Market Muse & Leveraging Topical Clustering To Maximize Your Forecasts
I reached out to a number of different SEO experts to get their #seobits take on a passion they have in the vast world of SEO for our #seobits video sessions. Luckily, Jeff from the killer software Marketmuse responded with this recording of his conversation on topic clusters and how they can improve your keyword research process. Enjoy the interview video, browse the transcript and overview/summary below for the recources and processes he walks through with Steve Jeske.
Transcript Of His #SEObits Session
I’m Jeff Coyle, the co-founder and Chief Strategy Officer for Market Muse. And I’m so excited to be joined today by Steven Jeske, our content strategist at Market Muse, We live our lives inside buildings and estimate and predict traffic. What we’re going to do is estimate and predict traffic potential more effectively. And you know, the reason for that is that we want to be able to go to, you know, the budget holder. I’d like to be able to go to Steven, who decides what market news writes and say, Hey, here’s a great idea for a piece of content.
But I want to be able to deliver the why, and I want to be able to estimate the upside potential of this page, maybe a new page, or creating this new page and updating a collection of pages that may be connected to it so that I’m not just talking about this, you know, myopically or from a lens of, you know, with one data point, I’m truly getting close to predicting outcomes, and that’s the nature of what we’re talking about today.
So, Steven, the first thing we’re going to talk about is, obviously, what is the north star that we need to knock out of the sky? What do you think? Topic volume, Jeff? 50. So, search volume entails calculating and predicting how much traffic a page will receive through search volume.
Isn’t this fundamentally problematic? It’s not just because every search result is different, but that’s critical, right? So every search result, the RES, and the organic results are at different points in terms of pixel depth and how many pixels you’ve got to go. Some search results are on page one. Or continuous scroll, or whatever you want to call it, will have different click-through rates?
But then, only looking at the search volume for this one word doesn’t tell the story of the page. So it’s separating, you know, the search volume of a word from what this page is going to get. That’s critically important to get. And once you put those two points in place, you can see that there’s just so much error.
The concept of term pool
One way to think about this mistake is as follows: you’re going for one word, that focus topic, and that target keyword, right? And it’s going to also naturally be about a lot more things. It’s going to rank reasonably well for some things, maybe great for things you didn’t predict, and maybe not so great for others. But those are all going to combine into something crackable.
Term pool, right? So I’ve got my focus term, and then I’ve got this trackable term pool, all the other stuff that it’s going to rank for. Guess what? That’s going to be more than you’d expect based on search volume for a single term. So, you know, the old Hy-Phothes hypothetical here is, “I’m going to focus on this keyword and get its 1,000 search volume and it’s ranking fifth.”
So, Steven, how many clicks do you think that will get? So you might look at the clickthrough rate on position five here as being about 5%. Okay? That’s not telling the whole story. Even if you’re only accounting for the pool, you’re doing your research to know what that term means, right?
There are other words that we’re going to get access to. So rule number one, search volume by itself, forgets about all the other words you’re going to rank for, right? So you need to do better. But there’s another piece to this, right? There are piles of words that you won’t be able to predict. So there’s a concept called, and I’m gonna drop to a share screen.
There’s a concept called the “pool multiplier,” and these are the words that you can’t predict that this page is going to get connected to because they may only be searched for once in Google; they may be low-volume terms that your favorite keyword research tool, whether you use Sim Rush at RS Market Muse or something else, isn’t going to have data for because maybe Google AdWords doesn’t report that data.
So there is no place for supplements, whether they are clickstream supplements or calculation-based supplements, which are available across all platforms. Are you either taking Google data and supplementing with clickstream data or supplementing with a proprietary math problem, usually done by pixel analysis of the search result?
So no matter how they’re doing it, those are the calculations that you need to fix, but there’s always going to be this pool.
The importance of topical authority
The cool thing about this pool is how authoritative you are. On this topic and semantically related topics, the more topical authority and authorship you have, the more access you have to that second orbit.
So I always like to think about this target word. The first orbit is my known term pool; the second orbit and beyond, I’m not sure, right? However, the larger and more powerful your cluster, the more traffic it will have access to. And this is very much me. So you could choose a target word that has a thousand search volumes.
Well, that second orbit of a collective of words could be a thousand to 2000 visits. Right. which is, you know, 20, 30, 40 max, which you would predict by only doing click-through rate analysis. So far, so good. Absolutely. Yeah. And so, by taking that information, it’s going to allow you to take better next steps too.
So you see a page ranking well for some things and not for others. So what do you do? Would you improve this page for that term? Do you begin writing on adjacent pages? It’s going to help you with content strategy to be thinking about this in the first place. In second place, it’s going to allow you to estimate better.
