Episode Transcript
[00:00:00] Speaker A: Hello, I'm Jeremy Rivera, your unscripted SEO podcast host. I'm here with Erica Breaker. Again, give yourself an introduction to the folks who didn't catch the last interview. And if you didn't watch it, it's available on unscriptedseo.com and yeah, take it away.
[00:00:17] Speaker B: All right, awesome. Thanks for the intro. I'm Erica Brager. I'm director of strategy at 10 speed. I oversee the strategy team and innovation.
I've been in digital marketing a little over 10 years at this point. Most of that time has been spent in B2B SaaS, SEO as well as content.
[00:00:39] Speaker A: Excellent. I'm curious, what's the most consequential facet of SEO right now?
[00:00:49] Speaker B: Could you clarify a bit?
[00:00:51] Speaker A: Well, I mean, we have different disciplines. You know, we've got, you know, content side, we've got link building, we've got technical SEO, we've got conversion rate optimization, interacting with other teams.
What is the most consequential element or fraction of SEO to focus on?
[00:01:13] Speaker B: Or I would say speaking broadly, machine readability. And if we are were to zoom into that, it's technical SEO, because if a machine cannot easily parse and understand your site, then it's going to be much more difficult for you to show up in aios, you to show up in AI mode in organic search and LLM search.
It really spans much further than organic these days.
[00:01:46] Speaker A: What's the most common pitfalls that you're seeing companies or sites fall into where they have holes in their clothes that they weren't aware of?
[00:02:01] Speaker B: I would say the number one is page speed as well as technical debt that has ballooned from URL Updates.
So with PageSpeed in the past, I witnessed clients kicking it down the road. Kicking it down the road. And now we've reached the point in machine readability where if the page is over a certain size, it's much more difficult for the machine to read it. And if it's over a certain size, it's much more difficult for it to load on a mobile device, for it to load on desktop, and it could really fracture the user experience.
And that just, it's a snowball effect.
And then as far as URL updates, we actually ran into this with a client this week.
They sunsetted a blog, but it wasn't redirected. And because it was so frequently linked across the site, it created 44xxs. And I've witnessed this with other clients in the past where you sunset one URL, you update one URL, but you, you don't redirect it, it creates 33xxs, 34xxs and that can very quickly balloon the more URLs that are updated.
So we've really gone back to this fundamental is the site fast? Does it load well on a mobile device and does the site have strong technical health?
[00:03:31] Speaker A: What are some of the indications that you point to if a client's like does it really matter, you know, how fast my site is? Are there any studies that used to underpin that or any statements from Google or any statements from LLM tools that you kind of surface and say hey, this is evidence X, this is evidence Y. This is some of the citable reasons why we focus on pagering or page speed.
[00:04:05] Speaker B: Sure. With PageSpeed, a lot of what I'm doing is a comparison. It's a combination of using Google's guidelines for or user first content, user first for the site.
However, we also have to factor in that a machine needs to parse the site in order to present it to the right user. So if a client is struggling to maintain their position in an LLM to maintain their position in the SERP or they've lost it, that's when I'm doing a comparison across the competitors to say, okay, the page speed on your site is significantly slower and these other two, three competitors have pulled ahead of you. And this is the biggest difference that I've been able to identify.
[00:04:53] Speaker A: I like the Walmart study that showed their process of optimization for conversion was tied heavily to load and site speed.
There was an improvement of large improvement in conversion rate when they got to two, two seconds at a minimum. But they were also able to release statistics that showed even beyond every quarter of a second faster below that that you could reduce it, there was even more conversion rate benefits. In other words, the curve of benefit did not decrease with the decrease in speed. So even as you approach near instantaneous, there's still money to be uncovered or recovered however you want to view it. Like for their type of, for E commerce at least.
You know, obviously not every site's need E commerce but it does, you know, I think it does carry over, you know, so you can take that lesson 100% directly for, you know, if you have the E commerce sites you can take that and say hey, here's the Walmart study and conversion rate optimization. So if it's that important to users that can draw literal cash out of them faster because it loads faster. I mean there's an intuitive level to it, but you know, there's also like a functional aspect to it for behavior analysis of like, if somebody's, you know, waiting on a phone, they're more likely to disconnect.
