Mike Duboe:

Eli, thanks for joining. You have been a good friend for quite some time now on all topics around SEO and now AEO. I know we've been chatting a lot about this in private and you have some spicy takes, and so I thought it was a good idea to bring you on and air those out for everyone here. We actually had an episode a few months back with a mutual friend, Ethan, where we got into the fundamentals and basics around AEO, and a lot's happened since then. So I thought the headline for this episode is really moving from what is AEO to how to operationalize it alongside SEO and how to measure it, and maybe some more of the advanced topics around this. But before jumping into it, folks might know you, you wrote a book called "Product-led SEO." Tell us a little bit about what you meant by that and then explain which of those viewpoints still hold in an AEO world, how those lessons translate to this new paradigm.

Eli Schwartz:

Yeah, thanks for having me. I'm excited to be here and do this face-to-face. We're just always constantly texting and having these debates before ChatGPT even came out. So obviously I wrote my book before AI came out. So my book, "Product-led SEO," was my viewpoint, which really was kicked off [when] I was working at SurveyMonkey and I started moonlighting, and I pitched doing some SEO consulting and the CEO of this company said, So basically you're saying write some content and build some links and that's SEO. Why can't I just hire someone on Fiverr to do the exact same thing? And I didn't have a really good answer for that because he was probably right, but obviously I'd done more because I generated $200 million a year off of SEO for SurveyMonkey, so I did better than someone could do for a hundred bucks on Fiverr. So I started thinking about what that was, and then I encapsulated that and started first by blogging, and then I gave it a name, which is product-led SEO.

And that's the idea of building something for the search user. So a search user is going online to find a solution or an answer or a product or a service, and you want to create that asset for them. So at SurveyMonkey, it wasn't me writing a bunch of content about how they should write surveys because they would learn from my content and then go use Google Forms to do that. I wanted them to give them the process and what that product might be that they're finding and then it brings them into something that they're going to create and the reason to benefit the company. So that would be product-led SEO. So examples are Amazon, so Amazon's product, obviously, it's eCommerce, so I'm looking to buy a product. If you want to know what eCommerce looks like when you don't do product-led SEO, you look at eBay who wrote a lot of content but then didn't sell any of the products because you just read the content, or Walmart actually did something similar.

Another example of product-led SEO would be Zillow. So Zillow created a product that when you search for home valuation, you find the home valuation and it brings you into Zillow's funnel. So I started writing about that and thinking about how is it that you build this product for a search user? And the book obviously was relevant when I published it, but I think even now it's more relevant because we've removed some of the grayer, blacker tactics from SEO. The counterpoint to anything, any time I brought up product-led SEO, was yeah, but I wrote a lot of content and I bought some links and it worked, so why would I put all this effort into creating good user experiences when that other thing is just so much easier? And now what we're seeing is that other thing doesn't necessarily exist as much anymore, and it's certainly not easier. So what's left is building product experiences for a user. I would think that it's more relevant today than ever, and I'm actually seeing better sales now with the book than I did the year it came out.

Mike Duboe:

Actually, before moving to the AEO stuff, I want to pressure test one thought with you. So I think oversimplification of SEO, at least in the years I was a marketer, where there was on-page and off-page SEO, off-page being all the content, backlink generation, etc., and then on-page being kind of technical architecture, basically ensuring that your site is crawlable and shows up well. As you're articulating this, I'm envisioning a lot that lives in the middle there, actually, and you're describing product output that actually — designing product, actually, for these engines. Is that right? Do you think the on-page versus off-page split is not the right way to look at this discipline?

Eli Schwartz:

Yeah, I think that's the wrong way of looking. So off-page, for anybody that's newer at this, is really building links, and there was a tactic on its own of building links, acquire an actual link for it, and on-page would be more like how you structure the page. So I think that what gets lost, and this has been around for a lot before AI, is that on-page wasn't that effective in the last few years because Google figured it out or Bing figured it out. You didn't have to have the best structure in the world, you could figure it out. So there was an argument many years ago against using platforms like Wix or Squarespace because they didn't have great SEO, and now those platforms are like 98% of the way there, and you're not necessarily going to see any downside from a structural technical standpoint.

So technical SEO has far less value than it ever had. So that would be on-page. What it is is, I would take a basic assumption that you do a good enough job of having an on-page site, but now what is the asset that someone is looking for? So assuming that your website is good, what's the asset? In an eCommerce case, it might be the eCommerce landing page. In a software case, it might be something higher in the funnel that tells you about the software or why you should use that software versus another software. So it doesn't really matter whether you structure that with the best title tag or not, as long as you've built the page around a good user and around a good use case.

Mike Duboe:

One of the arguments Ethan made is that early AEO could create asymmetric advantage and that startups could really start to win here faster than in SEO. Do you agree with this? What are you observing with those who are early to the channel and clients that you work with? How are you coaching them around this?

Eli Schwartz:

Well, I would challenge the entire premise that AEO even exists. Ethan and I have similar debates on text just like you and I do. I think the entire concept of AEO was invented by people looking to sell something additional to SEO. The same way you don't have SEM people — the paid side of the house — suddenly decide that SEM is dead and there's this whole new thing to do just because there's TikTok or influencer marketing, they've just adopted it. Like, yes, search marketing is everything that we pay for visibility. And I think of SEO as the same. I don't think mobile killed SEO. Obviously the UI completely changed when we flipped to doing more mobile searches versus desktop. That doesn't mean SEO died, just [that] SEO transformed. The same will exist when we're in driverless cars and the medium we're using is now a screen, or maybe there's a screen on your steering wheel, or maybe we have glasses and we nod at stuff, and that's still going to be SEO.

