How To Do Remarkable Customer Research With Rand Fishkin From SparkToro [AMP081]
Where do your customers hang out? What kinds of things do they like? What publications do they read? Customer research involves a lot of leg work, so does this information even matter? How can you leverage such insight for SEO?
Today, we’re talking to Rand Fishkin, founder of SparkToro and author of Lost and Founder. He is a powerhouse in the content marketing and SEO world.
Some of the highlights of the show include:
- Background, origination, and purpose of SparkToro
- Reaching/Researching Audiences: Slow, frustrating, and inaccurate process
- Companies spend money contracting agencies for a list of top customers, blogs, podcasts, and events
- Bones of Audience Intelligence: 1) Identify audiences across channels; 2) Know audience density; 3) Use trustworthy and valuable metrics
- How to obtain, benchmark, filter, and analyze data
- Data Points: Which to focus on and where to get them
- Social Network Profiles: Report follower count and engagement
- Biases generate unrepresentative data influenced by SEO
- Significant sample sizes and diverse groups are needed for true coverage
- Examples of missing specific audiences
- SparkToro lets you find people who practice specific fields
- Does current audience intelligence data represent the market as a whole?
- Improve SEO by helping audience accomplish tasks, and identifying and broadening link sources
Jordan: Customer research takes a lot of leg work. You need to answer questions like, “Where does our audience hang out? What kind of things do they like? What publications do they read?” And, by the way, do they actually have anything to do with their buying decisions or they’re simply affinities that really have nothing to do with our business. How do we marketers to this well and then leverage this insight for other things like SEO? To answer this question, I have the one and only Rand Fishkin on the show today.
Rand is the founder and former CEO of Moz and most recently, the founder of SparkToro, the author of Lost and Founder which drops next week, and of course, he’s a power house in the content marketing and SEO world. My name is Jordan with CoSchedule. Now, listen in on my conversation with Rand.
Hey Rand, great to have you on the show.
Rand: Jordan, thanks so much for having me.
Jordan: It’s my pleasure. I’ve been really excited to you. There are some big changes for you, can you tell us what you’re up to these days and what this next year looks like for you?
Rand: For the last 17 years, I’ve been at a company called Moz which I founded. Two weeks ago, well, as of this recording, we parted ways, I guess. I am doing a new company now. Starting a company called SparkToro which is focused on audience intelligence, sort of helping people find the channels, and sources, and people, and publications that influence their audience. I also have a book coming out called Lost and Founder, that’s sort of about the last 17 years of building this company. I think, ostensibly, about some of the biases that Silicon Valley startup culture can imbue on founders, and maybe some ways to avoid those. I’m starting a nonprofit called Project Event Safe and that is around making events safer for primarily women but really everyone. I’m staying busy, I guess.
Jordan: Yeah, it’s lot on your plate. I’m really interested to hear a little bit more about SparkToro and what you envisioned that looking like and sort of where did the aha moment or the idea come from. I supposed it was probably a slow burn too. You’ve been in this world for so long.
Rand: Yeah, yeah. Definitely slow burn. I think it was one of those less than aha moment and more a, “Gosh, this problem seems to keep creeping up and it doesn’t feel solvable or solved with software or with data yet. I wonder if there’s some who you do here.” Along with my co-founder—who’s still at his job so we’re sort of keeping that quiet for now—he did some experimentation and we looked at the data and we were like, “This looks interesting.” I’ll give you the full story.
Basic problem that we keep seeing is lots and lots of marketers have this process they go through. Jordan, let’s say you and I come out with a new lighting design product. Now we need to reach interior decorators and designers on the west coast with our new lighting design product. I don’t know about you but I have no exposure to that world, I have no idea where they hang out, I don’t know who they pay attention to online, I don’t know what they read, I don’t know if they listen to podcasts, I have no idea if they go to events. What are we gonna do?
What we’ll probably do is try and read a bunch of interior designers, maybe we’ll go to Google, and we’ll search for interior designers or interior design blog, or interior designer news, or interior design events. Then we’ll try from the top results sort of call together a few places and we’ll visit all those places, and we’ll see if they look appropriate. Maybe we’ll go to Facebook and Twitter, and LinkedIn. We’ll try and figure if those people are on there.
But this is a very manual, slow, frustrating, and inaccurate process. A lot of the time—I was doing this yesterday for classic cars—someone was trying to reach a classic cars audience. I go to Google and I looked for top classic cars blogs. The first result has a list of 17 blogs. I went to the top five or six of them and none of them had posted anything since 2015.
