How Britney Muller deindexed 75% of Moz.com’s site and saw success


Posted on January 25, 2018 in Business, CredoCast, Growth, SEO

Britney Muller is the SEO and Content Architect at Moz. She’s recently moved from Denver, my town, to Seattle to work in Moz’s main office. At MozCon 2017, she gave an incredible case study about what she did on Moz to see movement forward of Moz’s organic traffic.

In this video you’ll learn:

  1. How she identified tens of thousands of low quality profiles, and why she noindexed them;
  2. The results she saw from this;
  3. What she’s learned from trying to lose featured snippets in Google’s search results;
  4. Why she recommends noindexing pages instead of using robots.txt.
  5. How title tag changes killed conversion rates, and what she learned about broad vs specific.

Britney on Twitter

Moz.com

Transcript

John: Hey there, everyone. Welcome back. Today, I have with me Britney Muller who is with Moz, who actually just relocated to Seattle from Denver. So, I’ve known Britney for a number of years now, and actually, a funny story for you is the first time I ever met Britney, it was at MozCon in 2013, I wanna say and I actually thought that she worked for Moz. I was like, “You work for Moz, right?” She’s like, “No, I don’t.” It was, like, in the lunch line or something like that. And then, funny enough, you know, three, four years later, Britney joined Moz. And so, Britney is now… What’s your official title over there?

Britney: It is SEO and Content Architect.

John: SEO and Content Architect. So, Britney is basically responsible for improving the organic visibility of moz.com, which if you’ve been in the SEO world at all, you’ve read their blog posts and know that they’re an SEO and inbound marketing software company. And if you’ve been watching these videos, I had Rand on the show recently. So, Britney, if you would tell us a little bit about yourself, what you’re up to these days and also how you got into this crazy SEO and digital marketing industry.

Britney: Yeah. So, that’s kind of a weird story. So, I’m originally from Minnesota, went to University of Minnesota and graduated with a major in Strategic Communications Public Relations.

John: Really?

Britney: Yeah. And I just couldn’t find…

John: Did not know that.

Britney: Yeah. So, kind of random. But after college, I could not find a job that I wanted. So, I ended up just packing up my car and moving out to Breckenridge, Colorado to snowboard as much as possible. That was just kind of… My goal was to fulfil my, sort of, dream of being a snowboard bum for a season or two. So, while I was doing that, I got stir-crazy pretty quickly. There wasn’t a whole lot to do, so I started writing for a local realtor who introduced me to just basic programming and SEO. And the second I found out that I can figure out how many searches there are for ski-in/ski-out Peak 9, Breckenridge, single family home, my life was forever changed.

So it just became a big chess game for me. So, I started manipulating very strange things. I got into some trouble with some large national brands because I was raking above them. But it was just kind of like a big game for me and I wanted to see if I could do it. And then I joined an agency in Vail for a year. And then after that broke off and started PRIDE Marketing where we did strategic data-driven medical marketing. And then from that, yeah, from that, I kind of reached the traditional entrepreneurial burnout, as you would say, and just decided I kinda needed a change of course after five, six years of doing that. So, jumped onboard with Moz, and it’s been awesome.

John: That’s amazing. So you joined Moz beginning of this year-ish, is that right?

Britney: I joined Moz mid-July 2016.

John: That long ago? Okay.

Britney: Yeah.

John: Gotcha. So, well, hey, I didn’t know that PRIDE just focused on medical marketing. That’s really, really interesting. You know, the power of niching, you know, niching down and, you know, finding exactly, like, what you’re good at and what you enjoy as well. But, yeah. I mean, what I wanted to talk about today is… So obviously, you’re at Moz, and Moz is a huge website and it’s also a leader in the SEO space. If you work in SEO or you do any SEO like, you know, as a business owner, anything like that, like, you’ve heard of Moz, basically. And if you haven’t heard of Moz then you know Rand, who had the ridiculous mustache for a while.

So you came on as the SEO and Content Architect over a year ago. And I know that you gave a fantastic talk at MozCon this year, MozCon 2017. So, MozCon, obviously, is Moz’s annual customer conference. And I saw… Like, I really wanted to be there, you know, for your talk, to see it. I saw amazing tweets and all that. Everyone was just like, “Britney did an insane job.” And then I started, kind of, digging into it, which… Congratulations. Like, the biggest stage in the digital marketing world. That’s super cool.

Britney. Thank you.

