say automotive. So you have an automotive shop owner comes in to an agency and they say, I heard good things about your agency and I heard good things about SEO and you do a really good job of SEO. I want it.

And now five years ago, the agency would have been like, well, we’ll sign you up on an automotive package. We’re going to give you a blog post a month and we’re going to write about what a transmission is and how to change your oil. And at the end of the year, you’re going to have 10 posts a month. We’ll keep it even. We’ve got 120 posts and your SEO is done and every week we’ll meet with you and tell you about all the posts we’ve done in your traffic. That’s not necessarily going to drive in more business. Now, what the agency should do is they should make them indispensable. The agencies should say, can we come on site or if they’re not on site, can we turn on zoom and just watch how your employees interact with customers? What are the most common things that customers come in and complain about?

They should help them figure out. You write a lot. You think you want to write a lot about oil changes, but actually the most common thing is people come in for the check engine light. So let’s explain when you can or can’t ignore the check engine light and now we go through that flow of this is how we solve it and this is the price. That’s the thing that many people might be looking for when they certainly do any sort of searches. I have a check engine light. How much is this going to cost and do I have to do this on the first thing Monday morning or can I wait till my next off day? So the agency makes themselves indispensable and if they do this properly, they can have two years worth of things to do. If they don’t do it properly, they’re going to write the blog post and they’re going to get two months. Hey, you guys doing SEO, but I don’t see anything coming from it. If AI visibility doesn’t necessarily mean first click, how should agencies start treating AEO? Agencies are more indispensable. So the way they should treat AEO is they’re integrating themselves within the client and they’re saying

AEO is based on brand. So we also need to make sure you’re doing PR. Oh, you don’t do PR, we have our PR team. So now the agency is actually more involved. They’re selling something else. So you need to improve your brand visibility. Not in a fake way, not by spamming Reddit, but actual proving improve the brand visibility. And that’s what AEO should be because AEO is going to be that top of funnel. Should brands be worried about and focus on showing up in AI reviews? I’m not so sure. It really depends on what they’re trying to do. It doesn’t necessarily help you. Again, before we’re going back to our automotive shop, it doesn’t help you to be the one that Google has stolen all the information on check engine lights if no one’s going to come to your shop. What does help is for someone to for Google to say there is someone near you that charges this amount to fix a check engine light.

That’s the way they should approach it. What is the next question? Where can they be in that next place in the user journey? And that’s the right AEO. The wrong AEO is, oh, we used to focus on ranking number one, now we’ll focus on multiple having as many citations as possible, because that will be as meaningless as ranking number one. So the interception point in SEO and AEO is really what happens when you’re supposed to be the one that’s getting the click, because you’re the best fit for the intent at the moment. And that will range depending on who the client is and who the user is. So if we’re thinking about a mattress, which is a minimum of $1,000 purchase, and it’s something very specific, someone’s not just searching cheapest mattress, they’re searching something that they’ve heard a lot about. So if it’s a very specific mattress, they’ve done all the research and then they get to that point. And earlier in the podcast, we talked about top of funnel, mid funnel, bottom of funnel. This might be bottom of funnel interception. So now they know the name of the mattress, they know the year it was created,

they know the specific model, they know the sizing, the interception point is being able to be visible for that. Now if you move it still with an e-commerce to totally different product, like a microphone for your computer, the interception point is, well, it’s only going to cost $50 or $100. I do some quick research. What’s the best podcast mic? And then I get the name of it in the interception point is either having a blog post that explains why you’re a podcaster and you can recommend what is the best podcasting mic or just being the e-commerce site that sells it. So the investment you’re going to make in AEO or SEO is really about what do you expect the user to have or to see after they’ve done and learned everything from an AI overview or any sort of LLM response.