Al-Hidaya Research Lab
Empirical Research & Whitepapers
Original technical research, benchmark studies on AI LLM latency, hyper-local conversion algorithms, and web performance metrics.
Hyper-Local CommercePublished: 2026-08-01
Benchmarking Hyper-Local Search Signals & Conversions in Delhi-NCR (2026)
An empirical analysis of 120+ local SMB digital profiles across South Delhi, Noida, and Gurgaon evaluating proximity factors, WhatsApp click-through rates, and Google Business Profile engagement metrics.
Key Findings:
- Listings with direct WhatsApp catalog buttons achieved 3.2x higher conversion rates than standard contact forms.
- Page speed scores under 1.2s correlated with a 41% decrease in bounce rates on mobile networks.
- Localized schema markup (LocalBusiness JSON-LD) improved Google 3-Pack visibility by 68% within 30 days.
AI BenchmarksPublished: 2026-06-18
Evaluating Low-Latency LLM Agents for Enterprise WhatsApp Customer Support
Comparative study on fine-tuned Llama-3 8B vs GPT-4o-mini latency, intent accuracy, and token optimization when handling multi-lingual customer inquiries in English and Hindi.
Key Findings:
- Fine-tuned 8B parameter models running on edge GPUs matched GPT-4o accuracy for domain-specific appointment booking while cutting latency by 64%.
- Hinglish (Hindi written in Roman script) prompt templates reduced intent classification errors from 18% down to 2.1%.