They might think that the page is broken or buggy. And so whether or not you're selling an E Commerce product, you still are losing that percentage of users who might have gone deeper, who might have clicked through further into your funnel and converted.
[00:06:52] Speaker B: Absolutely. And I am guilty of bouncing when I'm shopping online and the site is loading too slow and I'm unable to see the product, the description, the category, whatever it may be. And that's a, that's a singular transaction for the business. Where in B2B SaaS we're dealing with monthly recurring revenue, annually recurring revenue. And that's something that's like duplicated over time. So we want to always be putting our best foot forward in user experience. And that's a big part of it.
[00:07:26] Speaker A: Would you say that the biggest culprit of slow load times is images?
[00:07:34] Speaker B: Images is a big part of it.
I've seen where images are incorrectly sized and then that affects the loading of the page and create layout shifts. Third party JavaScript Very often I see third party JavaScript firing before user facing content that can fire after.
Those are two of the big ones that I see. I'm sure there's many more. Also videos. Videos, if they aren't set to an iframe or they're autoplaying or motion graphics, those also tend to slow down the site.
[00:08:18] Speaker A: Do you suggest adding in lazy load so the first, you know, that first square of a screen like that prioritizes its load first and then the rest of the page. You find lazy load helps that or is it still, you know, because it you want the whole page to be digested whether the first part of the page, you know, happens quickly, you know, if it's still a monster of a page, you still might have problems with proper crawling, right?
[00:08:49] Speaker B: Yes. So you have to be careful with lazy loading and make sure all of your lazy loading is happening below the fold. I've seen where a blog image like the hero image of the blog or a hero image of a core page is set to lazy load and that can really scramble things. So you want to prioritize the loading in the HERO and then you can lazy load below the fold. And I have seen that make a big difference in page speed.
[00:09:20] Speaker A: Any thoughts on it?
Extra steps like leveraging WebP to serve your images in a faster format or like a CDN through cloudflare for image delivery from data centers that are closer to making A difference in those load times.
[00:09:42] Speaker B: I haven't dealt a lot in that. What I'll typically do if I notice that it's images that are holding back the client site, I'll present them a series of options and a lot of that comes down to their server and what their devs are able to do. Because every environment is going to operate a bit differently. But I like to present to them a good, better, best when it comes to image optimization. And in some cases we compress the images.
If we're operating in WordPress, we can use a plugin to help those kinds of things along, but it's all very situational.
[00:10:24] Speaker A: Are there any other steps as far as like page length or weight overall weight the of of the page in terms of kilobytes? There's something in the back of my mind and I can't recall it at the moment, but I remember there was a change in the past couple months that Google said something about the There's a change related to the weight of the pages and I can't remember what it was. So hopefully that's enough clue for you to grab onto. Otherwise it's just a threat.
[00:10:57] Speaker B: Yes, I believe the page should be two megabytes or less.
Ideally a lot less.
It's not that the crawler is going to pass you over, but it's going to take a lot more time.
So if your page is over 2 megabytes or it's nearing 2 megabytes, that's where you want to really zoom in and say what is it on this page that is adding the most, is adding so much to this page size? Is it a video? Is it an image? Is it a motion graphic? Is it a form fill? Is it a recaptcha?
It could be a number of things. And that's where we start our decision tree.
Do we replace it? Do we modify it? Do we remove it?
And that can start to significantly reduce page size and then ultimately load speeds for the user and the crawler.
[00:11:58] Speaker A: What about, you know, PDFs and uploading other documents that could be crawlable? I know Google has a separate unique encoding system for PDFs because it's written in binary and so it has to make it kind of converts it into an HTML that it can kind of read, but it'll follow links out of a PDF that you upload.
Any experience optimizing for PDFs or has it come up on any of your clients of looking how Google indexes or doesn't index PDFs?
[00:12:36] Speaker B: I haven't done A lot with that. When I'm dealing in PDFs, I'm typically making decisions about how to present bottom of funnel content and how to bring more users to it.
But as far as optimizing for PDFs, I think you have a lot more wiggle room than you do with the other pages of the site.
[00:13:02] Speaker A: What's been a really big challenge for one or more of your clients that you've recently unlocked a solution for?
