Most content on topic clusters describes the model in the abstract. This post does something different: it shows you what the model looks like in practice, at named brands, with documented outcomes and honest caveats about what the evidence actually says.
If you need a grounding in what a cluster is before diving in, the guide on topic clusters and pillar pages covers the full framework. Come back here when you are ready to see real examples.
The short version of what a topic cluster is: a pillar page covers a broad topic comprehensively, and a set of spoke pages each drill into one subtopic in depth. Every spoke links back to the pillar, and the pillar links out to every spoke. That bidirectional internal linking structure signals topical authority to search engines and makes it easier for users to navigate a subject area. The examples below show what that looks like at different scales, different industries, and different stages of maturity.
A note on the data before we go further: the figures above, and most reported metrics in this post, come from either Ahrefs traffic estimates (which are modeled approximations, not Google Analytics confirmations) or self-reported practitioner interviews. No controlled study exists that isolates topic clusters as the single variable producing a traffic outcome. We flag the source type for every number so you can calibrate accordingly.
What Makes a Good Topic Cluster Example?
A useful example tells you at least three things: the pillar topic and its scope, how the spokes are structured and interlinked, and what outcome the brand reports or what a third party measured. Many blog posts on this topic present cluster diagrams for fictional brands or cite no specific outcome at all.
The examples in this post meet a higher bar. Each one is a named brand with a publicly available cluster you can inspect, a documented structure, and a reported or independently measured result. Where the evidence is weak, we say so.
HubSpot: The Original Model and What Happened Next
HubSpot created the term "topic cluster model" in a foundational April 2017 post authored by Mimi An with contributions from Matthew Barby and others. The worked example in that post used "Content Marketing" as the pillar, with 16 spokes including content strategy, buyer personas, blogging, and content planning. The framing question for pillar scope: "Would this page answer every question a reader searching for that topic had, AND is it broad enough to anchor 20 to 30 posts?"
That question is still one of the cleanest tests for pillar selection.
When HubSpot applied the model to their own 12,000-post blog estate, they documented the process in a follow-up case study. The reported outcome: positive month-over-month growth in first-page keyword rankings, and hundreds more keywords moving toward page two and three. No specific traffic numbers were published alongside these claims.
A more recent and more rigorous data point comes from Ahrefs, which independently audited HubSpot's SEO strategy in June 2024. Ahrefs found that HubSpot's blog generates approximately 8.2 million estimated organic visits per month and maintains five major content hubs that collectively receive links from around 36,000 locations across HubSpot's own properties. The critical finding: despite that massive internal link investment, the hub pages themselves generate very little direct organic traffic. Ahrefs described this as ironic given HubSpot's role in popularizing the model.
The lesson is not that topic clusters do not work. It is that the value of a cluster often flows through the spoke pages, not the pillar page. The pillar attracts links and signals topical depth; the spokes rank for long-tail queries and generate traffic. Expecting the pillar itself to drive traffic at scale is a misreading of how the model works.
What to copy from HubSpot
The pillar-scope test ("broad enough to anchor 20 to 30 posts") is directly useful. The internal linking structure, where every spoke links back to the pillar and the pillar links out to all spokes, is the minimum viable implementation. The mistake to avoid: treating the pillar page as the traffic engine rather than the authority anchor.
Zapier: The Self-Building Hub
Zapier's remote work content hub at zapier.com/learn/remote-work/ is the most thoroughly documented example of what Ahrefs calls a "self-building" hub: a structure that grows over time by incorporating related articles that were not originally created for the hub.
When Zapier launched the hub in 2017, it linked to 14 pieces of content. By 2023, when Ahrefs published a case study of Zapier's SEO strategy, the hub had grown to more than 50 links across seven categories, incorporating articles published throughout the intervening years. Ahrefs measured the hub as having accumulated 4,400 backlinks from 1,100 domains. Estimated monthly traffic to the hub itself was 1,100 visits, which is modest.
That traffic number is the interesting part. The hub's primary value is not the traffic it generates directly. It is the topical authority signal it builds and the backlinks it attracts. Ahrefs estimated Zapier's blog overall at 1.6 million monthly visits, worth roughly 3.7 million dollars in equivalent paid traffic. The remote work hub is a structural contributor to that authority, not the direct source of it.
The self-building approach is worth understanding in detail. Zapier did not write 50 articles specifically for the hub and launch them all at once. They published content related to remote work over several years, then added those articles to the hub retrospectively. The hub became more comprehensive over time with less per-article overhead than a planned cluster build would have required.
What to copy from Zapier
Start the hub with a small number of high-quality pieces. As you publish related content, add it to the hub rather than letting it float as an unconnected page. Build a consistent internal table of contents that readers can navigate. The hub compounds without requiring a large upfront production commitment.
Drift: Link Acquisition Through Hub Architecture
Ahrefs cited Drift's chatbot content hub at drift.com/learn/chatbot/ as a strong example in their 2020 resource on content hubs. The hub used a custom table-of-contents design linking to subpages covering how chatbots work, the benefits of chatbots, and common use cases.