Jordan: Oh, that’s extremely helpful.
Rand: Yeah. This is really a frustrating type of world. I don’t know any tool that I can really use where I could type in classic cars or interior designers and get a list of the places and people that actually influence that audience. The blogs that they read, the events that they go to, the people that they follow on Twitter, the Facebook groups they are part of, the email newsletters that they get, the podcasts they’re listening to, that’s what we think we can solve using sort of a big data, profile graph approach. Then exposing that information to marketers so that they can sort of see like, “Yeah, look, a large number of the interior designers that we’ve been able to identify on the web all listen to this podcast.” That’s what we’re hoping to solve.
Jordan: That’s absolutely huge. We’re content people. I’m a content marketing strategist, this is my world, and it’s so interesting because as I hear you talk, I’m like, “Yeah, I have that problem too,” because everything we do is so data-informed and data-driven like we all say or that way. It’s like, okay, what’s the most important thing? Knowing your customer and how to add value to them and how to reach them in non-spammy ways. How do you do that? “Well, you kind of go search and then see if their profile on Facebook is open but I hope that they don’t know you’re trying to…” This is exactly it. It’s very gorilla in some ways.
Rand: I know a ton of companies that basically when they have this work to do, it’s usually a PR agency or content marketing agency or an influencer marketing agency, they contract these agencies and they’ll pay $20,000-$40,000 for a list. It’s just a spreadsheet, essentially. A spreadsheet that has, “Alright, we think these are the top 20 people on Twitter. We think these are the top 20 Instagram accounts. We think these are the top 10 blogs. We think these are the top five podcasts. We think these are the top three events. Give us our $40 grand.”
Jordan: Right. 50% upfront, naturally.
Rand: Yeah. A lot of cases, they’re doing $40,000 worth of work. They’re doing a big survey. Sometimes they’ll use SurveyMonkey audience or something like that. Sometime they’ll have a customer list and they’ll go email them one by one to try and get them to take that survey. Sometimes they’ll do as some sort of a giveaway or whatever. A lot of times they’re getting on the phone and interviewing these customers or these audience members to try and figure that out. It’s a ton of work, kind of like polling for an election. You’re dialling, you’re trying to get a significant sample set, you’re trying to reduce the deviation so that you have a high confidence in a role. Whew, man. Intense.
Jordan: In lieu of getting Nate Silver to work with you, we can hire SparkToro to do this for us. As you’re putting this together and as you’re building this, what are the bones of audience intelligence that you think like these three things arbitrarily, like these are the things you gotta know to be good at this. That’s what you’re focusing on servine up with SparkToro.
Rand: I think the three big ones for us in solving this problem are you need to be able to identify audiences across channels. It’s not good enough to just say, “Hey, here’s the people I follow on Twitter because many channels and many audiences are not exclusive to a single network or a single source. Diversity of source and having full coverage, that’s a critical piece. A second big one is knowing how dense your audience is in a particular location or around a particular source.
For example, maybe it is the case that lots of classic car enthusiasts, in fact, we saw this in our data set, lots of classic car enthusiasts visit the Fox News website. But the density is low. Even though technically, the coverage might be higher, but there’s a lot of your audience in that place, it’s not serving that particular need. That’s more of an affinity. That can be helpful, it can be helpful to see, “Okay, I know that they go to this website,” but what’s more helpful is finding the highly-dense place. The place that they go to that nobody else really goes to. That’s what you’re interested in. Interior decorators, “Oh, great. Lots of them go to BuzzFeed.” Everybody goes to BuzzFeed. That’s not helpful.
Rand: Right. Lots of them listen to the NPR of Moth podcast. Not helpful. Everyone listens to that.
Jordan: If you’re human, you like it, right.
Rand: Yeah, exactly. Exactly. What you’re looking for is that density. The third thing is you need metrics that you can trust. By that, I mean a lot of the reports that I’ve seen from agencies and in-house marketers include very surface level metrics. Mostly follower count. “Here’s this blog, you should get on this blog. Here it selects a rank and here’s how may they have on Twitter.” “Oh my god, that’s not helpful.” Neither those metrics are trustworthy or valuable. I think thanks to the Russian bots, I think we all know that we can’t trust Twitter follower counts at least. But Alexa rank hasn’t been good since the late 90s so we should really stop using that.
