The Pillar-Cluster Model, Refined

The classic pillar-cluster model has aged. Here’s the 2026 take — same structure, sharper rules.

What changed

Google’s NLP understands topics, not URLs. The pillar-cluster model assumed Google needed explicit hub-and-spoke structure. It doesn’t — but the structure still helps humans and parsers. The change: tighter cluster boundaries, fewer cross-links, more decay work.

The 2026 model

1. Pillars are tight. 3,000–4,500 words, not 8,000. Comprehensive on one sub-topic.
2. Spokes are specific. One query per spoke. No “Ultimate guide to X” spokes.
3. Cluster boundary. A spoke belongs to one pillar only. Don’t link spokes across pillars unless genuinely relevant.
4. Internal links. Spoke → pillar (one). Spoke → spoke (siblings, 2–4). Pillar → spoke (all).
5. Decay work. Quarterly review. Refresh stats. Add 2026 section if needed.

What doesn’t work

• 5,000-word pillars that try to cover a whole niche.
• Spokes that duplicate content.
• Cross-cluster linking that confuses the topical signal.
• Pillars never updated.

The exception: the comparison cluster

“X vs Y” and “Best X for Y” queries form their own cluster. They convert highest. Keep them separate from informational pillars — they cannibalize each other if mixed.

The trap

Cluster boundaries become personal preference. Validate with SERP overlap (see our keyword research guide). The SERP is the source of truth.