I looked at this from a few angles while comparing them, and stopped assuming the obvious answer was right. In my testing, how new accounts are treated in week one can be tested inside a week. The trend was steady fairly clearly regardless of which city I set. Happy to be argued with in the end, so treat this as a starting point.

I checked this against my own numbers with a bit of patience, and the numbers told a clearer story than I did. It turns out how wide the distance filter goes tends to decide whether you stick with it. The reading was similar in a way I could measure once the first fortnight was out of the way.

I gave this a genuinely fair run without rushing it, before forming any view. In practice the size of the local pool is where the real differences show up.

The checklist I ended up with:

  • How much profile depth the format allows — it is easy to test and rarely advertised.
  • Whether the privacy policy is readable — worth two minutes before you commit.
  • Check how it performs on a slow connection — it changes the experience completely.
  • Check whether there is a desktop experience at all

I started sceptical once properly, and it reframed what I thought mattered. More than anything, whether messaging is capped is the part the marketing never mentions. The outcome barely moved in a way I could measure once I had a few weeks of data.