So I would say the form factor of search will continue changing. So we don't really need another name. There's a push and there's a pull. So pull is like I go and request something, push is I'm requesting something and then I push an ad at you. So pull is everything SEO. We can keep that name or just make a new name, but I don't think there's SEO plus AEO, and I don't ... If we need to make a new name, it shouldn't be AEO, it should be something that is like "organic optimization" or something like that. But SEO will stick for now. So I don't think there's something new. But as far as the question of is Ethan right or not? I'd say it depends. I'd say most companies don't do a really good job of SEO at all. And what we're seeing now is that they're being exposed for having bad SEO.

They never had an SEO strategy. There was nothing they ever did to optimize for actual product experience, for actual conversion experience. They were generating traffic that may or may not have turned into conversions, but it could have accidentally turned into conversions. One thing there's been a big debate around on LinkedIn is monday.com because they specifically called out AI is destroying their top-of-funnel business. I would say that monday.com never really had effective SEO because they were just writing a lot of content for their space and hoping that someone would read a blog post about something that monday.com wrote, and then they can retarget them or peel them off and try to bring them into the conversion funnel. But that content they wrote was never really in that conversion funnel to begin with. So in that sense, they didn't do SEO, so therefore they can't say SEO died, they just were exposed and the medium changed and top-of-funnel disappears. As far as companies having asymmetric advantage if they jump into AEO — which, again, I'm not so sure that it should exist — it only matters if there's something that is new around what they're doing.

So I've typically said that B2B companies — B2B software — should not do SEO at all because when I think about it from a product-led SEO experience, is there a product that search users are looking for? Potentially not. There's a handful of landing pages saying this is what the company does. There's not a wide range of thousands of pages that anybody should invest in from an SEO standpoint, it's just a few pages. And those few pages might cover a handful of search terms, but again, not thousands of terms. AEO becomes different — or answers become different on AI search — because now there is a reason for someone to go and search for these things and discover something new on a search engine because you're doing comparisons that are more relevant to what they're doing. But that's actually brand marketing. So I would say B2B software should never have done SEO — they should have invested all those dollars in doing something else — but there's asymmetric advantage if they can figure out how to do brand marketing that shows up in AI answers.

Rishabh Jain:

So Eli, I think I am understanding the distinction here, which is basically that, in the world of SEO, there was what you would describe as a hack to showing up in search that would drive traffic to your site, but it was not actually SEO because you weren't actually fulfilling the user's need in part of the consideration funnel. One of the things I get curious about is let's just take any B2B software example where it's something that's more complex to use. So let's just say Salesforce, you can configure it in any way that you want. And so writing a lot of content about various configurations, would you describe that more as product information and therefore is actually more of the variety of what is actually useful to the user who's trying to fulfill a product use case? And so in the way that you frame this, examples of how to use the software to fulfill a use case and having content about that is actually the right way to do it and has always been the right way to do it, and therefore will get picked up by an answer engine.

Eli Schwartz:

Yeah, so I'll use Mixpanel as an example because Mixpanel is a company that I consulted with early on [and] is one of my first clients, and that was where I discovered that B2B software potentially wasn't the best fit. So I've consulted with multiple companies, Mixpanel was the first. So, we did a great job. We followed the complete playbook for SEO for Mixpanel, but it didn't ladder into any conversions. And the reason why is because, at the time, there was really — and there's potentially still — only really a handful of analytics tools. So if you're about to make a purchase of an analytics tool, which you're going to bake into your product and the switching costs are incredibly high, you're going to consider every single one of those products. So the Mixpanel SEO was like, We could create the best page on top analytics tools or ways to understand LTV or all the things you might do if you were doing for an analytics tool, but it's not going to generate more conversions because even if Mixpanel was on page 100 of search results, you were going to find it because you weren't going to make that purchase without talking or at least viewing all of the different options.

So in that case, there were a handful of things to do around Mixpanel that were going to be effective conversion SEO pages. For example, Mixpanel versus Google Analytics, Mixpanel versus Omniture, or something like that would be very effective because that, you're in the consideration phase. But what is analytics? What is LTV? Yes, it would generate search traffic and you can make thousands of those pages, but that's not very effective from an SEO standpoint. Does that answer that?

Rishabh Jain:

Yeah, that makes a bunch of sense. I guess one of the things it makes me curious about, and maybe this is in the spicy zone, is if you take that view that it is harder to make SEO to be effective for B2B, it is also the case that a lot of the purchasing happening right now for tracking of how you show up in answer engines is happening from B2B. How would you reconcile those two things? And again, a marketing investment is always an ROI thing, and it's just the case that SaaS is more expensive and therefore it's easier to justify the expense. How do you think about the fact that that's sort of how the landscape is playing out right now?