John: And well-deserved as well, of course. But I heard that you basically talked about how you…the work that you’ve been doing over the last year or so to really improve Moz’s own organic reach. And so that’s what I wanted to dig into. I don’t know too much beyond it from that, but I think it’s really interesting to talk about a case study like yours, being very transparent and literally telling us, like, what are the things that you went and did, you know, where did you start, if you have a ginormous website like this that there’s a ton of opportunity but you also have, you know, a ton of, like, low-quality stuff. Can you tell us that story?

Britney: Yeah. So it’s super interesting. I mean, I did what any of us would do when you start at a new big company. And I did the full technical audit. I mean, I ran the gamut on everything. And something that kept, sort of, popping up for me, and I noticed over and over again, was whenever I would try to dig into the site’s index pages in search, so whenever I would do a site:moz.com, I would take a really, really close look at the first 100 results, so those first 10 pages. And what I noticed was over 56% of those were spammy community profile pages.

John: Wow.

Britney: Yeah. Like, 56%, it was insane. And they were doing well and these people had, you know, whether they were bot-created or not, they had pointed thousands of backlinks to them and there was just so much spam around that whole area of the site.

John: And we’re literally talking tens of thousands of pages here, right?

Britney: Yes. Yeah, that have been around forever, you know. So that was a huge red flag for me, just in the scheme of things. And I knew that eliminating those could make a huge impact overall. So that’s, sort of, where I started was which of these changes are relatively easy that can make the biggest impact. And so sitting down with some of the developers, we were discussing options to meta noindex profile pages under 200 points, and then start to clean out some of the stuff. And the results that we saw were absolutely phenomenal. I mean, almost instantly, organic traffic went up, rankings went up. I like to think it, sort of, reconsolidated our own site authority, you know, and really cleaned up our index. So, I love to kind of share that story and some insights.

But something that I have heard a lot back from, after the conference, is people trying to do this and there are many really skilled SEOs that might have a hard time implementing something like a meta noindex, just because they haven’t done it before. So the most common mistake I see people making is they meta noindex a bunch of pages, but they still have those pages on a “disallow” in their robots.txt file. So if you’re disallowing the bots from crawling those pages in the first place, they’re not gonna respect or pick up your meta noindex. So, I think that’s an important weird little nuance to make. And when we did it, I even left the… We had a profile sitemap, community profile sitemap, that I left in place for a while, just to make sure that those were all being seen by the crawlers.

John: How long did you leave that in place?

Britney: I would say, maybe, like, two weeks, a week to two weeks. Yeah.

John: Okay. And so you watched that, like, you watched that sitemap or that, like, set of sitemaps to see. Did you split out these, like, low-quality ones into, like, into its own separate sitemap, and submit that separately? How did you go about that?

Britney: No, because I wanted them all to be, sort of, recrawled anyway. So whether or not, you know, they had five points or 500 points, I wanted it to be evaluated and then cleaned out. And then after that, I did remove the profile sitemap and now… So if someone leaves a comment link, that will be followed to their profile page and it’s a more natural way for the Google bot to pick those up. But initially, with all of those noindexes, you know, I just needed to make sure that those were being picked up.

John: Gotcha. So you didn’t actually go and delete those pages, those, like, low-qualities or profiles off the site?

Britney: No.

John: Okay.

Britney: And it was relatively simple. It was just through htaccess, so it was just a rewrite. Yeah.

John: Yeah, totally. Yeah, super. Yeah, not that part. And then the… What’s the logic behind the, you know, if someone leaves… Well, first question is, what’s the logic behind if someone, you know, that has one of these profiles that’s underneath 200 points, leaving that…having a followed link, like, from the, you know, from the comment that they make on a blog post back to their profile that’s noindexed. At some point, basically, do you plan to, like, if they… You know, I mean, A, you’re trying to encourage, like more engagement, right?

Britney: Right.

John: But also… So, do you remove the noindex, you know, automatically when someone hits 200 points?

Britney: Yes.

John: Okay. And why did you settle on 200 points?

Britney: We had a couple meetings on it and that was, sort of, the comfortable level that everyone felt, because I think it was already in the guidelines, I would have to check to make sure, but I believe it was already in the guidelines that if you reached 200, your website link becomes a “dofollow” on your profile.

John: Right.

Britney: So that was still, sort of, in play. And then we thought in order to respect that, we would remove it at 200.

John: Gotcha. Gotcha.

Britney: Yeah.

John: That makes sense because then, also those profiles have, you know, more comments on them and they’re much more likely to build out their information. So it’s really like a robust, like, profile for someone in the SEO industry.

Britney: Right. Yeah, yeah. Exactly.