A couple of the other things that we’re gonna try and do is get metrics that match up well to when this, for example, when this Twitter account sends a tweet, what is the average impression count? How many people does it reach? Same with Instagram. How many actually see an Instagram post rather than just how many followers does this person have. Is buying Instagram followers is how every “influencer” close their audience size and shows off the importance. We need to be able to dig deeper and to say, “Does this actually get engagement? Does it actually get shared and liked? Does that happen consistently?” Okay, now this person clearly does have an audience that has high engagement with their profile, therefore, more important.
With a blog, I think that the data that’s best out there is quick stream data. SimilarWeb and Jumpshot collect clickstream data from millions of browsers and users and then they report back on that. It’s expensive to get access to but I think we’re gonna probably pay for the API and then expose those visitor counts to audience who can say, “Okay, this classic car website is highly-dense with classic car enthusiast. It only maybe gets maybe 25,000-50,000 visits a month but this might be a place you wanna hang out.
Jordan: You mentioned metrics you can trust. I think that’s so important because you can look at a dashboard and it can be really pretty and you can have pie charts and stuff and be like, “Oh, this is such good data.” But it might be a lie. It might not actually reflect reality. I was just talking to someone who used to work for a pretty big profile company that does metrics. He was like, “Listen, I hope this isn’t being recorded but you need to ditch them because their attribution actually doesn’t work. But they haven’t told anybody. I’m like, “I think you can get sued for stuff like that.” Anyway, that’s a whole other conversation.
Let’s talk about then how can you get data like that, that is trustworthy? How do you benchmark that and be really confident in that?
Rand: The theory behind this solution right now is that let’s hope that in the future SparkToro has tens of millions of profiles with data about what they’ve shared, what they’ve liked across a bunch of social networks, what they’ve linked to if they have a blog, what’s pointed to or talked about in their profile, in their bio, all that type of data. We basically search for interior designers and we find, “Okay, in our data set we’ve got 17,000 interior designers.” And then it’s just a venn diagram.
We overlay all of those 17,000 and we say, “What’s the intersect? Where do most of them share stuff from most often? What podcast have we seen them like, share, listen to, link to, seeing their profile comment on, etc., etc.?” Then, “Alright, we have 47 of these 17,000 people like this people which is unique to interior designers. We don’t see many other people liking and sharing this particular podcast.”
Jordan: That’s that density idea right? “Yeah, okay, density, got it. Got it.”
Rand: Great. High-density of this particular audience. And then we’ll go and look at clickstream data and say, “Okay, according to them this podcast gets between 1000 and 5000 visits every month. We’re gonna guess that that means probably 200-500 listeners per episode.” We report that back.
Jordan: Could we talk a little bit about what kind of things we could do now to get better at customer research and sort of maybe filter it for us? What data points should we be paying attention to and where do we go get them?
Rand: One of the tools that I actually really like—it’s public, you don’t have to pay for an API and code it out—is called Similar Webpro. I mentioned them earlier. They collect clickstream data. You can basically go to them, put in a website and see other websites that people visit from there, before or after that, as well as what other websites that are commonly visited by this website’s audience that don’t really go to other place. When we look up interior designers we see, “Oh look. They visit Architectural Digest. Oh, weird. They visit Sacramento Design Expo website. Ha, okay. That’s really interesting. I guess that must be a big deal.” Then they’ll estimate traffic based on the clickstream data.
For websites that are getting say 100,000-200,000 visits plus per month, they’re pretty good, pretty accurate. I was impressed. Actually, one of the other guys emailed me last week, I think it was, and said, “Congrats on SparkToro,” and sent me a little mini report from one similar web of how much traffic they estimated that we got. They estimated 47,000 visits in our first week and we actually got 60,000. Very close.
Rand: Really impressive. For a brand new website? Really impressive.
Jordan: It’s really nice to hear too. They weren’t like 120,000. It wasn’t an inflated number. Reality was better than they predicted but that’s really cool.
Rand: Yeah. They’re a great way to sort that out. Another thing that I urge folks to do is if you’re at a social network profile, don’t just report on follower count. Go look at the last 20 or 50 posts—whatever that is Facebook post, or Instagram post, or tweets, whatever—and then report on how much engagement did each of those get. There. Are there lots of likes? Lots of shares? That’s a good sign. That’s a sign that that account really has engagement. If those numbers are low and the follower count is high, probably they’re not reaching very many of their followers.
That’s a manual process, which sucks, I know, but it can really save your bacon in terms of where you spend that energy especially if you’re doing the classic influencer marketing thing, trying to get these people to post about your topic.