Eli Schwartz:

Yeah, okay. Thanks for framing the question like this because this allows me to show this difference between brand marketing and SEO. So the reason why I don't think a tool — let's keep with Mixpanel — shouldn't have done SEO in the past is because there's only a handful of queries that would matter. Analytics tool, like again, conversion tools, like, "What is analytics?" Not a conversion query. That Wikipedia could answer. But analytics tool, or I don't know, mobile analytics. Again, a handful of terms, a handful of pages. Now when you're doing your research to figure out which tool to use, even if you took a sales call, even if you took a demo, you go and read all the websites. So you would read Reddit, you would read, this isn't my space, but I don't know, let's say Emarketer, you'd read all those articles on people that have made these decisions and have taste tested this, this is less obvious when it comes to B2B, but could pretend that it was like a consumer tool like you were purchasing, I don't know, a MacBook.

There are dozens and dozens of pages you could read describing that MacBook experience again. So thinking about, now let's flip back to B2B. So that's where you would do that research. What happens with AI is AI has disintermediated — or the answers have disintermediated — all of those websites. So it distills every single one of those journalists into an AI summary. In the past, I would say do brand marketing, do a bunch of outreach to get featured by all those tools, make sure those influencers — let's call them influencers because they're writing content, they're influencing your decision — have tested Mixpanel, and they understand the experience, they understand why it's better than Google, and they write those detailed articles. Now, those exist less, and you could just feed that content by creating ... Again, you're getting — well, we could dig into it —but you're getting citations. Now AI summarizes exactly what those 30 websites would've done in just one answer. And again, I still don't think it's SEO, I think it's brand, it's just you're doing brand for the AI engine regardless of whether it's Google, Perplexity, Claude, you're doing brand for that. Some people are calling that SEO, but I think we're essentially robbing the SEO budget to pay for brand marketing.

Mike Duboe:

Well, so Eli, aren't you, in this example here, are you not describing AEO? You described — at least my simplification of what you just described — is instead of spending all this time trying to generate content to go and get backlinks, go and actually get cited by the right influential sources. The work that's required to get that done will pay off in you ranking in these engines because now these engines are pretty effective at ingesting all of that stuff and just surfacing the relevant results based on those signals. So isn't that kind what AEO is? And maybe we could debate, people should be spending money on monitoring tools, but that effort is conducted by content and SEO/AEO teams, right?

Eli Schwartz:

It's not. I think the outcome, the effort is conducted by brand teams. I can't imagine there's a case where a brand did not do any good brand efforts and the answer is going to be different than what the brand has already done. So as an example, one I always look at is what are the top SEO tools? You're never going to find some random tool you'd never heard of that did a better job of — okay, let's pretend it's AEO — did a better job of AEO, but their SEO and brand marketing was terrible. So if they've done a good job of SEO, which means they've also probably done a good job of brand, because SEO is now reflective of brand efforts, you can't just build bad links. You have to actually get good links, which is branding efforts. So it's unlikely that the answers are going to be not reflective of what SEO is. It'll be nuanced, of course, but I don't think it's worth the entire effort of creating a new channel.

Mike Duboe:

Okay, so maybe let me tee up an example. So I'm the CMO for a midstage B2B software company that, over the last three months, has seen my volume ... seen ChatGPT as a referral source go from basically zero to, let's say, 30%. I'm not doing anything to influence that, but I see that and that's a very promising trend. What should I do? If I'm your client, what do you advise me to do in that case?

Eli Schwartz:

So I would look at that answer. The best thing about ChatGPT and Gemini now is it tells you the sources and those sources are going to be examples of other places you should reach out to. So the same way that if you had a PR team, they would find out the — again, we'll call them influencers, but they're not like LinkedIn and Instagram influencers, they are people that influence purchases. So the same way you got that visibility in those places that became citations for the answer engines? Do more of that. But I would even push back, you're seeing a ton of referrals from ChatGPT, it just means that ChatGPT has done a really good job of summarizing the work you've already put out there. From an investment standpoint, I would invest more in a PR agency. I would take all that answer engine budget you have, invest in a PR agency to get you more visibility and invest in a Reddit agency to promote you on Reddit because that's where people are making those final decisions. And not because Reddit influences the answer, but because Reddit is that final place.

Rishabh Jain:

Maybe we can dig in a little bit more into the sources and what you're observing as the sources. One set of sources is what you're describing as journalism influencers, or like journalism influencers. So they have some authority because they're associated with whatever magazine or newspaper. One is Reddit, which is social influence. One example at least I've seen is actually interestingly for a lot of B2C or commerce transactions, there's a lot of retailers that get cited, and retailers are a source of influence. What are some of the sources that you're seeing and how should we be thinking or how should the audience be thinking about what sources to be influencing? You mentioned the PR strategy, but what other mechanisms should they be thinking about with respect to influencing the sources?

Eli Schwartz:

So I think, at its essence, all of these engines are trying to imitate a human experience. In the human experience, if you're looking to do something completely offline, you're buying a new pair of shoes and you want to know what's a quality pair of shoes. You've grown up, you're ready to not just wear flip-flops to work, you ask your friends, but you ask your influential friends. You go on the subway and you look at the shoes that other people who you want to look like are wearing. You look at the ads on the subway or wherever you're going and you see these ads and you judge — again, all subconsciously — you judge the quality of the ads by saying, Wow, that's a really creative ad. They must have had enough money to hire a good ad agency. That must be a quality shoe.

