Static sites are usually described as brochureware, but the AKREtrack build treats static delivery as the final stage of a small ETL pipeline.
The moving parts are deliberately boring: a scheduled Worker fetches a daily CSV, stores the raw file in R2, writes structured rows to D1, compares the latest snapshot with the previous one, and triggers a Cloudflare Pages rebuild when there is useful new data to publish.
The useful trick is keeping each stage accountable. Raw source files stay available for audit, parsed holdings are queryable, diff output is generated from stored state, and the frontend remains a fast static Astro site.
That shape fits small public data products well. The runtime work happens at ingestion time, while readers get a cached page with no application server in the request path.