A Deep Dive Into Marketplace SEO With Ethan Smith | EM Group Chat #176

Welcome back to Everything Marketplaces, where we talk with founders and leaders from some of today's top marketplaces. So this is episode 176, which was really a great group chat that we just had with Ethan Smith. Ethan is a leading, if not the leading mind when it comes to SEO for marketplaces, and is currently the CEO of Graphite, which is a boutique growth and SEO agency. Ethan has worked with leading marketplaces like Thumbtack and Upwork on their SEO strategies, and also teaches the SEO strategy course at Reforge. So this is really a great chat with Ethan where we dove right into SEO. Started off by covering why marketplaces are uniquely positioned to leverage SEO as a growth channel, an overview of the main types or buckets for SEO, examples of SEO playbooks and strategies that leading marketplaces have used to drive growth. Ethan shared lots of actionable tips for founders looking to get started with SEO, some of the most common mistakes made when implementing some of the strategies. We discussed the future of SEO with AI and also had a great group Q&A. I really enjoyed this conversation and you're going to find it a great watch to the end. So Ethan, as I mentioned to you right before we started recording, you know, you're of course a leading mind when it comes to SEO for marketplaces and one of the most requested group chat guests on the topic. So I'm really excited to have you join us here today. I wanted to start off by saying a huge thanks for taking the time to do so in advance. So before we dive into SEO, I think it might be great if you can start off by briefly sharing your background. For those here that might not know you.

Sure. Thanks for having me. So I'm Ethan, I'm the CEO of Graphite. We're an SEO agency and we work with a lot of companies like Notion, Webflow, Thumbtack, previously did work with, so Thumbtack's a marketplace, Ticketmaster. So done a lot of marketplace work.

Yeah, definitely. That sounds like it. So I guess, you know, what initially led you to SEO? And I guess when it comes to marketplaces, it sounds like you've worked with a lot of, you know, leading marketplaces that have been able to leverage SEO as a, you know, growth channel. So, you know, what makes marketplaces uniquely positioned to be able to leverage it?

What makes marketplaces uniquely positioned is number one, the addressable market's usually very large for a marketplace for SEO as a channel. So the channel has substantial opportunity. And then the second thing is you can auto-generate or do programmatic strategy and especially if you have a lot of SKUs. So the more SKUs you have, the more pages you can auto-generate and the more content you can auto-generate. There's many important execution details to actually do that, but you have all this stuff versus a company like MasterClass where everything has to be done manually.

Yeah, we'll probably get into the execution here, but I guess, you know, it might be helpful too. Are there maybe specific types of marketplaces? Like, I know you mentioned like kind of having like a high SKU count, but maybe, you know, say for instance, like a local services marketplace or, you know, products and goods marketplaces that might be a you know, positioned to leverage it more so versus others?

Not really. Essentially everyone can leverage, every single marketplace can leverage SEO. The exception is small addressable markets. So if you have a very niche audience, like I was talking with a company that was doing a job marketplace for oil rig workers, the addressable market's not that high. And so that might not make sense there, but anything like commerce and travel and local, all of those have a lot of opportunity and the more SKUs you have, the more it makes sense.

So I guess like as founders, you know, what are some maybe steps that we can take to see if, you know, to actually kind of research, you know, the size of the addressable market?

The main way is to find other competitors and go on SimilarWeb and look at their traffic for non-brand SEO. So when we do a project, we'll just take a category. Let's take commerce, for example. If I'm an early stage commerce company, I'm not gonna look at Amazon and Walmart because I'm not gonna actually be able to compete with them. But I might look at, you know, I would do 3 tiers like high, medium, low. Based on authority and then look at the traffic for whatever tier I belong in authority-wise, and then look at their non-brand SEO traffic in SimilarWeb, and that'll be a good proxy for my opportunity.

That's going to be super helpful. So, so I guess before, you know, we jump into some of the kind of specific strategies and also to, you know, examples of playbooks that marketplaces have used, I thought it might be helpful. Can you just give us like a high-level overview of the different types of SEO and maybe kind of break it down into the buckets?