As of Ahrefs' 2020 measurement, the hub had attracted over 500 linking domains and an estimated 6,400 monthly organic visits. Note: these are Ahrefs tool estimates from 2020, and current numbers may differ substantially given Drift's subsequent acquisition and rebrand.
What made the Drift hub notable was its design. Rather than building a generic blog index, they built a structured learning resource with a visual architecture that communicated comprehensiveness at a glance. That design is a signal in itself, to readers and to sites considering whether to link to it.
What to copy from Drift
Hub design matters. A well-structured hub page with a clear visual hierarchy and obvious navigation between sections earns links more readily than a list of loosely related blog posts. Treat the hub page as a product, not a content index.
Hotjar: B2B SaaS Clusters Tied to Conversion
The Hotjar case is the most commercially compelling, and also the least independently verified. In a 2022 practitioner interview with Ben Goodey, Sean Potter, then Senior Content Strategist at Hotjar, described their approach and reported results.
Hotjar's cluster structure used 5 to 8 articles per cluster, starting small and expanding based on performance signals. They published 50 to 100 new articles per month using a combination of in-house staff and freelancers. Cluster topics came from customer and prospect conversations, not solely from keyword tool metrics, which is worth noting as a sourcing approach.
The "Website Tracking" cluster was the specific example Potter described: a pillar guide, a how-to on user activity tracking, a comparison article, benchmarking content, and a privacy-focused variant. Each piece addressed a different stage or angle of the same user need.
Reported outcomes (self-reported by Potter, not independently audited): 47% organic search traffic increase and 20% increase in product sign-ups, measuring January to November 2022 versus the same period in 2021. The Website Tracking cluster alone was associated with a 112% increase in sign-ups and a 66% increase in new customers.
We flag these as self-reported because Hotjar's analytics were not independently accessed or audited. Potter is a named, credentialed source, and the figures are internally consistent, but treat them as strong practitioner evidence rather than verified data.
What to copy from Hotjar
Sourcing cluster topics from customer conversations, not just keyword tools, surfaces queries that are commercially relevant but may not appear in volume-heavy keyword research. The 5 to 8 article cluster size is also a useful starting point: large enough to signal topical depth, small enough to execute without a massive initial investment. Track conversions at the cluster level, not just traffic, to understand which clusters are doing commercial work.
Backlinko: Hub Architecture Built for Authority
Backlinko's SEO Marketing Hub takes a different structural approach from the pillar-and-blog-posts model. Rather than a blog with internally linked posts, the hub is a standalone learning resource with approximately 53 articles organized into eight sections, navigable via a sticky table of contents.
Backlinko describes the hub's performance as ranking for 29,000-plus keywords and driving over 158,000 visitors, with nearly 165,000 backlinks. These are self-reported figures, so treat them as directional rather than audited. The YouTube Marketing Hub, built with the same architecture, is cited as ranking for 17,000-plus keywords.
What is notable here is the hub format itself. Backlinko does not use a standard blog post as the pillar. They build a dedicated resource with its own URL structure, navigation, and visual design. The hub functions as a product in the content estate, not a category page.
This approach requires more design and development investment upfront. The tradeoff is a resource that reads as authoritative because its format signals comprehensiveness, and that earns links because it genuinely functions as a reference.
What to copy from Backlinko
If your pillar topic is competitive and important enough to justify the investment, a purpose-built hub page with its own navigation outperforms a long blog post with H2 headings. The format communicates depth before the reader has read a word.
The Patterns That Appear Across All Examples
Looking across these five examples, a few consistent structures emerge.
Hub pages attract links; spoke pages generate traffic. This shows up in both the Zapier data (1,100 hub visits but 4,400 backlinks) and Ahrefs' HubSpot audit (massive internal link investment, modest hub-page traffic). The cluster value accrues to the spoke pages ranking for long-tail queries. Pillar pages are link targets and authority signals, not primary traffic generators.
Cluster size varies more than you would expect. HubSpot recommends 20 to 30 spokes per pillar. Hotjar uses 5 to 8. Ahrefs documents examples at 8, 40-plus, and 700-plus pages. The right size depends on the topic's breadth and your production capacity. A tight, high-quality cluster of 6 articles outperforms a sprawling cluster of 30 thin ones.
Self-building is underrated. Zapier's approach of incorporating existing content into a hub retrospectively is lower-cost and often produces more authentic coverage than building a cluster from scratch. Audit what you already have before planning what to write.
Bidirectional linking is non-negotiable. Every example uses explicit links from spokes to the pillar and from the pillar to each spoke. This is the minimum structural requirement. Spokes that do not link back to the pillar are disconnected from the cluster's authority signal.
Cluster topics sourced from customers outperform those sourced only from tools. Hotjar's practitioner account makes this explicit. Keyword tools surface volume; customer conversations surface intent. The highest-converting clusters address real decision-stage questions, not just informational queries with high search volume.