All these things are subconscious, and you're trying to, again, distill what is something of quality, what is worthwhile, what's an investment you should make? The LLMs — again, they're initially not really AI, but they're getting to the point where they're starting to think — they're trying to emulate that human experience. How do humans understand brand and they're looking at all those sources. So there's no quick hack to like, well, if you spam Reddit now you're going to show up in ChatGPT because that's too easy to hack. No user is going to trust everything they see on Reddit. They shouldn't trust everything they see on Reddit. It's a little bit of a less reliable place, but it's a bunch of opinions and LLMs are trying to distill what those opinions are, which is why for now you can poison the LLM responses. But they're trying to get to a point where they can look at a wide array of sources of things that are influential in a space.

So let's say for software, Stack Overflow might be influential. For finance, Reddit might be influential. So there may be things that are influential today, but as they improve, they'll look at, again, their user experiences, their clicks and try to understand what becomes the most influential. But where they're ultimately trying to get to is exactly the way humans think. How do you decide what is the best pair of shoes to buy? What is the best restaurant to go to? I actually saw this in real life. I was in Houston and I saw an urgent care clinic and it was called Urgent Care Near Me. Obviously, they were trying to get people that search "urgent care near me." That's very creative of them, and they may not have been high quality, but you'll just find them. But if you're actually looking for a quality urgent care, what are the things that you would do to be that quality?

So again, as humans, we're going to ask people, we're going to look at the location, we'll map it to see if it's really near me. So the engines are trying to get to a point where it's not, you just call it Urgent Care Near Me or Best Pizza or Most Delicious Ice Cream. It's like, what is actually the urgent care near me? What is actually the best pizza? What is actually the most delicious ice cream? So that's what the engine is trying to do. So the same way humans will make these judgments, the engines want to make those judgments, which means that it is an incredible, diverse set of citations and sources that they're going to do that. So if you want to be most effective in AI, you want to be the best brand. You want to actually produce the best product. So there's less hacks today than there were, which means that if you're a startup and you're trying to get ahead, you can't just figure out how to hack SEO and you'll get a bunch of visits. You actually have to be the best. So good news and bad news.

Rishabh Jain:

(laugh) The thing that it makes me wonder, and one of the provocations that Ethan had, and I'm curious what you think about this, is "best" means different things for different people in different contexts. And what it actually then points to is that the level of nuance that you have to get to to actually answer "best" is extraordinary. For example, if you just said something like "running shoes" or "best running shoes," I would describe that as the previous generation of how people queried for something, right? It's like extremely short queries, extremely broad category. But the problem is that that actually does not answer "best" because Mike and I might run differently. We might run on different terrains, we might have, I have a tight IT band. There's all of these things that now, in principle, I might actually — and actually, I do this — I actually give way more nuance to the answer engine such that a "best" for me can show up. What that suggests, though, is that somehow that content has to exist, and this is sort of what Ethan provoked, at least for me when we were talking to him, is that you have to produce content that is extraordinarily granular and nuanced because now all of a sudden queries are longer and are more nuanced. And I'm curious, when you think about "best," how do you think about "best" and would you agree with that view? Do you have different views on what "best" means? How would you recommend someone interact with this?

Eli Schwartz:

I love that you brought this up because about, I guess it was two years ago, I wrote on LinkedIn that there's going to be an SEO apocalypse because I saw an early version of what was then called SGE. I was bothering a Google engineer, and I bothered him so much, he just showed it to me and I was like, Oh my God, this is the scariest thing I ever saw. There's no reason to click anything below search results anymore. I just see what they called SGE and now is AI Overviews. So I wrote about it. I didn't say that I'd actually seen it, but I wrote about what I thought Google could do and that would be an SEO apocalypse. And most of the comments were saying that I was being alarmist. But I think, looking at it now, there was an SEO apocalypse.

We lost 40% of traffic, many sites even lost more than that. There's actually another SEO apocalypse coming, and this is something they announced at Google IO in May, which is personal context search. So you no longer need to be an expert prompter in order to say what "best running shoes" means for you. Google said they're going to use your context, but I think obviously they're going to use so much more than they're ever going to admit to. To them, context was like your YouTube watch history, your Gmail, your other Google touchpoints, but probably not. They're probably going to go so far beyond that. They're going to look at the words in your Gmail. They're going to look at, I dunno, pictures you've taken, if you use an Android, of other running shoes, they're going to look at, if you use Google Maps to run, how fast did you run?

They're going to figure all that out. It's available to them. And they're going to load that into prompts the same way that they load into prompts. If I search "best pizza near me," they load into prompts — or sorry, the query — what my zip code is, but they can load all the rest of that in as a long prompt. So that's going to be another SEO apocalypse because we're going to go from "best running shoes" or "best running shoes for trails" will now be "best running shoes for trails" and then the invisible query will be all the personal context they know about you. So we're going to explode from 100 possible queries into tens of thousands or tens of millions. Many sites are going to see some sort of huge reduction in traffic. The second part of that is, does the content need to exist?

