Content Cannibalization: When Your Own Pages Compete Against Each Other

Content cannibalization happens when multiple pages on your site target the same topic or query. Instead of one strong page that AI confidently cites, you have three weak pages that split the signal.

Put on ChatGPT's glasses. Someone asks about live chat pricing. Your site has: - /blog/live-chat-pricing-guide (written in 2024, 800 words) - /pricing (current pricing page, 400 words) - /blog/how-much-does-live-chat-cost (written in 2025, 600 words)

Which one should AI cite? They all sort of answer the question. None of them is clearly the definitive resource. So AI does what any reasonable system would do - it moves on to a competitor site that has one clear, comprehensive pricing page.

Content Cannibalization carries 2% weight in the Technical Plumbing tier. That's relatively low on its own, but the indirect damage is much larger. Every pair of competing pages dilutes the authority signal that could make one page dominant. It's not just 2% you're losing - it's the citation opportunity across every criterion those pages touch.

The scorer works by comparing page titles and content across your site sample. It builds similarity scores between page pairs, filters out site-wide theme terms (so a SaaS company isn't penalized for mentioning "software" on multiple pages), and flags pairs that exceed the competition threshold. The output is a ratio-based score: fewer competing pairs = higher score.

Content Cannibalization has a direct cross-criterion link to Topic Coherence. When your coherence score is low (below 6/10), the cannibalization checker becomes more aggressive.

The logic: a focused site naturally has multiple pages about related topics. That's normal and expected - a customer support company should have many pages about support topics. The scorer accounts for this by filtering out common theme terms. "Customer support" appearing in ten page titles on a support company's site isn't cannibalization - it's coherence.

But when coherence is low, those shared terms aren't evidence of thematic focus - they're noise from a scattered content strategy. The filter becomes stricter, and more page pairs get flagged as competing.

Additionally, when coherence drops below certain thresholds, the cannibalization score gets capped at 6 regardless of actual pair counts. A scattered blog with a few competing pages is treated more harshly than a focused blog with the same number, because the competing pages are more likely to actually confuse AI about which page to cite.

Translation: fix coherence first. Many cannibalization problems resolve naturally when you consolidate scattered content into focused topic clusters.

Finding duplicates: Search your own site for your key topics. If the same query returns 3+ of your own pages, you likely have cannibalization. Check for: - Blog posts and product pages covering the same topic - Old and new versions of the same article - Slightly different titles covering identical ground ("Live Chat Guide" vs "Guide to Live Chat" vs "Complete Live Chat Guide") - Category pages and individual pages with overlapping content

The consolidation playbook: 1. Pick the strongest page for each topic (highest word count, most links, most traffic) 2. Merge the unique content from competing pages into the winner 3. 301 redirect the losers to the winner 4. Update internal links to point to the consolidated page

When to differentiate instead of consolidate: Sometimes two pages legitimately serve different intents. "Live Chat Pricing" and "Live Chat ROI Calculator" both involve live chat costs, but they answer different questions. In this case, make the differentiation explicit: - Give each page a clearly distinct title and H1 - Ensure the opening paragraph states what makes this page different - Cross-link between them with descriptive anchor text ("See our pricing" from the ROI page, "Calculate your ROI" from the pricing page)

Canonical tags as a last resort: If you must keep two similar pages (e.g., regional variations), use canonical tags to tell AI which version is authoritative. But consolidation is almost always better than canonicalization.

Start here: search Google for site:yoursite.com [your main product term]. If you see 3+ results that look like they answer the same question, consolidation is needed.

Content Cannibalization carries 2% weight in the Technical Plumbing tier. Sites with zero or minimal competing page pairs score 8-10/10 on this criterion. Sites with 3+ pairs of competing pages score 3-5/10. Sites with widespread duplication across their content inventory can score 0-2/10.

The 2% weight is deceptive because cannibalization's damage extends far beyond its own criterion. When two pages compete for the same citation, neither page gets the full authority signal. A single 2,500-word comprehensive page about "live chat pricing" will outperform three 800-word pages covering the same topic - not just on cannibalization scoring, but on Content Depth, Internal Linking authority flow, and Query-Answer Alignment. The competing pages fragment the signal across all these criteria.

In practice, we've seen sites gain 5-8 points on their overall AEO Site Rank from a consolidation sprint alone. One e-commerce site had 12 blog posts covering variations of their core product topic. After consolidating them into 4 clearly differentiated guides (each 2,000+ words), their AEO Site Rank jumped from 42 to 51. The content was largely the same - the architecture changed.

The "refresh" that creates a duplicate. A content team publishes "Live Chat Guide 2025," then publishes "Live Chat Guide 2026" without redirecting the old one. Now both pages compete for the same queries and neither has full authority. Always 301 redirect the old version to the new one, or update the existing page in place and change the date.

Blog posts that cover the same ground as product pages. Your /features/live-chat page describes your live chat product. Your /blog/complete-guide-to-live-chat describes... your live chat product. AI sees two pages answering the same question and picks neither. The blog post should answer a different question than the product page - "How does live chat reduce support costs?" vs "What features does our live chat include?"

Tag and category pages that duplicate content. WordPress and other CMS platforms auto-generate category pages, tag pages, and author pages that surface the same content. If your "customer support" tag page shows the same articles as your "help desk" category page, that's cannibalization. Use canonical tags or noindex on the weaker version.

Localized pages with identical content. A /us/pricing and /uk/pricing page with the same copy but different currency symbols creates competition. If the content is truly identical except for locale, use hreflang tags and canonical references to establish the primary version.

Landing pages for different campaigns. Your Google Ads landing page and your organic landing page cover the same product. Without canonical tags pointing to one authoritative version, AI treats both as competing content. Canonicalize to the organic page.

AI engines don't explicitly check for cannibalization the way our scorer does. Instead, they encounter the symptom: multiple pages from your site that could answer the same question, with no clear signal about which one to cite.

ChatGPT handles this by defaulting to the page with the strongest overall signals - highest word count, most structured data, most recent date. But when two competing pages are roughly equal, ChatGPT often cites neither and moves to a competitor site that has one clear, authoritative page. This is the real cost of cannibalization - it's not that AI picks the wrong page, it's that AI picks someone else's page.

Claude applies explicit deduplication when it encounters multiple pages from the same domain that cover the same topic. It selects one representative page based on content depth, recency, and schema quality, then ignores the others. If your best content is on the page Claude discards, you lose that citation entirely. Claude's selection criteria favor pages with Article schema and clear dateModified signals, so ensure your canonical page has the strongest markup.

Perplexity assembles answers from multiple sources and avoids citing the same domain twice for the same fact. If it finds three pages on your site about live chat pricing, it picks one and moves on. The selection is based on content density and answer directness - the page that answers the pricing question most concisely in its opening paragraph wins. The other two pages are wasted content.

Google AI Overviews applies its existing canonicalization signals to resolve competing pages. If you've set canonical tags correctly, AI Overviews respects them. If you haven't, Google's algorithms pick a canonical for you - and they don't always choose the page you'd prefer.