Yes. So I want to reimagine what actually the types are. So the typical types are programmatic, editorial, and technical, which I don't actually agree with. So, but I'll define those and then I'll redefine what I think that it should be. So programmatic would be programmatically generated pages like browse pages and product pages. Editorial would be articles and then technical, quote technical is sort of this loosely defined thing, but that would be the infrastructure, like your internal links and your metadata and things like that. I think that this is not actually the right way to think about it. So I think that we need to decouple programmatic and editorial, which is how content is made with the page type. You can have a page type with programmatically generated content or editorial content. Any of those page types could have programmatic or editorial. So I I think instead we should start with the page type and then we should talk about how the content is created on the page. The page types, these many different page types, the traditional marketplace page types are browse pages, which would be, you know, red shoes, product pages, which would be an individual product. And then there's a bunch of other kinds of page types beyond just article. So articles usually are these high volume, extremely low intent pages. that sort of float over in the blog and you get a bunch of visits and they don't convert, which is not what we want. The other page types that I think are very interesting are listicles, which is sort of an editorial form of a browse page, a product review page. So a lot of times product queries, product review pages outrank the product page. There could be how to do X, like how to perform this task or how to solve this problem, tools, templates, all these different kinds of page types. And all of those page types could have programmatic content, they could have editorial content, human written content, they could have AI content. But programmatic editorial is how the content is created. But we start with what the right page type is.

Yeah.

And so I guess, you know, when we think about, you know, programmatic and then editorial, is there one maybe that, you know, might make sense to kind of start at an earlier stage versus the other?

I think I would start with the highest intent ones. And usually for marketplace, highest intent and lowest scope. Usually that's listicles. So if you took an example with Thumbtack, let's say that they didn't have anything at all. I would start with listicles. So plumbers in San Francisco. And the reason I would start with listicles is because the scope is not that high. The intent is very high and the traffic potential is also high. So if I'm Thumbtack and I have a limited, you know, I'm like 6 months in and I just raised a seed round. I would go manually create for the top 10 cities, plumbers in San Francisco. And I would, you know, go into Webflow, no backend, manually select some of these. Don't worry about building browse pages because then you need a whole search engine and then just manually create 10 of these listicles. The listicles are probably gonna outperform the browse page. They're not gonna cost that much money. There's no backend and it's really high intent, really search volume. So that's how I would start.

And so you mentioned as far as, you know, are these listicles kind of based off like location and kind of categories or?

A listicle is literally just a duplicate version of a browse page, but reframed as an article. So literally, and we did a test with one company where we literally just recreated their category page, browse page as a listicle with the exact same set of SKUs and the listicle outperformed the browse page. And, you know, at scale you can actually auto-generate listicles, but that's what I mean by that.

No, it's gonna be really helpful as far as, you know, the kind of first steps for getting started on this. So I guess if we, you know, this is kind of like the earlier stages, but, you know, say for instance, you know, kind of like a little bit later stage, you know, what do these kind of listicles look like in that kind of editorial, I guess like the volume of content that some of these marketplaces are kind of producing?

Usually hundreds, not thousands. So, you know, Wirecutter is a good example. How many Wirecutter articles are there? There's probably 1,000, maybe 2,000. There's never going to be tens of thousands.

Got it. And I guess, you know, so we don't get, you know, go too deep in as far as, you know, editorial and kind of, you know, zoom out and think about like the overall kind of like SEO strategies, you know, could you maybe like give us some examples of, you know, marketplaces and maybe we could use, you know, Thumbtack as an example of just like their overall kind of SEO strategies and maybe the playbooks that they use?