I know that some right now are trying to write a bunch of content on their website, but I'd say you don't necessarily need to match content or string match content the way you always did in traditional SEO. Because actually when I started SEO — and Mike used this long time ago — I was working for an automotive review site and we had to have the word "car," we had to have the word "cars" and "vehicle" and "automobile," all the synonyms. Now, of course, you don't need any of that. You need to have basic ideas about what your product is, but you don't need to match the strings. You don't need to assume any of those things. So you can write about your running shoes, like it has cushioning foam, and then they can infer from what you've described, all those other things. And even better — going back to my brand marketing point — you don't need to write it as long as someone writes it and creates that citation, which it doesn't need to even need to be a link anymore, it just needs to be a citation mentioning this Nike running shoe has cushioned whatever and quick bounce back, and then they can infer all the relational words to it. So I'd say it's even easier. You don't need to load up all of those words, figure out all those words, and then load them all up on your page. You just need to make sure that enough people have tried it that describe the running experience, and you have enough of the descriptors on your site that it can be expanded into all of that.

Mike Duboe:

You bring up an interesting dynamic here. It reminds me of a conversation we had with Dmitry from Perplexity on this pod a couple of times.

Eli Schwartz:

I tried to listen. It was too smart. So kudos to both of you on that. (laugh)

Mike Duboe:

Well, you're hitting on a very similar theme, which is the importance of them owning the browser means that they have access to it, not as much as Google, but a lot of this same kind of personalization signals that you're referring to here. In that conversation, we tried to dig into what are the different ad products that they could build around that? And the conclusion was that they're going to lean more into personalized offers and things like that, which are essentially a form of advertising. But maybe if you take what you're describing to the next logical conclusion of some of the stuff that Google and Gemini are launching, what type of new ad products does this unlock? And I know you're more focused on organic, but the extension of SEO on the paid side of things, how's this adjusting your priors on that?

Eli Schwartz:

Yeah, so I would just add on one thing that I did hear from Dmitry, which is he doesn't want AEO to exist either, but for a different reason. And don't want it to exist ... I don't like that there's manipulation in SEO and I think if there's manipulation in AEO, the LLMs are doing a bad job. We want them. As consumers — and most marketers don't do this enough — just think how you want to be as a consumer. We've all had the poor experiences. I know that my parents have told me, Oh, I clicked on something and I signed up for a credit card. It was like "Chase credit card for you." And I was like, Oh please, now we got to change all your passwords. We're like, we don't want that to exist. So as consumers, we want, when you do an LLM search ... Actually, a good example of this is Pinterest.

So a lot of what Pinterest has — because is Pinterest is a search engine. You click on something — and I do it for my kids. They ask me to print out a Superman they can color and I click on it and I don't land on "Superman I can color." Someone figured out how to essentially spam Pinterest — which again is a search engine — and land me on something else. So as consumers, we want LLMs to not be manipulated. So I agree with Dmitry on that. The LLMs are not doing a good job if there is an AEO element. I think SEO should not be manipulated, therefore we take it one extra level, the LLMs are even smarter and they can summarize all that bad content, and that should be a good future for us as consumers. But going to the point about ads, I'd say this is a touchy subject for Google because it reduces access.

Google is able to have a lot of page views on search because you couldn't really figure out what you wanted, so you did a search and then you refined your search. So in each of those cases, you're refining search, you had an ad, and maybe you just gave up and you clicked an ad. But as search becomes smarter, and as we talked about personal context, if now you do a search or a prompt and it just got it, it knows who you are, you're done. You've gotten your answer, you probably don't even need to click an ad. So I think this becomes a very challenging space for anybody, right? Perplexity is different because they're starting from zero and don't have any ads. But for Google where they've allowed access to so many small businesses who can out-create larger brands, that sort of disappears and it becomes like whoever has the most money can buy all the most visibility.

So I think there may be an element of, again, brand ads, we're going to have some sort of ads sponsor an entire category. So Google figures out you're looking for shoes and they'll offer it up to Nike and Adidas and a certain amount of inventory they'll just give to them. And again, it won't be targeted. Or there could be an element of they bring back — I forget what it's called —Kontera. Do you remember Kontera, way back when, where you could buy ... It was a network where you could — I think AOL bought them — you could buy clickable links and content and then you pay on a CPC. They could bring that back so answers will have those sponsored ads. But again, I still think it's so much less inventory and it's less satisfying for users if you just have one hyperlinked word "shoes" and you click on it and you're like, Oh, I didn't want to go to the Shoe Warehouse. I had no intention on being there. I'm not ready to be there yet. I'm just reading. So I think that that's a touchy balance for them. So it's either one of those where they do a bunch of brand ads or just have sponsor links, but if you have more efficient search, the outcome is that there's going to be less ad inventory.

Mike Duboe:

Yeah, it's interesting. So you are making the case that, compared to the last gen of search engines, ad revenue is going to be a smaller percent of maybe an overall revenue mix on these answer engines. Is that kind of the case you're making? And maybe let's take ChatGPT — Google is maybe a little bit different — let's go to ChatGPT for a sec because they've showed some of their cards on where they're heading on the ads front and what they're doing with commerce as well. What do you think, if you extrapolate that out maybe a year or two, what do you think that monetization surface area looks like?

Eli Schwartz:

I don't think commerce works, because I think Google has tried a lot of these things and it never really worked for them. And Google has a lot more that they could build on. With Google, I don't know the exact number, but I'm an Android user, so my credit card is already loaded into my Google search. So if I wanted to buy, they have my Gmail, they have my shipping address, and they have my credit card, and still Google has not found it very efficient to own that entire shopping experience. I know OpenAI is going to try to do that, and they're going to try to again disintermediate the websites completely, but then they own a certain amount of logistics. So I don't know that that becomes something for them. I think that it still ends up being search and OpenAI has to figure out some sort of ad monetization model.

