The client site that changed without anyone deploying

Got a panicked message from a client last month. Their checkout page was showing wrong prices. Nobody on the team had pushed any code in three days.

Turned out their headless CMS had a scheduled content update that a marketing intern had set up two weeks prior and forgotten about. Changed the pricing tier labels and the numbers underneath. The intern had left the company by that point.

It took us about two hours to figure out what happened. The git log was clean. The deploy history was clean. The CMS audit log showed the change but nobody thought to check there first because "nobody deployed anything."

The gap in most monitoring

We had uptime monitoring, error tracking, performance monitoring. All green. Because technically the site was up, throwing no errors, and loading fast. It was just showing wrong prices.

Content changes are a blind spot for most monitoring setups. You check if the server responds, if the JS throws errors, if the page loads in under 3 seconds. You dont check if the actual content on the page is what it should be.

What we added after this

Scheduled visual snapshots of critical pages. The checkout page, the pricing page, the homepage hero section, the legal pages. If any of these change visually between snapshots, we get a notification with a before/after comparison.

Its not perfect -- legitimate design changes also trigger alerts. But tuning out the noise is way easier than missing a pricing error thats been live for 18 hours.

We check about 30 pages across 4 client sites now. Most days, zero alerts. When something does fire, its been worth investigating every single time.

The irony is this whole category of monitoring is dead simple to set up. Screenshot the page, compare to previous, alert on difference. The tooling exists. Most teams just dont think about it until something goes wrong.