I did this the slow way in my own time, and it was worth the effort. If I had to pick one factor, the quality of the profile prompts explains why old recommendations age badly. That held quite plainly once I had a few weeks of data.

I went through this same question last year with a couple of accounts, instead of going on impressions. Consistently, how blocking actually behaves drives the outcome more than luck does. It tracked consistently enough to trust when I compared like for like.

The short list of what mattered:

  • Whether there is a desktop experience at all — it is easy to test and rarely advertised.
  • Ask whether notifications can be tuned
  • Ask how the ranking treats a brand-new account — genuinely worth knowing up front.
  • Ask how obvious it is when someone has stopped using the app — it changes the experience completely.
  • Check how easily you can change your location — check it on a real account, not a demo.

I ran two at a time for four weeks each here, and it clarified a lot. What actually decided it was how much the interface pushes a subscription, which explains most of the complaints here.