The Argument Against Topic Clusters and Pillar Pages


Caroline Gilbert of Siege Media published an article in 2018 questioning the merits of the topic cluster model. This blog post has gained a great amount of traction and is something people often find when researching (or questioning) HubSpot's recommended approach to content planning. On the surface, the headline and subheads seem to indicate that planning around clusters is rarely an effective strategy. However, if you take the time to read and truly understand each point Gilbert makes, it actually serves as more of a guide to the proper way to build a cluster plan, while touching on some common pitfalls to avoid.


The Original Article

Below, we address each point the Siege Media article makes and how these can be applied to a successful content strategy. This is not intended to totally discredit the original piece, but more to garner insights we can apply moving forward. Before we dive in too deeply, I want to give you the opportunity to check out the full article and allow it the opportunity to earn your trust. There are some incredible points made here, worth dissecting and putting into context, which will hopefully help us refine our cluster planning.



Their Introduction

Content marketing at scale is hard. Even publishing a few times a week means 100+ posts in the first year. And if you have an even more accelerated calendar, you’ll be left with tons of content to dig through — leaving your first posts never to be read again.

Creating some kind of content taxonomy is helpful,
not only to keep you sane, but make sure your readers can seamlessly get from one post to the next relevant one. Recently, “topic clusters” is the new buzzword to help achieve this.

—When the Topic Cluster Model Doesn’t Work

"Content marketing at scale is hard."

This is true. The original article is intended to help those producing a torrent of new content, or those sifting through a massive volume of existing content, from becoming overwhelmed or misguided in the process. Planning and understanding is important to make sure no efforts are wasted and you don't end up with a mess of content that is thin and ineffective. 100+ posts per year implies you are publishing at least two articles per week, but even at a much smaller volume of one article per month, it is important to make sure the content you are producing is deliberately and effectively working toward an ultimate goal.

"Creating some kind of content taxonomy is helpful."

But to be clear. Taxonomy is a way of classifying and organizing content. There are two different structures at play, though, which need to be understood: one is organizing content for the visitor; the other is organizing the efforts of content developers themselves. These do not need to be organized in the exact same way. A site structure taxonomy is designed to facilitate user experience and includes things like navigation items, topic tags, and url structure. A content plan taxonomy is built around select search queries and topics you wish to rank for.  Both categorize related content—but one is for the user; the other is for your sanity. It's important to avoid the mistake of believing that content only has to be planned around how it's organized on your site.


Misquoting HubSpot

Before this comes off as a call to grab your pitchforks against HubSpot, I should clarify that it’s a solid model. It’s just too often people adopt the new shiny thing because, hey, if HubSpot is doing it, we all should do it, right? But that’s not the right mindset. HubSpot even shares in this article when their model isn’t a good fit.


—When the Topic Cluster Model Doesn’t Work

"When their model isn’t a good fit."

Hang on! What is described in that quote are not examples of when the cluster model isn't a good fit, but rather a description of when a proposed piece of content is or is not a pillar page.

There are, however, a few key points we can take away from the article (or, I guess, from HubSpot's article):

  • If you’re trying to get the page you’re working on to rank for a long-tail keyword, it’s not a pillar page. It's probably a supporting blog post, part of the cluster.
  • If the page you’re working on explores a very narrow topic in great depth, it’s not a pillar page. It's probably a supporting blog post, part of the cluster.
  • If the page you’re working on touches on many aspects of a broad topic, it’s probably a pillar page. And each of those "many aspects" are potential supporting blog posts, part of the cluster.


First Point: Pillars are too niche

Like HubSpot said, pillars should go after very broad terms. This means that the terms you target should be high level, high search volume, and high competition. If your pillar page is too long-tail or niche, you’re not going to populate it with enough [supporting content] to make it look robust enough.

For example, let’s say you’re customer service software Broadly. You already have a blog that’s updated weekly and you’re looking for a new blog architecture. If we were following the topic cluster model, an example pillar may be “customer feedback.” Seems broad enough, right?

But if you look at the long tail related to that potential pillar, you’ll find that volume drops off after five to 10 phrases like “customer feedback questions” and “customer feedback methods.” Even a vague keyword like “customer feedback” only has so many potential long tail after that.