And what I will coach you on today is an estimation. Isn’t this page going to get very little traffic? So, why will I get at least this much? Why? I want to predict my outcomes effectively. I want to be efficient with the content that I create. So if I look at the averages, Steven, we’ve done the studies on this.
What’s the deal with the number being 10% for content efficiency? Is that right? Yeah, that’s about it. Yeah. So, that’s where you are. Your team is creating content, and for every 10 articles you make, one of them achieves the goals. Predictively. Yeah, right. When we work with teams, we’re getting that number up to 40 and 50%, so we’re going to talk about that and how we might be able to influence it, but we also want to predict, because if you’re predicting a minimum of traffic, you’ll likely get it, but then you’re only right 10% of the time.
Right, that’s tons of And, you know, all of these processes and all of these quick tricks that people use go out the window.
The benefits of being predictive in content creation
So we want to take that away. We want to be more predictive of what’s going to be successful. And then we want our estimates to say, “I’m going to at least get this much traffic when there’s an upside, and here’s why.”
And the benefit will come from our existing authority on the topics. Alright, so let’s show this process and why it failed. Using only those data points in Seru And then let’s look at the common hack and the trick that has two of my favorite workflows. So we’re going to look at this as a market muse.
Okay, the first thing I’m going to show you is Topic cluster 260 Okay, cool. So again, if you were just doing this, I’m going to rank for the topic cluster. How much traffic would I get? So I’ll take 5%, 10%, and 26. Okay, cool. So I’ll get a 10% click-through rate. And I’m going to guess 26. Nope. I need the term “pool.” So what might I do?
I might add up all these search volumes and then try to calculate them again. I’ll still be missing out on a lot. I’m still missing quite a bit. All right, so we’re going to put a pin in that for search volume. We all know you can’t do that. And you’re competing if you don’t understand the x-ray, the pixel depth, or what’s out there.
You’d be better off not doing it anyway. Yeah. I want to dive into keyword difficulty, though. So Steven, you know, give us your perspective on general difficulties like that; you know, the industry has more to do. Sure. I mean, it’s sort of like the same situation with search volume. The issue with keyword difficulty is that it essentially applies to the entire thing, which means it applies to everyone, right?
It’s a generalized metric. If it applies to everyone, remember the old adage, “If it applies to everyone, no one trusts,” right? Yeah. I mean, you got it, Steven. It’s exactly like search volume. Search volume is great, right? It gives me an order of magnitude of interest and demand.
So it’s a whole lot. Perhaps a little. only a little bit, right? That’s all it’s giving me—a directional order of magnitude. The same thing goes for general difficulty, and I’ll give you the quick litmus test of search volume last time, your search volume, and your estimated traffic.
Go back and look at it when you achieved the ranking, and how far off were you? Right? You’re going to be significant. I believe the averages we looked at for the teams were wrong the majority of the time and by more than 100% or something wild like that. You don’t think those are difficult, you know?
The Limitations of Averages
I like this litmus test. Right. So this is a 50, and this is a 63. Okay. What would you do differently? Right. Yeah. These are orders of magnitude, and it’s also very biased. And it’s biased because I’m bringing some prejudice about who I am to the table. So I’m saying that I believe that I’m not using data for this, and thus I can use off-page metrics in isolation.
So, what does that mean? The old process of looking at the top 10 ranking pages and saying, “What are their DAS? What are their off-page factors?” and saying, “I’m a 60.” The average here is 60, so I can rank. It assumes that you have authority on that or related topics because these are topic clusters. But could Market Muse rank for, you know, handcrafted Colombian coffee if it was a 60
No. Right. We don’t write about coffee. We have nothing. We have no expertise. So you’re bringing in some bias naturally, but you don’t know how to quantify the bias, right? The bias is a competitive advantage. The competitive advantage Is topical authority, right? The punchline here, Steven, is that Market Muse calculates topical authority.
We all know that. We know that. But the most important thing you take away here is that if you don’t have that, you’re bringing in the bias and the error, right? Because you’re guessing at that, and then you’re guessing at averages, and you’re guessing why other people ranked without actually looking at the quality of their content or their breadth and depth of coverage.
Okay. So, no, I’ve said a mouthful here. But you look at these pages and say that you can make those assumptions. So here’s the other methodology that I love to get a quickie estimate, right? However, it does not provide you with a dependable fork. This thing. Let’s look at a page and say, “I’m going to give you one trick that’s really good, and then I’m going to give you one that’s very error-prone.”