[00:13:16] Speaker B: I've been really focused on technical cleanup for a number of my clients.
There's one client that said we're cleaning up two years worth of tech debt in 90 days and I'm on this roller coaster with them. I'm sending them new things every week, whether it's to do with page speed, whether it's a 3xx or a 4xx cleanup.
But what I aim to do with every delivery of like hey, work on XYZ the first half of this week is what is going to move the needle the most.
[00:13:55] Speaker A: I am curious, what does that look like in terms of conversations? You know, I find like with larger enterprise clients that have technical debt, it usually has come from like a structured system where you know, the SEO isn't fully in the driver's seat. They might have a more complicated setup. You have to push PRs or get tickets over to developers to fix things or, or involve more resources.
What's been your experience with clearing hurdles to get SEO done that aren't SEO itself but get the challenge of getting SEO argued and budgeted and implemented because
[00:14:45] Speaker B: of the time it takes to do implementation and get buy in for it. On the enterprise side, I try to stick with the global items first, the, the big ticket items because that's what's going to move the needle. And when it does move the needle after implementation, that's where you can go to all of the different teams that this win touches and communicate the win and show how this win as we add on more and more technical implementations that that's going to help them.
So you really have to work piece by piece and you have to work with the pieces that are going to make the biggest splash first to get buy in to continue to build trust across the organization. Before you can go for, I guess the fruit at the top of the tree
[00:15:35] Speaker A: have LLM tools, AI tools, Claude, Claude code, ChatGPT being added to your stack given you extra leverage to be able to slice, dice, analyze and find these technical issues faster. I mean we've had screaming fraud for, you know, two decades now.
So technically the information has Always been available to kind of mostly analyze a lot of these technical SEO things. But is there anything that's changed in your tech stack that's made that process, that analysis process, the visualization process of these things much easier?
[00:16:16] Speaker B: It has helped a lot with organizing and acting as a second set of eyes.
Parsing through a screaming Frog crawl can take some time. You have to know what you're looking for.
If you can upload chunks of the screaming frog crawl into Quad, it's able to surface things and prioritize much more quickly than if you were to just look through the crawl on your own. I mean, you, you obviously need to make your own set of decisions before feeding it to Claude. But Claude can organize it into graphs that look nice to present to your clients, where if you were to create graphs on your own, it would take a bit more time,
[00:17:04] Speaker A: Do you think? You know, with Claude and its ability to visualize stuff, you know, I have so many MCPs and APIs available now, you know, and it's funny because, like, Google Analytics had an API for a long, long time and so did gse. But until I started getting into Claude code, I didn't think to just make it a default function for myself. Oh, hey, I need to add this so I can just pull this data in. And it seems to me, you know, Claude, like a good implementation, lets you make a very quick dashboard that can pull from, you know, clarity and GA and GSE at the same time and make it triangulated analytics dashboard that isn't beholden to, you know, a product manager's, you know, development cycles. Like, I, I give full respect to the people at Ahrefs and full respect to SEMrush and anybody who's building SaaS, especially in the SEO space.
I've worked for some, I've worked for many and consulted with many others, but they're all limited in terms of their production cycle.
And so the fact that I need a dashboard that shows all three of these channels is something that they don't care about. I might put in a future ticket, but now I can literally just put in my own dev ticket to my. To Claude and get a browser tab open to visualize this data.
Do you think that changes the nature of SaaS and SEO SaaS?
[00:18:45] Speaker B: Do you mean in terms of reporting and analysis?
[00:18:49] Speaker A: Yeah, I mean, like, if we think about, you know, and this is probably kind of existential, but if we're, you know, able to use Claude code to get these visualizations that help us specific to this thing, you know, and we have these API connections to the data, then I don't have to log into, you know, a specific tool to get that visualization. And it's almost like, you know, a lot of the SaaS tools out there are kind of like these UX UI experiences to help kind of dumb things down.
But if we're, you know, kind of moving to an age where if you're a consultant, you know what you need, you know what you want, you just need the data to be able to make that conclusion, then is there as much of an argument to even have a SASS versus a data vendor source that's doing interesting things for the service versus their feature release being some other UX or UI element within the SaaS itself? Does that question make sense?
[00:20:03] Speaker B: Yeah, that makes sense.