Let's do it. So Thumbtack is a good example. And, you know, most examples that people know about are larger companies like a Canva, and those strategies will not work that well for earlier stage companies. The reason why Canva is able to do well, they have the core strategy down and they have tons of authority and their on-page strategy does not need to be 10 out of 10. It could be 7 out of 10 and they'll still beat everyone because they have so much authority. So most of the playbooks for programmatic are for very large companies with super high authority, which will not— and earlier stage or even mid-stage companies will not succeed with the exact same playbook. So I'll describe Canva and then I'll describe the add-ons that have Thumbtack or that other people would need. So for Canva, they're one of the most successful marketplaces and they have browse pages for templates. So invoice templates and then there's user and they have thousands of these. So there's user-generated templates and then there's a grid of the templates. There's links to other browse pages and then there's page-level editorial content at the bottom. And that's like the core structure of any browse page. And that may or may not work for most companies. And for Thumbtack, it probably, you know, maybe they're ranking number 6, but they're not gonna win. But earlier stage and mid-stage companies can beat a Canva with better on-page. You'll never beat them with better authority, but you can beat them with on-page. So I'll describe what we did with Thumbtack. So the first thing is that for a Canva, you have a grid of, let's say, 20 different templates. But if I'm Thumbtack, or if I'm, you know, Thumbtack for, you know, a Canva competitor, there's a few ways you can do that. The first way is that on each item in the browse page, you can add words and you can add metadata. So for Thumbtack, there's a list of the services that the plumber provides, and that's keyword-rich content. And then the second thing is in the early days we would do snippets. So, and like Yelp and TripAdvisor have snippets, Yelp search pages, you'll see have review snippets. So adding these snippets adds lots of keyword rich data for free. So you're programmatically selecting these, you don't have to spend money on that. So for Thumbtack, let me know if there's too much detail, but for Thumbtack, we started by manually writing snippets on each pro and we basically hired an army of snippet writers. And the reason why is because you can't show the same string of text for the same pro on 100 different pages. Every single snippet needs to be unique on that page. So, and it needs to be keyword rich. So we basically created an army of snippet writers following a guide of like, write, you know, keyword rich data, include these keywords. Each snippet was unique. And, you know, typical programmatic strategy would say, well, it has to be automated. You cannot manually do stuff because it doesn't scale. It actually does scale. So we got it down to about $0.25 per snippet and the number of pages that you care, actually care about is gonna be about 1,000. It's not gonna be a million. Even if you make a million, there's gonna be 1,000, 2,000 that actually bring in all the traffic. So any of these programmatic strategies, manually doing stuff actually does scale. So we did, we did manually written snippets and then as we had more reviews, we removed those and then we selected reviews. The reviews also, you cannot show the same review snippet on every single browse page when that pro shows up, just like Yelp, same restaurant's going to show up on many browse pages, but the snippet, the review snippet has to be unique. So then you can, for each individual pro, you can automatically select, let's say 50 different snippets from their 10 reviews and then have different snippets show up on different pages. So that's thing one. The second thing is internal linking. So I, early in my career, Search Console used to have internal link counts, actual internal link counts. They have fake internal link counts now, but they had real internal link counts. And we had, I was at a shopping comparison company. We had like 5 million product pages. And not many of them got traffic. And so I went in and saw that the pages that didn't get traffic had no links or had like one link, one internal link. So then we added a section that randomly linked to other products, which is not great for users, but we randomly linked to products and now every single product got links and traffic tripled. So then for Thumbtack, we did a more principled approach, which was like a related but spread links. And because a bunch of the browse pages got 0 links, 1 link, 2 links, because they had facet-based links like other service categories and city, other cities for same service category. And what happens is there are a bunch of browse pages that didn't get links. So we added this module that spread the links more and traffic went up 25%. The last thing was title tags. Happy to go into more detail on title tags, but those were the 3 key things, which is the on-page, Content, the internal links, and title tag optimization.

That's super interesting. So I think one of the things that we're always, you know, our guests kind of comes up often here in the community is, you know, how to kind of measure their ROI for SEO kind of in the earlier stages. And I know you mentioned some of the, you know, even, you know, some of this can actually be done, you know, manually, of course. So, you know, how do you think about that, you know, measuring ROI and the kind of timeline on that as well?

Mainly, you can think of it as a lifecycle. So first thing is, let's say that we create 1,000 pages. So first thing is, did they get indexed? You can check that by looking at Search Console and make an XML sitemap. It'll tell you how many pages are indexed. So is it indexed? Second thing is, is it getting a little traffic? And then third is, is it getting a lot of traffic? And then fourth is, is it actually converting users? And at each point in that lifecycle, wherever the fall-off is, there's a different intervention. So if it's not getting indexed, maybe it's not linked to properly. Maybe the tags are sort of incorrectly If it's indexed, but it's not getting it, it's not getting any traffic at all. Maybe it's the wrong page type. Maybe the on-page is not good. If it's getting a little bit of traffic, but it's not actually climbing and getting a lot of traffic, maybe it's low search volume, the content's not optimized, or you just don't have enough authority. And then finally, if you're getting a decent amount of traffic, is it converting? So the main success criteria that I would be looking at ultimately would be visits to start. And the reason why is because for a lot of these pages, they won't convert well until you add conversion optimization. So the most common mistake that we have in our projects is we get a bunch of visits to the blog, let's say, and they're not converting. And then we stop, we stop working on it when, if you did conversion optimization, actually would work. So the main success criteria we do is visits, and that's pretty straightforward, just looking at Search Console or GA. And then once you're getting traffic, Start working on conversion optimization.