Sources: Ahrefs content hub case study (Hardwick, 2020) for Drift; Ahrefs Zapier SEO case study (Makosiewicz, 2023) for Zapier. All figures are Ahrefs modeled estimates, not GSC-confirmed. Drift data from 2020, Zapier data from 2023. Selection reflects documented examples only -- this is not a representative sample of all topic cluster outcomes.
Pitfalls to Avoid When Copying These Models
Treating pillar page traffic as the primary success metric. The evidence consistently shows that hub pages attract links but often underperform on direct traffic. If you optimize for pillar-page ranking and ignore spoke-page performance, you will misread how well the cluster is working.
Publishing thin spokes to hit a cluster size target. Hotjar's 5 to 8 article clusters outperformed expectations because each article was genuinely useful, not because the cluster was large. A cluster of 6 depth-appropriate spokes is better than 20 shallow ones. For a systematic view of where your existing content is thin, the content gap analysis guide covers how to audit at the topic level.
Skipping the internal linking. The structural connection between pillar and spokes is not optional. It is the mechanism through which the cluster signals topical authority. Many content teams publish the content and skip the linking work. That is not a topic cluster; it is a set of thematically related posts.
Neglecting to link spokes to each other where relevant. The examples above link pillar to spoke and spoke to pillar. The strongest clusters also include spoke-to-spoke links where one article naturally continues from another. This makes navigation more useful for readers and reinforces the cluster's topical coherence. For the full approach to this, see the internal linking for SEO guide.
Not auditing what you already have before building new clusters. Zapier's approach of self-building works because they recognized existing content that could slot into the hub. Most content teams have relevant content that is not yet connected to their cluster architecture.
How to Map Your Own Cluster
We have found a four-step process useful for moving from example to implementation.
Step 1: Identify a pillar topic. Use the HubSpot scope test: is it broad enough to anchor 15 to 30 spoke articles, but not so broad it becomes a category? "Content marketing" qualifies. "Marketing" does not. "Instagram Reels" might be too narrow.
Step 2: Audit existing content. Before writing anything new, list what you already have that covers the topic area. These are potential spokes. Identify which ones already link to a pillar and which are floating.
Step 3: Map the gaps. What subtopics are you missing? What questions do your customers ask that are not yet answered by an existing page? These gaps are your spoke candidates. The content gap analysis guide covers systematic approaches to finding them.
Step 4: Build the bidirectional links. Update the pillar to link to all confirmed spokes. Update each spoke to link back to the pillar. Where spokes naturally follow from each other, add spoke-to-spoke links too.
Then publish the cluster, monitor which spokes generate traffic and conversions, and use that signal to prioritize the next round of spoke content.
FAQ
How many articles should a topic cluster have?
There is no universal rule. HubSpot's original framework suggests 20 to 30 spokes. Hotjar uses 5 to 8. Ahrefs documents clusters ranging from 8 articles to 700-plus. The right number is enough spoke pages to cover the topic's real subtopics, not a specific count. Start with 5 to 8 high-quality spokes and expand based on which subtopics are underperforming or missing.
Can I build a topic cluster around content I already have?
Yes, and this is often the most efficient approach. Zapier built its remote work hub largely by incorporating content that already existed. Audit your published content, identify pieces that cover subtopics of the pillar, add internal links, and build or improve a pillar page to anchor them. New content fills the gaps the existing content does not cover.
Does the pillar page need to be a dedicated page or can it be a blog post?
Both formats work, and examples in this post use both. HubSpot and Zapier use dedicated hub URLs with custom navigation. Backlinko builds standalone resource pages. Many teams use a long-form blog post as the pillar. A dedicated page with its own navigation communicates comprehensiveness more clearly and often earns more links, but the structural requirement (bidirectional linking with all spokes) is the same regardless of format.
How do I know if a topic cluster is working?
Track three things: (1) organic traffic to spoke pages over time, not just the pillar; (2) featured snippet capture rate across the cluster's keyword set; (3) if you can tie it, conversions or sign-ups attributed to organic traffic from cluster pages. Backlinko's SEO Marketing Hub attributes strong backlink growth. Hotjar tracked sign-ups at the cluster level. Pillar-page traffic alone understates cluster performance.
What is the difference between a topic cluster and a content hub?
The terms are often used interchangeably, but there is a structural distinction. A topic cluster is the architecture: a pillar page with a set of linked spoke pages. A content hub is more often a design pattern: a dedicated URL with custom navigation that houses the pillar and provides visible links to all spokes. Every content hub implements the topic cluster architecture, but a topic cluster can be implemented without a dedicated hub page.
The evidence across these examples points in one direction: topic clusters work most reliably when the spoke pages are genuinely useful, the internal linking is complete, and the cluster is evaluated on spoke-level traffic and conversions rather than pillar-page rankings alone. The brands that report strong outcomes built clusters around real customer needs and maintained them as the cluster matured.
At SparkBlog, we treat the content estate as the unit of planning, which means clusters are a natural starting point, not an afterthought. Building a well-structured cluster from the start, using a content brief that specifies the pillar and the spokes, makes the entire downstream production process faster. If you are working on briefs for a cluster, the how to write a content brief guide covers how to encode the cluster architecture into every brief before drafting begins.