John: Interesting. Interesting. What sort of lift did you see?

Britney: I should bring up my slides so I’m not just like making up numbers. Yeah, pretty awesome. Let’s see. Honestly, the whole thing really surprised all of us.

John: Yeah.

Britney: Just because I think those pages had been around for so long. I think it sent a really, really big signal when we did that, you know. It was, yeah, tens of thousands of pages.

John: Yeah. I mean, Moz is not a small website.

Britney: No.

John: I mean, you have… I’m just doing your site:moz.com search and I see… Okay. I see 45,000. Well, that’s really interesting because I see 45,300 results. I don’t know if those are being filtered at all. So basically, like, you noindex, like, a third of your site.

Britney: Oh, yeah. When I started, there was over 100,000 indexed pages.

John: Okay.

Britney: Here. Okay. So, I stand corrected. When I started, there was 175,000 indexed pages.

John: Wow.

Britney: Yeah.

John: Wow. So you’ve literally cut that down by… You’ve removed 75% of your pages from the index.

Britney: Yeah.

John: Wow.

Britney: So, let’s see. To get to, like, specific results here. It was crazy how spammy some of them were.

John: Yeah.

Britney: So, at the time when I did site:moz.com community users, there was 71,500 community user profiles and I brought that down to 1,490.

John: Wow.

Britney: And the result was…we saw a lift in organic users, almost 9% the following month.

John: Wow. Across the site?

Britney: Yeah. And then we saw a lift of 13.7% year-over-year for organic traffic the following month.

John: Wow.

Britney: Yeah.

John: That’s crazy. And that’s just continued to build?

Britney: Yeah. I would say, as of lately, we’ve, sort of, plateaued a little bit. And that’s something I’m, kind of, working on with a bunch of new interesting projects, just because I’ve been doing so much speaking, so it’s really hard to juggle all that stuff. But what I’m working on now is trying to implement a better data-driven sitemap. You know, I think all too often, we just throw up, kind of, almost shitty sitemaps, you know, based on…

John: Their pages.

Britney: Yeah. Why don’t we do it based on popular pages or pages that get tons of traffic? So that’s, kind of, what I’m trying to work on next to get some of that stuff re-evaluated.

John: Interesting, interesting. Very cool. Yeah. You know, I think this is an interesting point to kind of underscore. I mean, I work with very large websites, million-plus-page websites and, you know, in the past, I mean, think pre-Panda days, right? So, pre, like the low-quality content days. Basically, you would wanna spin-up as many pages as you could to get traffic, right? And not necessarily, like, you know, spammy user profiles, right? But just like all these different, crazy combinations, even, like, misspellings back in the day. Like, people would purposefully, like, target pages with misspellings, right? Crazy.

Britney: Yeah. Oh, my God. That’s right.

John: And of course, Panda came in, what? February? I think it was February 24th, 2011. And it was just like, “Boom.” It was like apocalypse, right? And then, like, a year later, in April, it was Penguin, you know, spammy links. But it’s really interesting, to me, how over the last five, six years, I’ve seen, time and time again, I did it on hotpads, I’ve done it on other clients where when you eliminate a ton of these, like, user-generated, super low-quality, you know, a lot of them duplicate content, especially on, like, marketplaces and eCommerce websites. You know, you’re really putting like… You see a good uplift, I mean, yeah, 9%, 13%, you know, that sort of thing. Like, that’s not uncommon to see. You’re putting more wood behind fewer arrows.

Britney: Yeah, exactly. Exactly.

John: So they’re not wasting their time. Did you do any, like, did you go deeper into, like, you know, analyzing your log files and seeing, like, what they were spending time on and that sort of thing?

Britney: Oh, John, I need you here, dude. I have been after our developers for so long to get me those log files but the way our server’s set up, we’re not capturing them because of the load balancer.

John: Oh, okay. Yep. I’ve seen that.

Britney: So they need to create, like, a separate cache for me, but that’s something I have not gotten yet. I’m dying to see it though, just because Search still tells us nothing, like, literally, nothing.

John: Totally. Totally.

Britney: So, I’m so frustrated. I would love to get my hands on, like, how it’s all resolving, where is it going, what pages does it hit over and over, all of that good stuff. Yeah.

John: Yeah, yeah. Gotcha. Gotcha. Stay tuned, that’s right. We’ll do another version once you’ve gotten those.

Britney: That would be cool.

John: Yeah, totally. So what else have you done, I mean, over the last…you’ve been there a year, 15 months-ish? Like, what are… Obviously, you’re doing a lot of speaking. I mean, every time I look at your, like, Instagram Story, you’re, like, somewhere else, just speaking at a conference.