I believe a lot of this is removing some of the barriers to entry as well as learning curve, especially on the visualization front or even like the data analysis front. Because I remember learning a bunch of complicated formulas and then having to learn all of these different platforms, having to learn how to use Data Studio, Looker Studio, all of these different moving pieces and then figure out more so within Data Studio how to cobble everything together from all of these separate systems where now I can use Pods, API and pull everything in that I wanted to see and I don't have to use complex formulas anymore or if I do, Gemini and Google Sheets can figure it out for me.
So it's, it's enabling, it's making certain things easier. But I do think that the learning part is important in the industry and it's, it's like a little bit of a worry for me that like, I didn't have Quad and chatgpt to learn SEO.
I had Stack overflow and all of these other forums that I would dive into Slack communities.
[00:21:30] Speaker A: I do kind of worry that the positions where people would have been given the bandwidth and the time and the budget to kind of learn SEO.
You know, people, you know, small businesses are like, oh, you know, I can just get this part of my SEO done, you know, by typing a few prompts to Claude.
And you know, it seems to me that there's, you know, going to be the lack of an underclass of learners who are at paid positions who, who can learn these things on the company dime. So speak.
Because, you know, there are, you know, increased capabilities for SMB owners or teams to turn to on some levels, you know, these AI tools to execute X or Y or Z or an impression by some of these marketing managers that they can do that. But it seems longer term, you know, after a few years of, you know, many fewer people being able to actually get paid to learn SEO, you're going to have just kind of the eternal class of everybody on the outside that kind of wants to learn it. But it's very different than like if you're working at a company and you're just like, okay, well you're now our SEO person, like, this is part of your job. Go learn it and do it. You know, that's. I think that's happening a little bit less or I worry that that's happening less and the repercussions thereof.
[00:23:11] Speaker B: Yeah, I'm seeing it a little bit as well.
And I don't think it's unique to SEO. I believe, like it's a cross industry thing.
[00:23:26] Speaker A: I think Lisa Rae on AI ads and beyond, like, she was concerned that a lot of the pay per click people who were given, you know, junior positions to try to manage ad spend, that's kind of cutting out for them, you know.
So yeah, you're right that it does seem like it's not just, not just SEO, but kind of industry, digital marketing, industry wide.
[00:23:58] Speaker B: Yes.
[00:24:01] Speaker A: What is, what's new on the horizon for optimization for LLMs? Like what is in your stack? When you think like outside of the core of stuff that is SEO, is there anything in the technical vein or outside of that you're thinking of as extra steps that you have to take for LLM optimization, you know, because, you know, worrying about crawl, setting up redirects like that, that's all still in the original SEO tech stack and speed. Is there anything genuinely new that you have to worry about before LLM specifically?
[00:24:44] Speaker B: Sure. There is a much heavier emphasis on EEAT as well as C schema with eat.
I'm going to start from the top of the funnel.
I've seen clients that are a bit resistant to refreshing top of Funnel, to creating top of funnel, and it's serving a very different purpose today than it did in the past. Like, we've moved far past the era of the 10 blue links.
Now it's an awareness driver and you need to make sure that you're showing up in these awareness driving moments. And as we move to mid and bottom of Funnel, that's where EEAT really comes into play.
So it's not enough to say that we can solve these pain points. We need to show stories from real customers, real users that have solved these pain points and what success looked like for them and that's contributing to your EEAT and all of These different pieces across the the funnel do contribute to how you show up in LLMs.
You need to tell a consistent story and you need that first party data to back it up.
And then with schema it's also machine readability.
So how are we supplying extra context to crawlers?
[00:26:02] Speaker A: Talk a little bit more about that.
[00:26:06] Speaker B: In the past I saw schema as a bit more of a lever for SERP features to say show up and people also ask.
But now schema is a lever to show up in AI searches because you are supplying extra context about what is shown on the page. So if you consider like review schema review schema is showing social proof and that social media social proof likely appears here B2B SAS on G2 on Trustpilot and it's connecting the dots from your site to other places on the web where people have spoken well of you.
[00:26:48] Speaker A: That makes sense.
Did you see Ben Wells report from OPAlerts about his research in identifying an actual specific AI ranking factor?
