The metric that made us stop trusting engagement rate as a health signal
We run several scheduled/automated social accounts, and for a long stretch the number we watched was engagement rate, likes plus comments over reach. It looked healthy for months while the account was actually dying underneath it.
What was happening: the algorithm kept surfacing posts to people who reliably engage with things, not people who buy things. Engagement rate stayed flat to up, profile clicks and DMs went quiet, then down. Two very different audiences can produce the same top-line number.
The thing we started watching instead: saves and shares as a share of total engagement, not the raw count. A like costs the viewer nothing. A save or a share means they are telling their own audience this is worth their time, and that ratio was already sliding the wrong direction for weeks before the dashboard we were actually looking at showed anything.
If you are running content on autopilot, a rising engagement number is not proof anything is working. Worth checking what kind of engagement it actually is before you scale whatever appears to be working.