I run a handful of small products. For two of them, the weekly report said the same thing every week: 0 visitors. I treated that as a fact and started deciding what to cut.
Yesterday I finally read the served HTML instead of the dashboard. Neither product had the analytics script on the page. They had a hosting-level analytics package I'd set up early on and forgotten, and nothing at all reporting into the dashboard I actually read every morning. The zero wasn't a measurement. It was the absence of one.
Then it got worse. One of the two had a production site sitting two days behind main. Every push built. Every push was also silently rejected at the last step: the host blocks a deployment when the commit author's email isn't linked to a GitHub account. That repo was the one I'd cloned into a different folder and never ran "git config user.email" in, so my commits went out as me@MyMacBook-Air.local. Pushes succeeded. GitHub looked completely normal. The domain just kept serving the last deployment that had passed. Two days of fixes were sitting in deployments marked BLOCKED that I never looked at, including the fix for an og:image that had been 404ing on every share.
The CLI didn't help: it prints BLOCKED as "UNKNOWN" with no duration. I only found it by querying the deployments API directly and reading readyState and errorMessage.
Both failures have the same shape. I was reading an instrument and calling it the thing.
What I changed:
If a product of yours shows a flat zero, open view-source before you decide anything about the product. It costs thirty seconds and I just spent two weeks not doing it.