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:
- Do the important routes return
200? - Are canonical URLs and sitemap entries consistent?
- Did the production alias move to the intended deployment?
- Does the browser console show real errors?
- Does mobile have horizontal overflow?
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.
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.