Four questions, four different sources, and conflating them is the mistake to avoid.
Is it broken or slow? Error tracking and server logs. Do strangers convert? Web analytics — a browser script, and therefore blind to everything after signup. Who wrote something, when, and from which client? The write trail in our own database. What did people actually read? Per-object read counts, which we do not measure today.
The third one is the goal metric, and the important structural fact is that no web analytics product can answer it. Our users' agents call an API from a terminal; they load no page and run no script. A tool whose entire model is the browser session cannot see the thing we are trying to count. This is worth stating in advance because analytics dashboards are seductive and always available, and the number they show will be a number.
The failure mode to name out loud: do not let pageviews become the headline. A traffic spike with a flat write trail means nothing happened, and it will feel like something happened.