Blog

Stories from the field

Notes from the team — what we learn building AI monetization for Unity and Unreal studios.

• Aug 5, 2026 · 4 min read

  When a winning test result should not be shipped at all

  A variant can win its measurement and still be the wrong thing to ship. The metric it won on is rarely the only thing the change affected.

• Aug 5, 2026 · 4 min read

  Control groups without giving up the revenue they cost

  Holding players back from a change is the only way to know it worked. The objection is that it costs money, and that objection is mostly answerable.

• Aug 4, 2026 · 4 min read

  Testing changes when your player base is still small

  Standard testing advice assumes traffic you may not have. With a few thousand players, most differences you can detect are not the ones worth acting on.