The SEO landscape has fundamentally shifted. While we once obsessed over individual keywords and their monthly search volumes, Google’s algorithm now thinks in topics, themes, and subject matter expertise. Yet most of us are still drowning in spreadsheets filled with thousands of individual keywords, struggling to see the bigger picture.
If you’ve ever exported your Google Search Console data and felt overwhelmed by the sheer volume of queries—tens of thousands of variations, modifiers, and long-tail searches—you’re not alone. The real challenge isn’t collecting this data; it’s making sense of it.
Here’s the uncomfortable truth: the topics Google considers you an authority on might not align with what you think are your core business topics. Understanding this gap is crucial for developing a content strategy that works with the algorithm, not against it.
The Limitations of Traditional Keyword Analysis
Traditional keyword research and query mining gives us valuable data, but it keeps us trapped in a keyword-centric mindset. When you’re staring at 25,000 individual search queries, patterns exist, but they’re nearly impossible to spot manually.
Consider what happens when you try to analyse this data conventionally:
- You might group keywords by product category, but miss semantically related queries
- Manual categorisation introduces bias based on what you think matters
- You lose sight of emerging topic clusters that users actually care about
- Business priorities cloud your view of where you genuinely have search authority
The solution? Let machine learning reveal the patterns hiding in your data.
Converting Keywords to Topic Clusters: A Technical Walkthrough
The process of transforming individual keywords into strategic topic clusters involves several sophisticated steps, but the concept is straightforward: use AI to group semantically similar keywords, then identify what those groups represent.
Step 1: Export Your GSC Performance Data
Start with a comprehensive dataset. Export at least six months of keyword data from Google Search Console, including:
- Search queries
- Impressions
- Clicks
- Average position
- Click-through rate
The larger your dataset, the more reliable your topic clusters will be. A dataset of 20,000- 40,000 keywords might consolidate into 300-500 meaningful topic clusters, depending on your site’s breadth.
Step 2: Convert Keywords to Vector Embeddings
This is where the magic happens. Vector embeddings transform text into high-dimensional numerical representations that capture semantic meaning. Keywords like “best running shoes” and “top trainers for jogging” might look different textually, but their vector representations will be remarkably similar.
Tools like Google’s Vertex AI text-embedding model or OpenAI’s embedding models can convert each keyword into a vector. These vectors exist in multi-dimensional space where semantically similar terms cluster together naturally.
Step 3: Apply Clustering Algorithms
Once your keywords exist as vectors, clustering algorithms can identify natural groupings. HDBSCAN (Hierarchical Density-Based Spatial Clustering of Applications with Noise) is particularly effective here because it:
- Doesn’t require you to specify the number of clusters in advance
- Identifies clusters of varying densities
- Labels outliers as noise rather than forcing them into inappropriate groups
Before clustering, dimensionality reduction through PCA (Principal Component Analysis) helps manage computational complexity while preserving the relationships between keywords.
Step 4: Label Your Topic Clusters
You’ll now have hundreds of keyword groups, but they’re just numbered clusters at this point. Use GPT-4 or similar language models to automatically generate descriptive labels for each cluster based on the keywords it contains.
A cluster containing “mortgage rates,” “home loan interest,” “refinancing costs,” and “mortgage APR” becomes “Mortgage Interest Rates and Costs”—immediately more actionable than “Cluster 47.”
What the Results Reveal About Your Search Authority
The output is transformative. Instead of managing 20,000 individual keywords, you’re now working with a few hundred strategic topics. But the real insight comes from analysing which topics drive performance.
The Authority Gap
When you attach impressions and clicks data to each topic cluster, patterns emerge quickly. You might discover:
- High-impression, low-click topics: Google shows your content for these searches, but users aren’t engaging. Your authority exists, but your titles and descriptions need work.
- High-click topics: These are your authoritative sweet spots. Google trusts you, and users validate that trust with clicks.
- Zero-click topics: Over 200 topics driving zero clicks? These represent semantic authority without traffic value—potential opportunities or content to deprioritise.
Business Alignment Analysis
Here’s where many SEOs experience a revelation. Plot your topic clusters against your business priorities. You might find:
- Strong search authority in tangential topics that don’t drive conversions
- Weak authority in core business topics where you need to dominate
- Unexpected opportunities where slight authority could be amplified
This misalignment isn’t failure—it’s intelligence. Now you can make strategic decisions about where to invest content resources.
Turning Insights into Action
Understanding your topical authority is just the beginning. The real value comes from applying these insights to your content clustering strategy.
Double Down on Existing Authority
For topics where you already have strong algorithmic trust:
- Create comprehensive pillar content that consolidates your authority
- Build out cluster pages addressing specific subtopics
- Update existing content to reinforce your expertise
- Acquire backlinks from authoritative sources in these topic areas
Bridge the Authority Gap
For core business topics where you lack authority:
- Audit competitor content to understand what Google expects
- Develop in-depth, expert-level content that demonstrates genuine expertise
- Build internal linking structures that signal topic relationships
- Consider whether the topic genuinely aligns with your expertise
Visualise Your Topic Landscape
Data visualisation transforms abstract topic clusters into actionable insights:
- Treemaps show the relative size of each topic cluster based on impressions or clicks
- Scatter plots reveal the relationship between impressions and clicks for each topic
- Heat maps identify which topics drive actual business value versus vanity metrics
Replicating This Analysis
The technical barrier to entry has lowered significantly. Modern AI tools make this analysis accessible to SEOs without machine learning expertise:
- Export your GSC data with all available metrics
- Use embedding APIs from OpenAI, Google, or Anthropic to vectorise your keywords
- Apply clustering algorithms through Python libraries like scikit-learn
- Use GPT-4 for automated cluster labeling
- Visualise results in tools like Tableau, Python plotting libraries, or even Google Sheets for smaller datasets
The entire process, once set up, can be rerun monthly to track how your topical authority evolves over time.
The Shift from Keyword Management to Topic Strategy
This approach fundamentally changes how you think about SEO. Instead of:
- Tracking rankings for hundreds of individual keywords
- Creating separate content pieces for minor keyword variations
- Losing sight of thematic authority in granular data
You’re now:
- Managing strategic topic clusters aligned with user intent
- Building comprehensive content that addresses entire themes
- Making data-driven decisions about content investment
- Aligning search authority with business objectives
The question isn’t whether your content ranks for specific keywords. The question is: which topics does Google’s algorithm trust you to speak authoritatively about, and does that align with where your business needs authority?
Moving Forward
Start simple. Export your GSC data and analyse your top 1,000 keywords first. Even a smaller dataset will reveal patterns you’ve been missing. The goal isn’t perfect classification—it’s better understanding.
As you develop this capability, you’ll stop chasing individual keywords and start building genuine topical authority. That’s not just better SEO—it’s better business strategy.
The algorithm has already made the shift to topics. The question is: have you?
Original Idea from Dan Hinkley