And quite frankly, there's really only been three big companies that have figured out really good ad networks, and it's Google, Meta, and TikTok. Reddit's been around forever and they could have a fantastic ads platform with their ability to microtarget users because they know a lot about users and the things they write, but the best you can do on Reddit is just bid for a spot in a subreddit, which is good. If you have the right product, you're willing to do that. But for many people who just want to use Reddit but just can't because it's so limited, building an ad product is very, very difficult. So OpenAI may do that, but it also means they get distracted from all the other things that they're building.

Mike Duboe:

You mentioned Reddit, and I wanted to ask this earlier, but since you just brought them back up ... In the spirit of tactical advice for teams that actually want to make the most of these channels, and I know you might refute some of that notion, but there is some stuff that one could do to influence results. You mentioned Reddit agencies and actually I'm not very familiar with that ecosystem. If I'm a brand that wants to make sure I'm showing up in Reddit the right way, what do I do?

Eli Schwartz:

Reddit is dangerous. There are two things that people are going to do with Reddit. There's some software products that announce that they're going to listen to the ... They're going to look at the LLM searches and they're going to listen to where the —quote-unquote listen — to where the citations are coming from, and then they're going to create AI content on Reddit. That is really dangerous because Reddit users are touchy. So if they see that content, they're going to get mad at the brand. And then I think eventually Reddit is very aggressive on fighting spam as a whole, just from an overall qualitative standpoint. But then the subreddits are aggressive at fighting spam. So I think they'll have to step it up just to push all this back. And then Reddit itself will probably get demoted in the engine so all of this will be useless. What a Reddit agency would do is it has created not fake people that are authentic, but real people who are authentic because they're real and they participate in Reddit as a community or brands can be coached on how they can participate in Reddit as the brand.

There are many great brands that are on Reddit. Google's a good brand, Google's on Reddit. There are Googlers who are on Reddit that will answer questions about any workspace product and they'll disclose that they're a Googler. That's what it means to participate in Reddit. So if you don't have that access, you can hire an agency who will, they're just like your LinkedIn agency or just like your X agency, they are you or they coach you to be you. So I think that's something that companies should do rather than AI-create Reddit profiles and AI-create the Reddit comments. Again, it'll make everyone in the subreddit mad at the brand and that's too dangerous to do. And I don't think that companies should use Reddit specifically to have visibility in the AI engines. I think they should use Reddit because it's a fantastic place to create those conversations.

The other thing, and this wasn't anything I planned on bringing up, but we should talk about Meta, because if you think about Reddit as a place that Google relies on to create the conversations and to learn about things, Meta also has an LLM search and they're not as aggressive with it, but they could become so much more aggressive. So on every single Meta product, whether it's WhatsApp, Facebook, or Instagram, your search is also a search into Meta's LLM, and all of those conversations are being created within Meta's ecosystem. So the WhatsApp — again, I don't want to think about it too much — but the WhatsApps we're having are feeding their LLM. The comments on Instagram — so not just the Instagram posts — but the comments on Instagram and the Groups on Facebook. Meta has this closed ecosystem that they don't need to buy data like Google does from Reddit, so they can know everything about anything.

I haven't used Facebook aggressively in a really long time, but there are very specific Facebook Groups that are very active, let's say patient advocacy groups or community groups where a neighborhood will be on Facebook, that is knowledge that is not necessarily available to Google anymore. There used to be forums, but forums is something that was deprecated when no one wanted to moderate them anymore. But a lot of that exists in the Meta ecosystem. So if Meta wants to start competing in the consumer search space, they can. They're not yet, and I don't think their LLM search is that good, but it is available to them.

Rishabh Jain:

I want to take this opportunity to tie a little bit your comment about ads to this comment about Meta, and I'm sure you've thought about this a bunch. I'm curious, would it even be in their incentive to create a consumer search product? Let's take your construction where the consumer search, which by the way, I agree, the less searches you have to do, the less opportunities for surface area, the less effective the ads are. That's the inverse of what Meta is. Meta is an entertainment system. And so actually what their incentive has to do is to consistently entertain and they have, with this data, an increased opportunity for ad surface, at least from the perspective that I see it. I'm curious, what has been your conclusion in terms of, if the ad opportunity for search is reducing, what do you think happens on Meta and what do you think brands should be doing about that?

Eli Schwartz:

So I'll answer the second part first. Brands need to think broader. Brands are always looking for shortcuts of like, Oh, well, let's figure out how to get AI visibility and spam Reddit. But instead of thinking they should think broader, we'll create real conversations where real people are talking about things. How do we make sure — and I wrote this in my newsletter today, so I dunno when this podcast comes out — but I wrote this in my newsletter today that NPS scores are a better indicator of AI success than domain authority. So focus on NPS scores. If you're doing that, if you care about your customers, you'll be showing up in Reddit, you'll be showing up in WhatsApp conversations, you'll be showing up in all those places, so you're prepared for whatever engine does take off and wherever those searches are happening. From an ad standpoint, I think that organic search or organic anything is never that effective because it only exists to the best of a user's ability to prompt or search and for whatever content exists, which is why ads are so successful.