Yeah. So I'm glad you actually mentioned, you know, one mistake there, but I guess, you know, what are some of the other kind of common mistakes that you're seeing marketplaces make when it comes to, you know, implementing some of these SEO strategies?

Yeah, mistake one is overscoping what you need to build. So I mentioned before, if you're starting out, you don't have to create a whole backend to create a million category pages. You could just manually create 20 of them in Webflow. So we use Webflow a lot. You can actually build stuff pretty quickly in Webflow, even search pages, browse pages you can build in Webflow. So overscoping it and having a really elaborate backend, make 20 pages 'cause you don't know how much traffic you're gonna get. You might not have enough authority. So if you don't know, don't build the whole thing, build 20 pages, launch them, and then if they get traffic, then build the other 2,000. and, you know, build an elaborate backend. Building 20 pages is, you know, very inexpensive. So start with a low scope thing to get signal and then decide to invest more after you've proven that those 20 or those 50 are actually performing. So that's mistake one. Mistake 2 is over-indexing on quote programmatic and AI content. So programmatic, you know, again, in the Thumbtack example, we look at Canva and they don't really have much content on their page and they're auto-generating all these and they're working or like Zapier Sheets plus Gmail works really well. It's not going to work for most people because the on-page content is not optimized enough and you don't have enough authority to get away with not having a lot of on-page content. So for a thumbtack, yes, go ahead and manually write snippets and that will scale. Don't worry about scaling to millions of pages. You don't need millions of pages. So, and then AI content. AI content can work, but you can't just say write an entire page and it's gonna work well. You need to heavily configure the AI with like unique metadata or a human editing pass at the end so that it's not just a straight ChatGPT response. And then the third is shutting the program off. You got a bunch of visits and they're not converting and assuming that you can like It's, you know, a common thing is it's 0.001%. And if it went up 50%, it would be 0.0015%, which is zero. So if you do conversion optimization, it's not 50% greater than 0.001%. It's like 0.001% to 1%.

Awesome. Yeah. So you mentioned AI as far as, you know, for generating content. But, you know, when it comes to some of the new services, so for instance, you know, Perplexity and ChatGPT, How are you seeing this impact SEO and what do you think some of the implications, you know, will it have moving forward?