[00:27:02] Speaker B: I did not see it. Tell me more.
[00:27:05] Speaker A: So he did a huge study and forgive me Ben for not being perfect on your quotes, but he did a huge study analysis with a the hypothesis was that backlinks still matter for LLMs.
And so he went at it from an anger of angle of vector embeddings. And so that's basically how LLMs work is they change the text that you put in into a string of let of numbers and then back out into letters.
And that's what it does all day, every day. But what he his premise was that you if you look at the. The I think there was four or five vector embedding types or common libraries of these vector embeddings that kind of categorize that content is within this specific niche or topic. You can think of it as like a map and everything that's related to you know, sewing is over here in 4, 5, 6 and everything related to automotive is 1, 2, 3. Right. It's very simplified but across those multiple models they still fall into those same number ranges universally. So what he proposed was he would pull the backlink profile for high visibility sites in AI results and look at the relevance of the page that was linking to that site to see if.
If it was provable that the backlink profile was even something that LLMs took into consideration. And it turns out that not only does it take it into consideration, but he was able to mathematically prove correlation positive correlation between a backlink profile and being in the top visibility for AI result slightly less so for branded linked mentions and slightly even, slightly less for unlinked mentions, but they are very close to each other.
So he was able to prove that your backlink profile and getting a nexus of number of backlinks from relevant sites that align with the topic of your page. And that was his correlation study was like if the links were coming from sites that were that had the same vector embedding for category, that was the signal. And so he proved interestingly that backlinks do matter for AI overviews, for, for AI rankings, ChatGPT and multiple different models. And he did across a bunch of different models. So I do hope to have him on the show and dig into it some more. He's releasing more of that data later this week, but his initial findings were out in the past couple days and I was just curious if anyone else had come across them yet.
[00:30:17] Speaker B: No, I hadn't seen it yet. It's so interesting.
Backlinks do matter. And it's something not specifically in the realm of backlinks that I've been talking about, but how you need to have alignment of how you show up across the web. And LLMs are looking for that confirmation that you're spoken about in the same way outside of your site that you're spoken about, that you speak about yourself on your site.
[00:30:45] Speaker A: And I think that that's kind of a shift of the thought process or the role that a lot of people have stuffed SEOs into. And I know that I was very happy in my little box in 2014 of like, you know what, you do the rest of the stuff at your business. I'm just going to work on these pages on your site. Okay. And then I'll get you traffic and you'll pay me money. You'll be happy. I'm happy you do whatever you want to do with your business. I'm just going to optimize these pages.
That doesn't fly anymore. There's a bigger picture. And we as consultants, as in House representatives, we ought to tie into the bigger picture of the business. You know, reputation management, pr, how the site functions, audience, you know, treating return customers with deference, I think is a big shift because for a long time SEOs have treated Google as kind of this endless ocean where you can always get more fish. And it doesn't matter the terrible conditions that you keep your current customers in, that won't fly anymore. You know, you have to tailor better experiences and provide more content for your, you know, anybody that you have converted, you want to keep them and you want to do more to put out content for you know, if you're a SaaS, more of your content needs to be helping your existence existing customers than just fishing for keywords to get new people in to convert.
[00:32:22] Speaker B: Yes, absolutely.
[00:32:26] Speaker A: What would be a key takeaway for people from today's talk if they could implement one thing?
What would you recommend that be?
[00:32:38] Speaker B: I recommend doing a a dive in to your page speed and sample across your site. Don't sample only a singular page because not every page is going to tell the same story. And if you are checking page speed across say a sample of four or five pages of your site, then you're going to be able to identify what the global fixes are and be able to make a bigger impact early on as you work down your list.
[00:33:11] Speaker A: Well, I appreciate you coming by. Give a shout out to your site, your agency, any resources that you've put out, how people can connect with you if they want to have more insights.
[00:33:22] Speaker B: Sure.
I'm at 10 speed and you can connect with us on LinkedIn.
We always are publishing new blogs. We just recently, recently put out a research report.
You can go and check that out.
You can also connect with me on LinkedIn as well.
[00:33:41] Speaker A: There we go. Well, thank you for stopping by, Erica.
[00:33:45] Speaker B: All right, thank you so much.