I think we've all bought things off of Meta's ads just because it is really good at targeting you and they know all those things. So yes, Meta could benefit by having the right ad and the right brand show up in the right place, and Google could too. I just think that they have to flip their entire model, which is always based on CPCs and bidding and quality scores. It might have to become interest targeting, which is like, I know you were planning a vacation when you're doing these prompts, but you're in the market for a new car so I'm just going to have to show you this ad. It may have to become like that, which is exactly what Meta does.

Rishabh Jain:

Got it. The follow-up that then drives to is if you think about those different surfaces where you're going to show up and the different types of opportunities you have to show up and it's like, okay, my best chance on the organic side and on the AEO side is really just product investment. So hey, here's the product and the UX investment that I'm going to make in order to make sure that that's driving the right outcome. It feels like in that world, actually, the best thing you can do is to double down on paid media as well in order to take advantage on both sides, right? It's like, I want to do the best thing from a product perspective, and I want to double down on the paid media side, which is weirdly the inverse of what brands have been doing. Just to be direct with you, we work with a bunch of brands. In 2023, 100% of brands were investing in organic over paid. And so the framing of how you allocate your budget in that context is really quite different. Am I understanding the framing of the budget allocation correctly that you would suggest?

Eli Schwartz:

Yeah. I think companies should spend as much money on paid as they're comfortable with, and then make a product investment on organic that's a long-term equity bet. So if you really understand your users, you're making this bet. It's not like, let me optimize this piece of content and then I'll move on to the next piece of content. You're making an equity bet to say, this is the thing that I know my users are looking for, and you invest in it from a product standpoint. In the meantime, acquire as much on paid, because if you have an efficient paid model and you're, again, working towards a CAC and an LTB that you're comfortable with, spend as much as is affordable for you because ideally it's profitable. And this other SEO bet, it's not like a other side of the coin, it's just a different marketing channel.

It's a different equity bet. So I don't think they should dig into each other. Whenever I hear, Well, we don't want to invest in SEO because we don't have enough money for paid. That doesn't make any sense to me. And the inverse, of course, is like, I'm only investing SEO, I can't afford paid. It's like, well then why can you afford SEO? This is a bet. You're making a bet which doesn't have a guaranteed outcome. Do what has a guaranteed outcome: Get your users, you know your model works, so spend as much as you can, and this is something else that hopefully will pay off. And this kind of SEO doesn't pay off immediately. You're building into a use case that takes a long time to build. So from all of my client engagements, it takes at least six months until they can even publish something that is effective, but then the larger the brand is, by the time they publish it, it starts accruing value almost immediately because they have that brand. Again, it's an equity bet. It's a long-term bet, and it is not the other side of the coin from any sort of ad spend.

Mike Duboe:

Eli, I want to shift gears and talk about the emerging services industry, not emerging, just the broader services industry around this space. In the past, the history of SEO would suggest that a very small percent of revenue goes to software and a lot ends up going to services, and the agency world is pretty fragmented there. And I know you refute the AEO term, but let's just, for simplicity, let's just call it that for a minute. In that space right now, there's a lot of venture money going towards the initial crop of monitoring software partners, and I know your view on that. Many of them are trying to attach into services. And then there are many services agencies, including ones that we really respect, like Ethan and stuff, that are helping attach the AEO offering onto what they're doing in SEO. How do you think this all shakes out?

Eli Schwartz:

I think a lot of services agencies were doing something that was not effective. They were producing deliverables, and those deliverables didn't drive ROI outcomes. So the deliverables might be a keyword plan. You don't really need a keyword plan. The product marketing team should know what the keywords are because the keywords are how you describe your product. You don't need that to come from Google Keyword Planner, but that was a very specific deliverable. An agency also might've produced an audit. I rarely produce any audits for clients because I can take one look at their site and say, Well, you do good SEO or you don't do good SEO, and I can base it on what their traffic performance or conversion performance is. I don't need to spend three months and charge them thousands of dollars to go and give them a report that tells them the exact same.

It's like when you go to a doctor, you know whether the doctor's going to say you're doing a good job or not a good job, unless obviously there's something bothering you. So I would approach ... Again, agencies do the same thing. They'll comment and they'll say, Well, I know your site kind of sucks, but I'm going to give you a report that really details the area that sucks, and you've waited three months and they still don't fix any of those things. And then the last thing that agencies have been doing, which has been completely disrupted, is content production. So a lot of this content production that can be farmed out has always been farmed out, and agencies have been upselling it and the content has become less valuable. But flipping sides here for a second, as consumers, how much do we just consume all this content that agencies have reduced on behalf of the clients?

Probably not much of it. The content we consume has been the authentic content — the Reddit post, the news article, the influencer post — not this long-form, 1,200 words on what an analytics tool is, but those have been the agency deliverables. So I think a lot of what agencies have been doing has been completely disrupted. And what I do and what Ethan does is provide strategic guidance towards brands who typically did not have the strategic guidance. And I think there aren't that many agencies that do that. So there isn't as much demand right now for it because I think we're in the life cycle where I am blaming Google, I've lost all my traffic, but it's Google's fault or it's the agency's fault that I've lost traffic. We're not yet blaming the fact that there might not be a strategy. I think we will get to a point where companies will start now looking at, What is my overall strategy to attract the organic user and was the strategy I had even effective? But right now, I think we're still in the Band-Aid stage.

