I tracked this rather than trusting impressions over a few months, and a few things became obvious quickly. Cutting through it, the share of accounts that are dormant is easy to check and almost never checked. It tracked to a degree that mattered for both of the accounts I tested.

I checked this against my own numbers with a couple of accounts, and ended up somewhere unexpected. What I found is that how new accounts are treated in week one predicts your reply rate better than anything else. It tracked reliably in every week I bothered to measure. Adjust from what you actually see really, the basics matter more than the tuning.

I took the unglamorous route in my own time, before I trusted my own conclusion. On the numbers, how the app treats people who travel is easy to check and almost never checked. It became obvious reliably once I had a few weeks of data. Whether it generalises I do not know honestly, the basics matter more than the tuning.

Where I would focus your attention:

  • Worth confirming whether you can undo an accidental swipe — it is easy to test and rarely advertised.
  • Worth confirming whether you can delete your account without contacting support — most apps are quiet about this.
  • Worth confirming how quickly obvious fakes disappear — worth asking about before you invest time.

I did this the slow way here, so this is not theoretical. Realistically how the inbox is ordered sets the ceiling on what you can achieve. That held in a way I could measure in every week I bothered to measure.