Yeah. So 2 short things. Thing one is I think these surfaces will be huge and it'll be a new channel and it's going to be very large. The second thing is that it's basically an evolution of SEO. It's not a completely different thing to optimize for them. So we, I posted some stuff on my LinkedIn about some stats around the growth of usage of these new AI services. And it looks like ChatGPT is, I think, about 85% market share or something, you know, almost. Essentially, you could just optimize for ChatGPT today and you'd be fine to ignore the rest of them. I think that might change over time. Like Google was not the biggest search engine at the beginning. It was Excite or AOL. And, you know, then that stuff changed. So we don't know if ChatGPT is gonna be the biggest surface and have 85% 5 years from now, but right now they have 85%. So if you're optimizing today, just optimize for ChatGPT. Second one is Perplexity, which I think is like 8% market share, and then the rest are really tiny. So there's definitely growth there and we're seeing conversions from there. And most of the impact from chat search and AI chat and search is hard to measure because if you ask ChatGPT what's the best credit card, you're probably not going to click on the citation. You're probably just gonna, and it says, let's say it says, you know, brand new card like Mercury Bank. Okay, great. I'm gonna go open a new tab and type in Mercury or I'm gonna search for Mercury in Google and I'm not gonna know that that came from ChatGPT. So you just look at referral traffic. It's undercounting the effect. So I would ask if you can ask post-purchase questions like, how did you hear about us? You'll probably see much higher percentage coming from ChatGPT and Perplexity. And then the third is how to optimize for these new surfaces. And the short answer is it's SEO version 4, or, you know, it's like an evolution of SEO. Most of the questions that you ask start with a search. So you want to optimize for showing up in the search like you would for Google. And then the chat summarizes the search results. So each of these surfaces will have their own search engine. They're going to be maybe 80% similar to Google, but there's gonna be another 20% that's different. But you optimize for all these different search engines for the chat, number one, and I can go into detail if you'd like. And then the second thing is that spycation optimization. So the chat is looking for consensus. And so even if you show up all the time with one result in the search results, you also need the other citations to mention you. So if I want to be the best credit card for businesses, I want my page to show up and I want the 10 other citations to all mention me. So it's kind of like a backlink, but a new form of a chat backlink. So for ChatGPT, let's say the first thing is we need to know the questions that actually matter. If we optimize for a question that no one asks, it doesn't matter. Just like if we rank number 1 for a search that no one ever searches for, it doesn't matter. So we need to figure out which questions are actually being asked a lot. And in SEO, you can get this information of keyword volume from the Ads API, but there is no truth set API for any of the chat services. So there's no data, there's no actual data. first-party data that we can get. So you need to guess what the questions are. And we're building, we're building a bunch of tools to do this. This doesn't really exist yet, but essentially what you can do is you can take the keywords that people are— this Google search keywords, and you can transform those into questions. So, you know, an easy way that people on this call can do this is go into Ahrefs, whatever keywords you want to rank for, try to transform them, try to manually transform them into a question. This is probably going to be pretty similar, the relative search volume for questions as for keywords. Okay. So step 1 is selecting the popular questions. Step 2 is tracking. So who's showing up for these questions? We built a question tracking tool that's not too expensive. So SERP So question tracking is the new SERP tracking rather than your position 3 for this keyword. Instead, you are, you are not showing up for this question. So we have a tool for that if anyone wants that. But then you want to look at your, how frequently you're showing up for a given question. It's like how, and when you ask ChatGPT best credit cards for business travel, it'll give you different answers each time you ask it. So you need to ask it multiple times. You need to ask variations of the question. You need to check perplexity. And so it's sort of a share of voice versus you rank position 3. So question tracking. Then you want to look for the patterns of what's in the answers. So if you ask best credit card for business travel, there are specific themes like the signup bonus and the APR and something else. So whatever themes of the content and the answers, you should have that on your page so that the chat is easily finding these things. And then 4 is citation optimization. So again, it's great if we show up all the time with our page, but we need all the other citations to mention us. So we need to know who the citations are and not just Forbes, but Forbes specific URL that is the citation. And then try to be on those pages somehow.

I'm gonna have to try to kind of summarize this and break it down so I can share it with founders after. So yeah, I know you mentioned a few different tools, but do you have like a kind of like a basic toolkit, you know, for founders that they can use?

My short answer is Google Search Console, which is free, Ahrefs for competitive analysis and keyword stuff, SimilarWeb, which is free. And then a content scoring tool like a Clearscope, MarketMuse, Surfer. We also have one.

Yeah. So we've covered quite a bit here and I think, you know, a lot of actionable kind of tips and, you know, even kind of specific strategies, you know, for founders to kind of, you know, use and kind of put into practice at the earlier stages. But, you know, when it comes to hiring for SEO, you know, I guess like what stage do you think it makes sense for, you know, founders to hire and, you know, what should they look for when they're hiring for SEO?

I think you first need to figure out what your strategy is gonna be and hire for that. And usually your strategy's not gonna be super technical. So, and the technical SEO people are the, there's the fewest of them, they're the most expensive and you frequently don't need them. So for Masterclass, for example, no one had any SEO experience and, but they were generally really good growth people who measured things and whatever we needed to build, they could build it. Like they could scale an article whole editorial team really well. That's what, that's the kind of person I would hire over the traditional SEO person. If you are building Zillow competitor, like maybe you do need a person with domain expertise, but a lot of times just getting a really good growth person who maybe they're good at paid marketing or maybe they're just a good growth generalist will actually probably outperform the traditional SEO person depending on what your strategy is. And especially if your strategy is heavy on the content side and light on the you know, programmatic side, you do not need a programmatic technical SEO person. So, and, you know, we're actually like when we hire people, we've sort of run out of people with a lot of domain expertise and we're trying to create SEO people. And so the way that we're doing that is to hire really smart generalist marketing people Or people with, you know, like some of our best people have done a little bit of SEO, but they've worked in multiple marketing functions and they're good managers and they're good at building things. And then we give them our playbooks. It takes them 6 months or so to learn those playbooks and now they outperform most SEO people. So that's the background I would look for.