Rishabh Jain:

I might ask this in what might feel like a provocative way, but I feel like we can do that. I guess one of the things that makes me wonder is, I feel like a lot of marketers need to feel a greater sense of control on their destiny. And so what you describe as deliverables is a mechanism, I would call that a control mechanism. So it's like, I get these deliverables, I know that these deliverables are coming to me, therefore I feel control, and I can report it to my boss inside of my organization, just like the agency's reporting to me. And I mean, at some level, that allows us all to feel comfortable in our jobs. At the risk of asking this in a blunt way, do you think that ... How would you tell the marketer who today drives value from that, where should they feel confidence and control from versus what you were describing as a long-term equity bet? Because ultimately somebody's got to be able to say something to their boss.

Eli Schwartz:

Yeah, no, that's exactly the issue that comes up with big companies, and I would be surprised if you face this at startups. So startups are obviously action oriented and they don't just deliver audits and deliver keyword plans. At the big companies, they're not really incentivized to think outside the box. And I think where SEO and most of marketing has to come in an AI age is we have to be creative again. A lot of marketing has become formulaic where here's the SEO recipe. I'm going to go do keyword research. I'm going to go write the content. I'm going to, well, I'm going to brief the content, which becomes a piece on its own, so I'm going to brief the content, get approval on the brief, then I'm going to write the content, then I'm going to edit the content. None of that really gets, in many cases, to a conversion.

So I think we need to bring back marketing to ... You guys did. This is the Intelligent Marketer podcast, so marketers need to be intelligent and think like consumers. What is going to move the needle here? What is going to sell product? What is going to get us to checkout? What is going to get us to ROI? And create those deliverables rather than checklist-style marketing of, These are the things that I have to do. In paid, it's always been a lot easier because the checklist can be gone through in an hour and either you make money or don't. But with organic, I've seen many cases where they go through the checklist over years and then they say, Well, our SEO isn't working. And I'll say, When did it work? And they can't really point to a time where that checklist drove any value, but they keep running off the checklist.

Mike Duboe:

Eli, maybe one last question to bring this home. So you've given a lot of commentary on reframing the AI opportunity more around brand and product marketing. Who do you think ends up being the most successful as a result of this channel over the next year or two? If we look back and say, Man, these companies just really nailed it and they really grew from these answer engines in a way that is beyond what we would've anticipated. Who are some of the brands that come to mind and what would they have done right, starting today.

Eli Schwartz:

I actually haven't seen anyone that's doing something incredibly right, because I think, again, we're still in the life cycle of tactics and Band-Aids of this is something that I used to do to win in SEO, and now I'm going to try to do this to win in AI. There's a client I'm working with, they're in a consumer space and they sell a product that has been completely disrupted. So they had an informational product, which drove a lot of revenue, a lot of ad revenue, but they also sell something where they drive leads. There is no engine that can drive those leads right now. The consumer needs to get to that endpoint. So that has to exist and they're pivoting their entire strategy of how do they showcase what the user has to come to the website to, again, finish the transaction. So I don't think I've seen any good examples of a company that's truly embraced it, truly pivoted their entire model from selling clicks and generating clicks and selling eyeballs to, This is how we're going to give the taste test of what we have and then come into the website and finish a transaction. And that's where we're going to end up being. So I think we're in this life cycle. You mentioned software earlier, but we didn't really talk about it, but we're in this place where like, Oh, there's AI, and the way I'm going to address the AI challenge is by buying a software that shows that I'm already doing well. When I worked at SurveyMonkey, there was a product we sold that was brand visibility. And I think brand visibility trackers do a significantly better job than creating LLM prompts that people may or may not use and trying to declare that your brand is winning. If your brand is not winning at the LLM prompt, you're not winning at brand.

You need to focus on the brand. And then likely, if the LLMs are effective and doing a good job, they will show that brand in an LLM prompt. So we can ignore that. That is not doing, again, you guys call it AEO, that is not doing AEO if you've just bought software. Doing effective SEO for an AI age is understanding how the user has changed. I think we've lost a lot of top-of-funnel search. What we're going to be moving to is middle of funnel. So users, as we talked about earlier, they searched "running shoes." There's a company I worked with, they were in the mental health space. They used to rank on the word "anxiety." There's no reason for any website to rank on the word "anxiety"; an answer will tell you what anxiety is. But they provide a solution. They provide therapy as a solution, and that's a mid-funnel solution.

So anxiety could be addressed by illegal drugs, legal drugs, meditation, any of those things. So they need to be in the mid-funnel place where someone has learned that they have anxiety, they've learned that there's different options for treating it. And from a mid-funnel standpoint, they bring it to the bottom of the funnel, which is, Get on the phone with us and we can solve your problem of anxiety with therapy. So brands need to pivot. I haven't seen, again, many companies really pivot to this. They're dabbling in, again, AI visibility tools or going onto Reddit to get citations. But I think we need an entire pivot around a future user that could be using ChatGPT, could be using whatever Google comes out with — their glasses — but that's the changing future.

Mike Duboe:

Eli, thanks for doing this. It's always a blast. Let's do it again in six months and see where we were wrong. (laugh)

Eli Schwartz:

Absolutely.