Technical SEO automation: what we automate, what stays manual
Crawl reports, schema deployment, Lighthouse delta tracking — automated. Strategy, prioritization, judgment — manual. The boundary matters.
Technical SEO has a clean automation surface. Crawl reports run on a schedule. Schema validates against a spec. Lighthouse produces a number you can diff. Internal link suggestions can be surfaced as PR comments instead of a deck.
What doesn't automate well: deciding which of the 400 issues the crawl found actually moves traffic. That's judgment, and judgment doesn't fit on a cron job.
What we automate
Daily crawl reports posted to the team channel, with deltas vs. last run. Schema deployments via PR, validated against schema.org before merge. Lighthouse runs on every deploy, with thresholds that block merge when performance regresses. Internal link suggestions surfaced as draft PRs on the content repo.
What stays manual
Quarterly priority calls. Which two of the twelve fixable issues are this quarter's focus? That's a meeting, not a function call. Anyone selling you 'AI-powered SEO strategy' is automating the wrong layer.
Client communication around tradeoffs. When canonical tags change indexing behavior, somebody has to explain it to the marketing director. Putting an LLM in that loop turns a five-minute conversation into a two-week support thread.
The system runs the system
The point of automation isn't to replace judgment. It's to make sure judgment happens with full information, not selective sampling. The team showing up to the quarterly call has every crawl report from the last quarter, not a memory of what stood out.