Yeah. So actually I want to jump in here and just go ahead and ask a question that a founder wanted me to ask in advance. And it was around using platforms such as Webflow. You mentioned for marketing pages and for blogs and so forth, and also using it on a subdomain. So do you, I guess, does using a platform such as Webflow on a subdomain significantly impact SEO? And then the second part to that they actually wanted me to ask was related to it. How do you unify analytics and tracking when using a subdomain?

My short answer is never use subdomains. For whatever reason, Google does not share authority on subdomains like it does if it's on the root directory subdirectory. Why? There's no good reason why, but that's how the algorithm works. So I would just never use subdomains. It'll slow you down and you won't reach full potential. And I've seen this multiple times where we'll start on a subdomain, traffic will be low, it'll take a really long time, and then you change it to the subdirectory and it triples or quadruples. So the impact is substantial. And it's actually not that hard most of the time to do a subdirectory. So for Webflow, you can just do a reverse proxy and make a, make it into a subdirectory. If you have your main web app and then Webflow, and that's in React, and then your Webflow is over here on a subdomain, you can just make it into a reverse proxy and you can add any Webflow instance to any site and have it be a subdirectory. So I would just never do subdomains and the scope is not high. The scope is like 3 hours or a day. It's, it's, it should not be that big of a deal. And then tracking and analytics is solved for you. Cool.

Hey, Amelia, do you want to come on?

I would love to. My question is, what are the most important tactics to pay attention to for editorial content in competitive categories? Like specifically if you're getting articles published that are important to the topic, but you're going against competitors with like really high DRs, Is there anything that you can do to jump up the SERP and get over them? We've played around with featured snippets and stuff, but yeah, if you're competing against very established competitors, what can you do to get editorial content to rank better, more frequently, et cetera?

Awesome. And do you have a specific key search or topic in mind that I can speak to?

For one example, MBA, where the scope of articles is actually somewhat constrained and there's MBA.com and Poets&Quants and these established players that have been in the field for a long time.

Yeah. So there's a few things. The first thing is that you might not, you don't, you're not going to have as much authority as, yeah, MBA.com, but you might have a lot of topical authority specific for specific schools or for maybe, you know, executive MBA or like you might find some niche that you're known for that you're going to outperform on. So whatever that is, and you can find that by looking at your keywords, We also have a free topical authority tool, which I can find and put in the chat, but find where your topical authority is, where are you ranking right now? And then double down on stuff that's adjacent to that. Topical authority is essentially your authority for a particular category over total authority. So if all your authority is all about Stanford Executive MBAs, but your total domain authority is not that high, your ratio's really high and therefore you're going to rank for that because you're going to be a niche expert. That's, I think, one. So leverage your topical authority. The second thing is the content score. So look at your content score. And the common mistake is for Stanford MBA to only look at Stanford MBA and not look at the other 50 variants of the keywords that you're targeting. So there's probably, probably like 300 different search, popular search keywords for the Stanford MBA topic. And you're not, you're only scoring on the head term. So I would do a clear scope where we have a tool for more of the keywords in that keyword group because no one else is looking at that and you're basically ignoring the whole mid and long tail. And what Google's doing is they're looking to see how many of these subtopic questions have you answered. And if you're the only one that knows the subtopics that are in the mid and the tail, then you're the only one that has that and your content score can beat an MBA.com. And then the third thing is title tag optimization. So if your landing page is how to get into Stanford GSB, I would go see how, see where that's ranking in Search Console, go into the keywords that you're ranking for. And there's probably gonna be some themes with any of those keywords with high impressions where you're ranking number 8 and you're not mentioning it. So an example of that would be how to get into Stanford. And then it might say like interview questions or templates or stuff like that, and it's not in your title tag and you're ranking number 80. So you might rank for it if you mention it. So put in interview questions. So how to get into Stanford MBA, interview questions, templates, examples, and then you immediately start ranking for those things. And we've repeatedly seen projects we've worked on outperforming huge high domain authority sites by doing these 3 things. So you, you can do that if you have enough, some sufficient amount of authority. And then, yeah, Louisa just posted in the chat the topical authority tool that you can use.

Super helpful. Thank you.

Appreciate it. Hey, Dimitri, do you want to come on?

Thanks, Mike. I have a question for you about the best way to quantify how well, like, a set of SEO optimizations is performing for the marketplace. And if the answer is experiments, can you talk a little bit about how you would set up the experiment? Would you randomize at kind of These canonical URL pages, which you do a geographic experiment where you optimize all the pages in a single, like, say Cleveland, and leave like the rest of the country alone. And how long do these typically experiments take to run?

