Automating the weekly report, end to end

Automating a weekly report means making it exist without anyone's Monday discipline: freddyxai's own metrics report now regenerates on a GitHub Actions cron every Monday at 07:00 UTC, in 16 seconds, for $0.

What does the report actually contain?

Live site stats read from the production sitemap — no self-reported numbers — plus the canonical 12-query GEO tracker: which buyer questions freddyxai targets, whether each target page is live, and the current citation count (0 of 12, re-baselined and published either way).

The receipt we refused to inflate

The standard automation pitch is "saves hours every week." We tried to measure that honestly: stepping through the assembly without the script took 48 seconds — but an AI agent did the stepping, at machine speed. That bounds the mechanical floor, not human minutes, so this page doesn't claim time saved. The receipt that survives scrutiny is different: the report now exists every single Monday, whether anyone shows up or not.

What makes it run itself?

Zero-dependency Python plus a Monday cron: generate, commit, push. The first scheduled run went green in 16 seconds, and every future report lands in the repo with a timestamp.

Key takeaways

• Scheduled or it doesn't count — a report habit beats a report heroic.
• Publish the receipt that survives scrutiny — we dropped "hours saved" when the measurement couldn't support it.
• 12 queries, every Monday, 16 s, $0/mo — the tracker is now freddyxai's canonical citation source.

Keep reading: the GEO baseline this tracker serves and the full case study.