Do you look at actual usage data instead of listening to feature requests?

It's so easy to mistake a loud minority for market demand. People will request integrations or dashboards but analytics almost always show they stick to just a couple of core tools anyway.

When we stopped building requested features and started focusing purely on fixing drop offs in our main workflow, our metrics improved a lot.

How do you measure whether a requested feature is genuinely going to help everyone or just satisfy one user?