Yeah, so I think Thumbtack would be a good example. Like, how do we add— adding a content block? Thumbtack has 100,000 pages, plumbers in San Francisco variants, and we want to add a content block. And we want to see what happens. Is that what you mean? Cool. Yeah. So some things— oh, okay. So title tags for that, I would just randomly select buckets. However, the common mistake with this is to include URLs with no traffic in your buckets. So let's say you have 100,000 pages and you split test buckets of 10,000, but each of your buckets there's actually only 10 pages in your 5,000 URLs that are getting the traffic. So you have all these noise URLs and then let's say that one page in your bucket dramatically increased and then you say, okay, well the, you know, it's very statistically significant, but it actually, if you look at the URL level, it's one URL that went up. So I would, for a thumbtack, you know, the earlier you are, the more that you have this issue. So I would manually go in and probably like randomly select buckets, but sort of clean them up, remove the URLs that have, or, you know, maybe do like high traffic, medium traffic, low traffic. We did this for a company where we did popular city, popular cities, LA, New York, mid-tier and lower tier, because there might be some different effects. So we bucketed them into different cities. We only selected URLs that had traffic and then we tested title tags, let's say. And then we looked at the results, but not only the totals, like the traffic went up in this bucket more overall, but we looked at the individual URLs to see if there were one or two that were skewing the whole bucket. And then you'll have cleaning data. So that would be a split test. And that is, so for title tags, you get results in 1 to 2 weeks, which is short versus something like adding a content block, which actually takes a longer time. So that I would do 1 to 3 months. Something like internal links would be like a month. So it's these 3 different types, which is short-term wins. Short-term wins would be typically title tags or like rich snippets. Medium is going to be things like internal links. Longer term is gonna be content changes. And then the 4th bucket would be, I'm adding new pages. Like I'm, I'm gonna add 10,000 new URLs in buckets. That would be like a month or so.

In your suggestion to remove the zero traffic pages, isn't that problematic in a sense that the lessons that you learn from your high traffic pages may not apply to those zero traffic pages? And so you don't know, like, if it's worth optimizing those at all, or if you just sort of abandon and Never come back to them.

Yeah. And for the zero traffic pages, I said you should remove them. I'll modify my answer. I don't think you should remove them. You should just separate them into a separate bucket because they might actually get traffic eventually. So I would test the zero traffic pages and for sure the changes for the high traffic ones might not carry over to the low traffic ones. So I would test those low traffic ones as well. And for the low traffic ones, you know, they're low traffic either because there's zero search volume and you could never perform Or it's that there's a lot of search volume and you could perform. So then you'd wanna test, how do I get from position 80 to 15 to 5, but separate that into a separate experiment so it's not noise within the, you know, the main bucket.

Fantastic. Thank you.

Appreciate it. Taran Deep, did you wanna come on? I know you sent me a question.

Sure. Hi, Ethan.

I wanted to ask you about the use of JavaScript to render on page.

Does that still inhibit the ability for crawlers to access content? And then does it inhibit maybe rankings from the lack of being able to assess the quality of that content? So just wondering your perspective on that. Short answer is no, it's totally fine. Here's the longer answer. So it depends on how you implement it. If you implement stuff server-side, there's no issue. That's all fine. If you're implementing stuff client side, it depends on how you're doing it. So if you're progressively loading stuff on the page, Google might not see it. And sometimes what'll happen is, and also if you're doing a combination of server and client, that can cause problems. So Google does 2 passes of crawl to your site. The first is a without script server side call. And the second is a client-side JavaScript call. If Googlebot thinks that it's got all the information on the server-side crawl, it will not— sometimes it will not crawl the client-side crawl. So I'll give you an example of that. It's common to have a product page where the photo description is loaded server-side, and then the reviews are loaded client-side. And Google looks at the page without the reviews and thinks they got everything that needed and ignores the reviews. Does not do the client-side call. So when you have hybrid rendering, the client-side stuff might get missed. And you can verify that by doing site:URL and then do, take a snippet from your client-side module and then put it in quotes. So site:product page URL. and then quote review snippet quote. And then if there's no, if Google doesn't have it, then it didn't find it. And if they do, then it did. But the hybrid stuff can cause problems. And then depending on, you know, if you're progressively loading in multiple modules client side, it's possible that Google is also seeing some but not all. So you could do the exact same check, site colon and then URL and then exact match snippet. in quotes from each of the modules, and then you can see whether or not it's getting crawled.

