Analyzing 8K+ Pages for $180: A Practical LLM Data Workflow

gaganghotra_ · x · 2026-07-21

A developer shared a hands-on case of using LLMs (Claude 3.5 Sonnet and Opus) to analyze data for a large client e-commerce site. - **Context**: The site is massive, featuring nearly 8,000 product pages, 183 collection pages, and over 4,000 blog posts. - **Workflow**: The developer fed 3 years of Google Search Console, GA4 (via BigQuery), and Shopify data into the models to analyze and identify pages that should be purged or merged. - **Cost & ROI**: Total token spend over a couple of weeks was around $180. It saved massive time compared to manual analysis, and the expense can be claimed as R&D and direct client deliverables for tax returns, making the net cost effectively zero.

Original post →

More from coding & agent

coding & agent channel →