“Look at the range of,” you’ll frequently hear someone say. On HFS or Sim Rush, it is predicting when you search for this result, and it says, basically, that if I create a page that ranks for the topic cluster, it will on average get me the average of all these traffic values.
Okay? Right. However, you can also look at it from a statistical standpoint. At least 97 visits and entrances, and as high as 624 Okay. Reasonable. Okay. Well, the error that’s built into that exists in that second orbit, that term “pool multiplier,” and it also tells the story of 97 versus 6-24. That’s wild.
The Inaccuracy of Reporting
That’s wildly different. Yeah. Thank you for 32. Compare phrases to one 20. Isn’t that four times? It tells a story about what it tells about YMCA music. Aren’t HubSpot, em, Rush authorities? They’re getting the bonus. Plus, it tells a story about the inaccuracy of reporting. So I’m just going to give you, you know, a quick example of this.
Let’s use one where we know the traffic. Five 12. Let’s see what SEMrush says. One 70. Okay, SIM says 174. Ari says five and twelve. What’s the real number? Do you know the real number? Yeah. You know what I actually do. It’s around; it’s been rising, but it’s still around 300 or so. I think 300
You’re right. Okay. So, if you don’t care about statistical reliability, you don’t have to worry about it. Well, it’s in the middle of the two. That’s right. Okay. But that’s lucky that it’s in the middle of the two, right? Yeah. So 5, 5, 12, more than 300. That’s a lot more, you know? Mm-hmm.
60% or more of 1 72 on 306 or less. What does that do? Yeah. That’s statistically completely unreliable. Right. And imagine that every one of these data points is also that unreliable. Mm-hmm. . Okay. What do you get when you get 10 unreliable data points and you smash ’em together and make another unreliable average?
Right, right. So just be thinking about that. Can you put a finger in the air If you get this search result, it will cost you around $300. Sure. Can you answer why? Not really. And do you believe that that’s going to be accurate and, you know, connected to reality? Now, the quick trick I’ll give you is that there are two tricks here.
You can’t use market news, right? If one, take all of these words and morph them into a. At the very least, you’ll have a trackable term pool. You can get a little better understanding of all this stuff. Then, based on the specific intent of your page, get really, really specific.
Finding Pages with Specific Intent
Find the pages that have your specific intent and try to look at the overlap. 56 over 61 for this one. and this is Jonas’s page, right? What’s that? Surp similar. The overlap of that similarity can give you a little bit closer to a better estimate without having authority. But man, that’s painful. That’s a manual process.
All of these things come down to what we already know to be a slightly unreliable process. So far, so good. Yep. So, as I typically do, I showed you the hard way. I showed you the error-prone way. I’m gonna show. Market Way So this is Market Muse’s inventory. This is the Market Muse topic inventory that’s telling me everything about Clusters 49 concepts that I’ve analyzed for this example, and these are the current rankings.
Using Market Muse’s Inventory
What Market Muse gives me is my competitive advantage. That tells me how hard something is going to be to rank for the general populace. And then, how hard is it? gives me the advantage of knowing that I might be able to maintain my existing ranking or grow by just updating a page or creating a new page, right?
So it’s giving me coaching on my momentum and how predictive this goal is, right? So let’s go back and say, “Hey Steven, I need you to create this page, please.” And you’re like, “Why?” And I say, “Well, I think it’s an easy win.” Well, I know we have a great breadth of coverage, depth of coverage, quality of coverage, lost page factors, and momentum here.
All of the topical authority that was combined represented a competitive advantage. And it’s telling me that if you go out and create one great page that talks about what a cluster page is from within the concept of topic clusters, what is that cluster page? What’s a node? What’s a support page, then? That should suffice.
Cool. Well, all right, now I have to get into the details. I have to understand—well, actually, Steven, we’ve got a collection of pages. We’ve got six or seven. We want to make sure they’re all intertwined. We can make better guesses knowing our estimates because we know we’ll have access to at least a thousand visits across all of these pages.
There’s also an advantage to the turn pool multiplier. We have that tower. We’re not starting work right away.on pink grapefruit seltzer water, right? Guess what? We might get lucky and get one word. It won’t be good, right? But if I’ve already got that Moe, I’ve got access to that second orbit. Now let’s take that a step further and look at the individual page and understand it, right?
So by understanding the individual page, I can see the concepts that tell the story of expertise, and I can understand who’s doing it. a great job with high-quality, comprehensive content, and who maybe isn’t doing all that at all. I can understand whether my page is differentiated or if it’s just like everyone else’s.