Appreciate it. Hey, Amelia, do you want to come on again? We'll get this in as one last question.

Awesome. To what extent, Ethan, do you think Google's EEAT updated guidelines with the expertise point? To what extent has that impacted unique content? To what extent should SEOs at companies be prioritizing unique content? And yeah, what are the thoughts on Google's EEAT guidelines these Yeah.

So there's 2 themes. One is personal experience and then the second is information gain and uniqueness. So they added personal experience and I don't have strong data on this yet, but I have strong opinions. So I think that we know that Google's looking at information gains. Information gain means uniqueness. We just don't know the weight yet. But What'll happen is if everyone's using these content scoring tools, these content scoring tools are saying your page differs from every other page in this way. So here's how you can make it more typical. So then everyone sort of has the exact same article and everyone's converging on the same content, which is not ideal. So a way to break out of that is to add some unique thing. And we know that there's a score for that, but we don't know the weight, but the kinds of uniqueness that you can add to a page That I think probably will have an effect are personal experience. So if you have an elopement page, rather than just describing generically the— rather than just describing elopement in Houston, add something like, and I eloped, or my friend eloped, and this was their experience with that, and use first-person pronouns like I, we. So that would be one, which is a personal experience. The second would be a quote. Like from an expert. So we interviewed Dr. Whoever and they said this about eloping or that, you know, we talked to a professor at Stanford and they gave this tip about applying to Stanford. Unique metadata. So if you have unique metadata, you can add that. These are all forms of uniqueness and most companies are not adding uniqueness. Most companies are optimizing for more typical, so they're not optimizing for information gain. And then the second part of that is experience. And I think part of why they added experience is because so many people are programmatically generating content and doing AI content, which lacks personal experience. People want personal experience. So I think that the weight of a personal experience is probably gonna— is probably higher and probably will increase over time because that's the only way you can break out of everyone else who's doing programmatic and AI content.

Awesome. Well, that was a great last question. And, you know, Ethan, this is such a good chat. So we really appreciate, you know, taking time to join us today. And I know we covered quite a bit and, you know, tried to pack it in. And so we really appreciate it. And also to, you know, helping answer some of the specific questions that, you know, we had here. And, you know, actually had, you know, I guess, you know, one last question for— usually it's part of my kind of parting question, which is, you know, if you could go right back to before you entered the world of marketplaces, you know, what you would tell yourself about them. But I guess here, you know, if we kind of, you know, slightly reframe that to SEO, you know, what would you tell yourself about SEO right before you kind of entered in the world of SEO?

I would say study. Rather than focusing on coming up with ideas and testing them, go talk to other people who have already done this and ask them what they did. Ask them in very specific ways. So go find someone, go find 5 other people who are doing marketplace SEO. Make sure that they're people where their traffic is actually good versus someone where their stuff isn't working because a lot of times it's not working. So find 5 sites where you can see that their traffic is growing. And it's performing well and you can see that in Ahrefs, find the person who works there, who works on SEO, ask them the most impactful things that are performing and then ask them about specific elements on the page. Like what about this? Did you test this? And what worked the best and what were the results of their tests? And they will actually tell you all of this even though it seems like they shouldn't and wouldn't, but they will. And then take the stuff that performs the best. And then test that yourself. Don't go online and read blog posts because most of those don't have actual data. Don't look at people and companies where they didn't actually cause an impact. And don't just rely on intuition because most of your intuition ideas will fail. Just take the learnings from others and you can fast-track it to impact. Awesome.

That's gonna be a really helpful parting advice here. And then last but not least, where can we keep up with you and follow along?

I post the most stuff on LinkedIn. So if you add me on LinkedIn, and I think we have a newsletter on our website and I send emails as well. So, or, you know, email content. And to my comment that I just made about getting learnings of impact and data, we actually do original research where we test the stuff that we discuss with data behind it. And we also have a blog. Maybe Luisa, you can post our blog. But we do actual research studies because we want to see whether or not this stuff works. So that's how you can find me.

Awesome. I'll include links in the recording. I know you'll get lots of founders here coming your way. So we once again really appreciate joining us here and thanks for the questions, everyone.