Britney: I know. This year has been crazy.

John: Yeah, yeah. But what are some of the other, like, things, I guess, campaigns or initiatives that you’ve undertaken there at Moz, that have helped you, you know, see results on a site like this?

Britney: Yeah. So, it’s funny. I have a lot of fun just being a creep on our GitHub repositories and seeing what things have been done in the past. So, like, I found one poll request from, like, years ago that talked about Google Guice. We’re Moz. It’s just hilarious.

John: Yeah.

Britney: But essentially, what the poll request was about was in the proper use of rel=canonicals for our local pages that were still in place as of a couple months ago. So finding things like that, and fixing them, and seeing the immediate lift, I have a lot of fun doing. So I’m really trying to work on that stuff, as well as just experimenting with schema, experimenting with our featured snippets. That’s kind of an ongoing campaign right now. I’ve done a lot of title tag meta description testing that has been actually, kind of, hilarious because…

John: Really?

Britney: Yes.

John: Tell me about that?

Britney: Okay. So, I initially thought going into this test where I took, I wanna say, maybe 60 or 70 pages. No, probably more around like 100. But I bucketed them into control groups, right? And then ones where I was just changing the title, ones where I was just changing the meta, and ones where I changed both, and then evaluating all the metrics after.

John: Yeah.

Britney: And in doing this…

John: So you benchmarked all of them? You benchmark, like, traffic, click-through, click-through rate, all that stuff?

Britney: Yes.

John: Okay.

Britney: Exactly, exactly. And this is funny because this is something I haven’t really talked to anyone about, but…

John: Yeah. All right. But, yeah, so we were talking about these title tag tests that you’re doing where you bucketed them into separate groups, you know, control and then experimental, benchmarked them, let’s go from there.

Britney: Okay. So, I thought we could cast a far wider net by maybe simplifying some of the titles and by, kind of, going from, instead of, so this is an example, instead of “SEO Best Practices for URLs,” I changed it to…so stupid, “What is a URL?” And then in parenthesis, I put “Uniform Resource Locator?” And it went from a 5.6% click-through rate to 2.99%.

John: Okay.

Britney: So that’s like a great…

John: Runway.

Britney: Yeah, runway. So, the more broad and, kind of, over-simplified I created these things, the worse we ended up doing. Another example, let’s see is, like, I changed our domains, “SEO best practices” to “What is a domain name?” That went from 5.25 to 1.95 click-through rate. But the things that did work consistently across the board was when I really just simplified the titles that were already in place, and then provided the answer in the meta description. So when I was testing both, I would do onsite research as to what keywords are people currently coming into this page for. So an example is, you know, in Moz Pro, how you have your search visibility score? Everyone’s always, like, “What is search visibility?” So, for that page, instead of it being “Guide to Search Visibility in Moz Pro Rankings,” I changed it to “What is search visibility?” And then I provide the answer. And it went from a 2.5% click-through rate to over 7% click-through rate.

John: Wow.

Britney: So that was kinda like the big awakening] for, you know, in at least this case, where we’re providing these educational resources, is you don’t wanna go to oversimplified and you do just wanna, kind of, answer those questions.

John: Yeah. Totally, totally. Yeah. Well, that one specifically was a… It’s like a Moz-specific question, right?

Britney: Right, right.

John: Why do you think the others, you know, that you tested going, you know, going broader from, like, SEO best practices, why do you think that one… I mean, did you lose rankings?

Britney: Yeah.

John: Okay.

Britney: Yeah, we did.

John: So was it that you basically, like, deoptimized it away from, like, you know, “SEO Best Practices?” And so, like, you weren’t getting it? You weren’t targeting those keywords anymore?

Britney: Exactly. It was way too broad and there were already so many broad resources out there that were, sort of, better suited and less technical than our, you know, “SEO Best Practices for URLs.” So I think Google picked up on that and was like, “No, I don’t think so.” Yeah.

John: Interesting. Have you done anything around, like, when you do have, you know, a broad, you know, topic like that, you know, you’re identifying that there are multiple… It’s ranking, kind of, for other head terms. And then building out, like, resource content around that. Like, I know you guys have done a turnaround, like, your resource content on the site. I’ve seen those resources and all that. Like, can you give us any, like, insight into that?

Britney: Yeah. So, that’s kind of been shuffled around so much lately with the new learning center that I haven’t been able to nail any of that down because we’re just still in the process of, like, transferring a lot of things from their old addresses to their new. So that’s something we’re just, kind of, currently making sure is well-adjusted before we tweak anything there.