Understanding Individual Pages
By analyzing this at the site level, we can see site-level gaps that will make all boats rise. And when all boats rise, those pages rank well for their target and trackable orbit. And then we, as entire massive pages, start ranking for that term pool multiplier. That’s what we want. So I look at this stuff; I’ve got coverage of a lot of these things, but I’m finding things already through cluster analysis, Steven.
Yeah. Gosh, that’s an easy winner. as I’m about to show you. All right. Pillar pages, you know, pillar content, you know, internal links That’s amazing, right? But I can also go in and look at the stuff that I’m ranking for and the stuff that I’m not ranking for through the lens of So I can see the stuff that I’m doing okay with on a general page, but specifically, I could get quicker paths to victory because I know my topical authority.
I can also do a better job estimating upside potential because of this. Okay. So understanding a site-level gap analysis through the lens of topical authority is the critical path to estimating more effect. Understanding who I’m competing with for this particular term topic
Identifying Topic Clusters
Cluster analysis can also tell me how much work I need to do for this specific page. And in this case, you know, we are doing well, but we want to make sure that we’re writing the best, most differentiated page in the world. We can then see what people have done. And in this case, people are talking about what a topic cluster is.
They’re not getting into topic cluster analysis, which is something that we do want to get into, and we can see the concepts that everybody covers and just see the concepts that nobody’s talking about. Topics related to Google search, cluster analysis, topic, and clustering are concepts that nobody’s talking about.
I like to say the money’s in the red lines of this view because these are things that the market is saying are covered. Look at my gaps if you were an expert, but no one is currently covering them. Clustering algorithm, relevant topics, topic cluster analysis, topic clustering, core topics, broad topics, related topics, digital marketing
Reaching the Minimum Amount of Traffic
These are things that, if we were to focus on them, would give us a leg up against the current field. And we already know that we have a great chance of succeeding if we focus on this word. So that gives me that one-two punch. Anything else we want to cover here? Overall, my goal here is to tell teams that they can do a better job getting to the minimum amount of likely traffic that this page will get if they use these practices.
And more of the content that they create will be successful; thus, their hit rate will be higher, and then they’ll be able to predict those outsized winners more successfully. Gosh, that’s a long way from just looking at search volume and multiplying it by 2%. Isn’t this Steve? Oh, it sure is.
Awesome. Well, thanks. SEO Arcade. Market’s Jeff Coy and Steven Jetski here. I’m hoping, hoping, hoping that you stop using search volume alone to predict and estimate page traffic. Thanks again.
Overview Summary Of Topic Clustering for SEO Forecasts
Jeff Coyle, the co-founder and chief strategy officer for Market Muse, is joined by Steven Jesky to discuss effective methods of estimating and predicting traffic potential. This is important to understand the potential of a new page or an updated collection of pages for market research purposes.
Forecasting SEO Outcomes With Topic Clusters
The north star of the discussion is the topic volume or search volume. However, calculating and predicting traffic based on the search volume is inaccurate, as every search result is different, and the organic results are at different points from a pixel depth. Also, only looking at the search volume for a particular word doesn’t tell the whole story of the page. Thus, it is essential to consider the decoupling of search volume from what the page is going to get.
Main Points Of Topic Clustering
The focus term or the target keyword should not be the only focus when estimating the traffic potential of a page. There are many other things that the page will rank for, and it’s essential to include them to get a clear picture of the page’s potential.
The term “pool multiplier” is an essential concept to consider when estimating traffic potential. This refers to words that you cannot predict will be linked to because they are low-volume or are not reported in keyword research tools. The more authoritative you are on this topic and semantically related topics, the more topical authority you have, and the more access you have to that second orbit, resulting in more traffic.
It is essential to get a clear picture of the traffic potential of a page to understand its market research value. Taking into account the term pool multiplier is important when estimating traffic potential. The more authoritative you are, the more topical authority you have, resulting in more traffic.
Search Volume and General Difficulty
- Search volume is great for understanding interest and demand.
- General difficulty gives a directional order of magnitude.
- A quick litmus test on search volume is to look at estimated traffic when achieving ranking and compare it to actual traffic.
- The majority of the time, search volume estimates are wrong by more than 100%.
Off-page metrics and topical authority
- The old process of looking at the top 10 ranking pages is biased.
- A topical authority is a competitive advantage in SEO.
- Market Muse calculates topical authority.
- Looking at the range of traffic values can be unreliable due to inaccuracies in reporting.
- A quick estimate of traffic can be useful but is not reliable.
- Unreliable data points can result in an unreliable average.
Takeaways on Topical Clustering
- Use metrics but be aware of their limitations and biases.
- A topical authority is crucial for successful SEO.
- Proper estimation of traffic requires accurate and reliable data.