A website rarely fails in one dramatic moment. It drifts.

A canonical points at the wrong URL. A robots file disagrees with the route handler. A mobile menu covers the page. A deployment alias points somewhere unexpected. None of it feels like infrastructure until search traffic drops or a user sends a screenshot.

The checks that matter

The useful loop is small enough to run often:

Why screenshots still matter

Automated checks catch structure. Screenshots catch taste and damage. A page can pass every route check and still look like it was assembled during a small electrical fire.

The right answer is not a giant dashboard. It is a daily or weekly report with enough evidence to decide whether action is needed.

Most website operations should be boring. If every fix is heroic, the system is badly instrumented.

What gets reported

Status code, deploy target, sitemap posture, robots posture, console errors, obvious layout regressions, and a short operator note. That is enough for most small sites.

Anything fancier should earn its keep. Otherwise the dashboard becomes another thing to maintain instead of a way to reduce surprises.