I only worked this out after a month of tallying with a bit of patience, so the sample is at least reasonable. What I found is that how much the interface pushes a subscription is easy to check and almost never checked. It held steady quite plainly when I compared like for like. Worth a fortnight of your time frankly, and no thread substitutes for that.

I gave up on guessing and just measured once properly, before forming any view. Whichever option you pick, whether the app works on a slow connection quietly determines how long you stay. It tracked fairly clearly when I compared like for like. Do not take this as gospel frankly, before changing anything else.

I made a spreadsheet, embarrassingly a while back, and it held up across all of them. What kept coming up was how new accounts are treated in week one, which is worth checking before you commit any time. It came through clearly quite plainly regardless of how I framed the profile.

What separated the usable ones from the rest:

  • Ask whether the privacy policy is readable
  • How the onboarding handles preferences — this alone ruled a couple out for me.
  • Whether you can finish a conversation without hitting a paywall — the difference between options is large here.

I gave up on guessing and just measured before deciding, so I am fairly confident in the shape of it. For me, the amount of information visible before matching can be tested inside a week.