—When the Topic Cluster Model Doesn’t Work

"Pillars are too niche?"

Not necessarily. This argument is that the head term, or keyword, the top-level pillar page is built around has a significant search volume, but the related long-tail articles which support the main topic tend to quickly dry up and have little search volume value, especially after five or so keywords. This makes it hard to come up with enough related long-tail articles to populate a full cluster.

I see exactly where you are coming from, but this argument is based in the past. It supposes that the pillar page will rank for "customer feedback," the first article in the cluster will rank for "customer feedback survey," the next for "customer feedback tools" and so on, when, in reality, the pillar itself should rank well for all of these terms if composed correctly. Each subtopic should be incorporated as a separate section within the pillar and link back to articles which explore these long-tail keywords more extensively.

It's also important not to feel obligated to only include sections that have "customer feedback" as part of the target phrase. Terms such as "customer reviews" or "net promoter score" should absolutely be incorporated into the "customer feedback" pillar, and have supporting articles in that cluster as well.  

The other factor ignored here is the value of related content. Linking to an article built around a low search volume term like "improving NPS survey response rates" may return minimal results, but demonstrates an overall topic relevance that is appreciated by search engines and, more importantly, the users being served your content.


Second Point: Pillars are too broad

I know — I just told you broad pillars are preferred. But there is a scenario when too broad also means off brand. In order to cover all the potential clusters under the pillar, you’re risking spending a lot of time writing topics that aren’t close to your product or service.

For example, if you’re an automation company like Zapier, your brand promise is increased productivity through automating daily tasks like email, messaging, and more. So a natural broad keyword may be “productivity.” And in fact, Zapier has a productivity tab on their current blog already.

While they probably didn’t set up this category with the topic cluster model in mind, attempting to rank for “productivity” isn’t the right mindset with this pillar.

The keyword “productivity,” if you Google it, doesn’t match to Zapier’s core business model. You’ll see a lot around productivity at work, but it focuses on a variety of methods like time management, healthy habits, and more. Unless Zapier wants to dedicate a good portion of their publishing to these other productivity points, ranking shouldn’t be a core goal.

—When the Topic Cluster Model Doesn’t Work

"Pillars are too broad?"

Potentially. A great piece of advice: when building out a content plan, don't go too broad. Zapier might think they want to be found for a term like "productivity." Until they consider user intent. Someone typing in the query "productivity" is probably not looking for Zapier. (They are probably just confirming the spelling of "producdivity".)

Don't go too broad... unless it's appropriate. It would be wasteful for Zapier to focus too much on a term like "productivity." Too broad, and not especially relevant. But a term like "automation?" Well, that is a term that likely has a comparable search volume and is far more inline with their target audience. Broad, when relevant, is worth the work.

Regarding "productivity" being a tab in the Zapier blog, that goes back to the different types of taxonomy. Blog tags and tabs are for user experience, not content planning. A "productivity" section makes a great deal of sense as a way to organize content for those visiting their blog, but as we've seen, probably not to frame a cluster.


Hang on a second. Did you say...

"In order to cover all the potential clusters under the pillar?"

Ummm. Not to get too distracted here, but there might be a fundamental misunderstanding in this original article about the relationship between pillars and clusters. A pillar is the centerpiece of a cluster. There are not multiple clusters under one pillar (unless we are talking about a resource pillar, of course). You should be building one cluster of related content around one central pillar page. This may be where some of the initial confusion comes into play. If you are trying to map an entire cluster of content for each section of a pillar, yes, you will run out of keywords pretty quickly. And, if you are building a different piece of content for each iteration of those keywords, you are going to end up with a ton of overlap. Speaking of overlap...



Third Point: Pillars have too much overlap

Most companies... focus on a specific industry and serve a niche need. Some clients have come to us with a list of pillar ideas that are way too similar, despite maybe having different search volume or SERPs.

Google is getting smarter and smarter at making connections. Even if today the SERP between “moving companies” and “moving” is slightly different, they are essentially the same word and Google is serving us very similar results. So, if I’m a self storage company, there’s no reason to create two pillars for each of these broad terms.