John: Gotcha. So you’re basically migrating that content right now, and then you’ll start optimizing it?

Britney: Yeah, exactly.

John: Gotcha, gotcha. Yeah, and that’s, I mean, I think that’s an interesting topic. I know Moz has talked about… I know Rand has talked about what, around, like, you know, moving content from, like, you know, a subdomain to a subfolder, then shortening the URL like that, that sort of thing. I mean, is that like an ongoing thing that you all are doing because you know that it works? I mean, other people in the industry will naysay it and be like, “Yeah, whatever. It could be this, that, or the other thing.” I mean, I’ve seen it a lot of times. Is that something you guys are investing in actively and doing?

Britney: So, that’s not primarily what we’re focused on. It’s more of a CMS thing for us right now, and just making it more accessible for people internally to make those changes. So we don’t currently have much educational resources, if any, that I can recall, currently living on a subdomain. But we do want them to be in more, kind of, mindful URL structures and places on the site.

John: Yeah.

Britney: Something that we have, kind of, experimented a bit with, that I stumbled upon months after I started, but we ranked for so many featured snippets, like, over 550 some featured snippets on moz.com. And I was trying, for the longest time, John, to, like, get new stuff into featured snippets, and it started driving me crazy because they’re so hard to get and a lot of them are really competitive and super tricky. And so I had this, like, lightbulb moment of, “Wait a second, what if I stopped trying to get featured snippets and I start trying to lose them?” I sabotaged quite a few of our featured snippets, just to figure out what that would be, right? What is the thing that I can tweak just enough to see us lose it? And it would surprise you, I started adding typos in meta descriptions.

John: Okay.

Britney: And we would lose them. And these are meta descriptions that weren’t even showing up in the featured snippet at all.

John: Right, right.

Britney: But having that misspelling caused us to lose quite a few. And then reformatting pages was, kind of, the other big one I noticed. Yeah, super, super interesting stuff. I still wanna play around more with that but I realized I’ve gotta be careful too.

John: Totally, totally. Yeah, yeah. I think it’s cool that you’re able to, you know, test that sort of thing, and basically, like, you have the, you know, the levity to reverse engineer it and say like, “Okay, we have these. Now, if I remove, you know, if I do this one task. Okay, now, we lost it. Let me add that back in. Okay, now we got it.”

Britney: Yeah.

John: All right. That’s a featured snippet, like, ranking factor basically, right?

Britney: Exactly.

John: So, yeah. Those are super important, super interesting, like, lessons to learn there.

Britney: Yeah, that was fun.

John: Very cool. Yeah. Well, Britney, thank you for taking this time to chat with me. I know, I’ve really enjoyed it. I love hearing how, you know, sites like Moz are doing SEOs themselves and the stuff that you’re learning. I think there’s a lot of takeaways. I mean, from this one, definitely, like, I mean, high signal-to-noise ratio. Like, you’ve reduced the noise on your site by… I mean, I’m amazed that you eliminated, like, 75% of your indexed pages, like, that’s it. That’s a ballsy move. And most companies, like, won’t do it. But you really showed it like, in certain situations, right? Like, if you have a lot of low-quality stuff and all that, like, it can actually show a really good, you know, improvement for you. That’s something that was holding you back. And honestly, that you can even do that without, you know, going to your log files and saying like, you know, “Okay, like, where are they spending a lot of time?” I mean, if you had seen those, you probably would have been like, “Oh, wow. They’re spending 70% of their time on this stuff that’s not gonna rank for anything. Like, how can we, you know, how can we move that?”

Britney: Yeah.

John: Yeah. I like that you’re taking, like, a very data-driven approach to it. And, you know, hopefully, this will also help out other people that, you know, have a large website or, you know, have a ton of, like, low-quality pages on their site to actually start running some of these tests if they’re able to.

Britney: Yeah, definitely.

John: Very cool. Well, Britney, as I said, thank you for spending time with me today. Where can people find you online if they wanna learn more about you, about Moz, about the work you’re doing at Moz?

Britney: Yeah. I think the easiest way is probably Twitter. So it’s just Britney Muller, is my handle, exactly how you would spell my name, B-R-I-T-N-E-Y M-U-L-L-E-R. That’s probably the best way, yeah.

John: Awesome. Good stuff. Well, Britney, thank you for…thanks again. And enjoy your time up there in Seattle, and hopefully, I’ll see you back down here in Denver soon.

Britney: Will do. Thanks so much, John.