Jordan: As I think about this—from a content marketer’s perspective—our huge focus is top of funnel. We wanna write stuff and create content that’s hyper relevant to our ideal customer, that’s like 101. We paid attention for years now. We’ve learned a ton for you about SEO and Whiteboard Friday is the best thing ever. I was usually like, “Oh, what shirt and hairdo is he gonna have?” That like the level that I was into it. But really, when we look at that like, “Okay, we’re trying to attract this very specific person.” Because we do the same thing. We pick up the phone, we call our customers, and I think that’s really good.
I’m thinking about your persona—at a very basic level—if you’re putting together kind of personas. Where do you seen people either do really well with them or sort of go off the rails with them where you’re not attracting the right people. Can you talk about that a little bit?
Rand: Two of the biggest problems I see will sort of tie back to the things that we’ve already been talking about but one is that you get biased by your existing understanding of the field. You’ve got 10 interior designers who were sort of friends and previous co-workers with one of your co-founders. That’s sort of the audience that you talk to the most and they have a certain bias. Maybe it’s a geographic bias, maybe it’s just you’ve got a small sample set, maybe they’ve all worked at the same firm, the firm did these unusual things in the field, and so you get different data than what is truly representative of broad field.
I see this happen to with a lot of people who look at the top of Google. If you look at the top 10 results in Google for interior design websites or interior design blogs, you’re gonna be biased not necessarily by the ones that are most popular but by the ones that are really good at SEO. Those are not always the same.
I see a lot of that bias creeping into, not just the reports, but also the practices. Marketers will get biased by the 10 interviews that they do with interior designers and they’ll be like, “Okay, I got it. The Boston Design Expo is the biggest one in the country. That’s the most important one. That’s where we’re gonna exhibit.” Not realizing that, actually, that was true 17 years ago and the co-founder of this company but today, they’re the 7th largest, and you just missed that because this small group that you’ve been talking hasn’t moved on past that or hasn’t had that exposure.
Getting a significant sample size of an audience, making sure that it’s a diverse group, that you don’t have homogeneity on that group. That includes demographic diversity, and also background diversity, and experience diversity, and company diversity. You need all those things to be able to have true coverage of the field. Those are definitely two problems I would urge folks to avoid.
Jordan: Yeah. It’s almost this idea of rounding out your data or taking your persona and making it 3D instead of 2D on the paper. You’re sort of making these assumptions or these hypotheses based on the smaller subset and so the real task is then okay, this is what we think. This is what we’ve heard. Now how do I overlay that with other data points? To see if that’s corroborated or if it’s like ahh, this does not seem to be true.”
Rand: Yeah, exactly. Exactly. A good example of this, in the SEO world, for a long time, I thought I knew all the major players and where folks were getting their information. It turns out I was out actually completely ignoring a huge, huge part of the SEO market probably for my first 12-15 years in the field which is web designers. Web designers do not spend any time in the SEO ecosystem. They spend it in places like A List Apart and Smashing Magazine and…
Jordan: Awards and all that stuff.
Rand: Yeah. Design Plus and these other places. Those places have articles by people about SEO. When I went and checked them out I was like, “Gosh, I’ve never heard of any of these SEOs.” I don’t know anything about them but they have been dominating the design market which is huge. There are probably, for every SEO, I think there’s probably 5 web designers if not 50. That is a huge miss because all those people are trying to do the same things that SEO are doing. They’re trying to help their clients who are getting websites from the get ranked higher. I had my biases. I ignored. I talked to SEOs. I thought I knew the field. I didn’t realize there’s also other people doing the same practice. They just don’t call themselves what we call ourselves and therefore, I missed them.
Jordan: How did you figure that out?
Rand: I think someone outranked Moz on Smashing Magazine with an SEO piece. I thought, “Geeze, Smashing Magazine.” That’s a lot of authority, sure. It makes sense. I found the site and then I saw they had a whole SEO section. They’ve been talking about for years. If I remember correctly, I think I actually ended up emailing with one of the editors there and contributing a piece myself. It drove crazy traffic, very, very good traffic. There you go.
Jordan: It’s sort of just keeping your eyes open and paying attention to who is playing in your sandbox?
Rand: If I had a tool, if I had a tool like this SparkToro thing, I could’ve gone to that tool and I could’ve looked for people who do SEO—not just who say they’re in SEO—but people who practice it, who share content about it, who write about it. I would’ve seen, “Hang on, there’s this whole other freakin’ world of designers sharing this content, and linking to it, and talking about it, and liking it, and amplifying it. I need to go pay attention to that.” Who knows, once I build this thing, it could be that I find two or three other worlds that I don’t know about. It wouldn’t surprise me.