—When the Topic Cluster Model Doesn’t Work

"There’s no reason to create two pillars for each of these broad terms."

Correct. There is no reason to try and build an entire collection of articles around the term "moving companies" and "moving." These are too similar and will likely serve the same search queries. Even if whatever keyword tool you are using tells you there is a different search volume on one query versus the other, this does not qualify them as separate topics to build content around.

Natural language processing is a search engine's understanding of user intent. Regardless of if you search for the terms "foot doctor" or "podiatrist," Google knows what you need. You do not need to build an entire cluster of content around the head term "foot doctor" AND an entire cluster around the head term "podiatrist." If you do, you will likely have very redundant content and too much overlap.  

Regardless if you search "foot doctor" or "podiatrist," Google knows what you need. You do not need to build an entire cluster of content around the head term "foot doctor" AND an entire cluster around the head term "podiatrist." 

This is a legacy way of looking at planning content, based solely on different iterations of keywords and their respective search volume and difficulty. The purpose of topic-based planning, rather than keyword-based planning, is to move away from this content overlap.


Final Point: Pillars compete with product pages

The final error I see with pillar page creation is potential competition with your product pages. It’s smart to build pillars that are on brand and close to product, but too close and you again run into the risk of confusing Google. Except this time, you’re fighting with your bottom funnel pages.

An example and potential solution that I like is Blue Buffalo’s blog. They have a main eCommerce nav at the top and a secondary blog nav with their top six “pillars.” One potential obstacle is their “Dog Food” page and corresponding dog nutrition topics. They’ve addressed this by deoptimizing the pillar anchor text (“Health”).

— When the Topic Cluster Model Doesn’t Work

"Competition with your product pages?"

"Too close, and you run the risk of confusing Google?"

I disagree. Yes, you need to help search engines understand what you want specific pages to rank for, for both product pages and articles. If you have a product page that would ideally rank for your product, and you have additional content that could theoretically also rank when someone is searching for your product, Google is not going to see two or three good results and think, "I don't know which to deliver, so I will deliver neither."

This is actually a great example of situations where articles can help bring added attention to your products. While product pages are absolutely among those you hope are delivered when someone searches for highly relevant terms, one of the shortcomings of these service pages is having a relatively limited amount of space in which to provide a wealth of information. It's often imperative that these pages are kept tight and to the point. As a result, you simply can't include enough content to warrant being considered the best result. Articles, however, can delve deeper.

In the Blue Buffalo example, if a searcher types in the term "puppy food," the company would logically want the Blue Buffalo puppy food product to be delivered. But what's more likely to get served—a Blue Buffalo puppy food product, or an article explaining what makes puppy food different from typical dog food? Do not avoid producing content that could potentially outrank product pages out of fear that it will. It doesn't matter how you get them there; just get them there.

Do not avoid producing content that could potentially outrank product pages out of fear that it will. It doesn't matter how you get them there; just get them there.


The balance of the article is framed as a selection of alternatives to topic clusters; but again, they're really just great suggestions to consider as additions to a proper cluster plan.

  • Properly using blog categories and subcategories
  • Categorizing content based on subject matter rather than the format it's found in
  • Linking priority content in the header or footer (or navigation)


All great advice.


Before we just tell our client "No"

I’ve been hearing a lot from clients about their own topic cluster model and how they’ve adapted it to their needs. This all sounds great!

But this model probably isn’t right for your business.

—When the Topic Cluster Model Doesn’t Work

Totally agree. If your clients are the ones crafting their cluster plan... it's probably not right for their business. They are not content marketers. Clients will always have input and try to provide their own direction, but ultimately, they rely on us to turn this feedback into a cluster plan that will work.

Cluster planning is not always the most appropriate approach, though not for the reasons cited in the Siege Media article. The real test is simply: Is long-term organic ranking going to grow your business? Sometimes, the answer is no. Sometimes, a more targeted ABM campaign, paid social, and landing pages will be more effective in the long run. Putting your efforts into content marketing can be a lengthy and challenging process, but will that traffic ultimately move the needle? We want to get that flywheel spinning, but not in the wrong direction.

Let us know what you think. Leave your comments below.