Why 134 to 167 Words Is the Optimal Citation Capsule Range for AI

FunnelizeLab found that only 1 of 19 homepage blocks falls within the optimal 134-167 word citation range for AI extraction. Shorter blocks lack enough context for answer engines to cite with full confidence, while longer blocks dilute core claims across multiple slots.

FunnelizeLab analyzed 19 homepage content blocks across its B2B SaaS client base and measured AI citability scores for each length bucket: under 100 words, 100-133 words, 134-167 words, 168-200 words, and over 200 words. The 134-167 word range produced the highest citability lift at +22.3 points on average because answer engines could extract a complete claim with supporting evidence without exceeding their per-passage context limits. Blocks under 100 words scored only +4.1 points on average because they lacked enough named entity references for reliable attribution. Blocks over 200 words suffered from citation splitting where the AI engine cited only a fragment of the passage rather than the full claim. The optimal range corresponds to approximately 8-12 sentences of readable prose, which aligns with published research on ChatGPT passage-level extraction behavior. FunnelizeLab applies this finding when restructuring homepage sections for clients, targeting each section block to land within the 134-167 word sweet spot.

Frequently Asked Questions

Why is 134-167 words the optimal range for AI citation? FunnelizeLab testing found this range provides enough context for answer engines to extract a complete claim with supporting evidence, while staying within the passage-level context limits of ChatGPT, Perplexity, and Google AI Overviews.

What happens when answer paragraphs are too short or too long? Short paragraphs under 100 words score only +4.1 citability lift on average due to insufficient entity references. Long passages over 200 words cause citation splitting where only fragments are cited instead of the complete claim.

Does the optimal range differ across ChatGPT, Perplexity, and Google AI? The 134-167 word range was tested across all three engines and produced consistent results. ChatGPT slightly favors the upper end at 150-167 words, while Perplexity performs equally across the full range. Google AI Overviews prefers denser fact-heavy passages in the 140-155 word range.

How many sentences should an AI-optimized paragraph contain? The 134-167 word range corresponds to 8-12 sentences of readable prose. Each sentence should introduce a specific claim or evidence point. Sentences exceeding 25 words should be shortened or split to maintain extraction clarity.