Jordan: With our new interior design business that we’re starting, if I come to you and I’m like, “Hey, Rand. I’m worried now after listening to this that I might be missing this whole segment like you’re talking about.” What would be the thing you’d counsel me or that person to do like this afternoon to survey and see like, are you missing Martha Stewart’s book club or something like that?
Rand: Right. Sure. I think one of the big things that you can do is look at how you got your current audience intelligence data. How do we know what we know about our target audience? If the answer to that is, “Well, we did a large scale survey through a SurveyMonkey audience or through a collection system that has a lot of statistical rigor behind it.” Yeah, maybe we can probably feel pretty good about that data. If the answer is, “Well, our co-founder had 10 friends.” Oh, let’s broaden our approach. We need to get out of our bubble.
One thing that I might consciously ask is is our current customer set accurately representative of the field as a whole or of the market as a whole? If we’re serving interior designers, do we have a strong bias to only the Seattle Northwest area? Do we have a strong bias to only folks who’ve been in the field 10 years or more? Maybe we’re missing a batch of the newer entrants. Do we have a bias that is only people who work on their own, nobody who’s in larger interior design firms? Another bias. Maybe we’re missing out on the larger market. All those kinds of things can help you discover and then fix some of those bias issues.
Jordan: I like to bring it in for a landing every week and ask you to put your Gandalf cap on and sort of go to this sage wisdom time. What would your best advice be for someone to take this understanding and knowledge and sort of leverage it—thinking back to content—leverage it for SEO to grain traffic and hopefully convert people. Is there some secret sauce in here that they could use to get a leg up in that world?
Rand: SEO is strongly based on a few things but two of the most important are are you doing a great job of serving the searchers task, helping them accomplish their tasks, getting them what they need. If you know your audience better, you can serve them better with the content that you create. If you know that, “Oh, wait. It’s not just people who have these problem who are coming to my website, it’s people with these three problems. I need to be able to solve and direct all of them to those that can really help your SEO.
The second one is SEO is still strongly based on links. If you can identify places that could be link sources for you, they could be talking about your subject, like my example of finding Smashing Magazine, and then realizing that they publish SEO content and writing something for them and getting all those visitors from there, perfect example. You wanna be able to broaden your link sources so that you can show to Google that you are important and notable across your field.
Jordan: Thanks so much, Rand, for being on the show. Real quick though, can you just give us a quick peek at Lost and Founder? I know it’s dropping next week. I’m pretty pumped about it. I got it pre-ordered. Can you just give us a little sneak peek, snapshots of what’s in there?
Rand: A lot of it is dismantling the classic Silicon Valley startup wisdom. I tackle things like, “Is an MVP really a good idea? When is it a terrible idea?” A minimum viable product. I look at fundraising and whether it’s this glorious and deserving of glory as it often gets and whether there’s some alternative paths. I take a look at some of the hard things that businesses face. I talk about my depression. I talk about Moz’s layoffs that they did a couple of years ago. I talk through challenging personal stuff like early days of the company when my mom and I were on the verge of bankruptcy and half a million dollars in debt.
I talk about the transition from being a services-based consulting business to a product-based business, a software business. What led us have success there. Hopefully, making some of those myths a little more mythological rather than reality. Also, telling a lot of personal stories from experience, and then drawing on other company’s experiences as well, and a bunch of research and data to try and back up, “Hey, is this just Rand’s opinion or is there a real point to be made here?”
Jordan: Right. Is it just a sample size of Rand or is there something more about it.
Rand: A lot of it is my bias. My hope is just go out there, build a company, make mistakes, just don’t make exactly the same ones I did. Hopefully.
Jordan: Fair. Love it. Hey, thanks so much for being on the show, Rand. It’s just a pleasure to talk to you.
Rand: I’m thrilled to hear it. I look forward to the next time.
Jordan: Alright, thanks, Rand. Remember, the bedrock of quality audience intelligence is one, identifying your audiences across channels and getting full coverage into where they’re at, what they read, and where they like to talk. Number two, taking that one step further in determining how dense your audience actually is around a given source. Number three, make sure you have metrics that you can trust and that matter. Follower count is great but averaging engagement is better.
Thanks so much, Rand, for being in the show and sharing the marketing wisdom with us. Rand is, of course, the former CEO and founder of Moz, the founder of SparkToro, and the author of the upcoming Lost and Founder which drops next week.