Join
Original Research & Experiments

SEO Experiments & Case Studies

Real hypotheses, transparent methodology, raw numbers, and verifiable learnings. Zero fluff, 100% test-backed evidence.

Flagship ResearchPublished: October 2026

Google AI Overviews Citation Study: Analyzing 100 India Search Queries

🔬 Hypothesis

Pages featuring concise 40–60 word direct answers and original data tables receive significantly more citations in Google AI Overviews than generic 2,500-word articles.

⚙️ Methodology

Tracked 100 conversational and procedural queries in Google India (logged-out browser, neutral IP). Recorded citation sources, sentence structure, and schema markup.

74%
AI Overview Trigger Rate
68%
Had Direct Answer
81%
Used Tables / Lists
92%
Valid JSON-LD
💡 Key Finding & Takeaway:

AI search engines prioritize quotation density over total word count. Pages with clear definitions immediately below H2s were cited in 68% of generative answers.

Technical ExperimentPublished: September 2026

Does llms.txt Actually Influence AI Search Engines? A 60-Day Experiment

🔬 Hypothesis

Publishing a structured /llms.txt file improves crawl frequency and semantic entity accuracy in Perplexity and Claude search agents.

⚙️ Methodology

Implemented /llms.txt with concise markdown summaries and tracked server logs for hits by OAI-SearchBot, PerplexityBot, and Claude-User over 60 days.

1,420+
AI Bot Hits
+24%
Entity Accuracy
< 0.01%
Server Overhead
2.1x
Index Speed
💡 Key Finding & Takeaway:

While not yet a verified Google ranking factor, llms.txt is heavily fetched by autonomous research agents and provides clean context without HTML parsing noise.