What result should you optimise for when you’re testing creative?

I run couple of creative test for better ROI:

1. IPM test in mobile app install campaign on FB with only FB newsfeed placement in 1 geo to gain ~ 100 installs

2. IPM Winners move to ROAS test in VO campaign with all placements in 1 geo.

3. ROAS winner move to the main VO campaign to compete with old best perfroming creatives

In my experience, it's ideal to have your optimization as close to the desired business results as possible. It sounds like in your case that's account creation of first game play. I consider that my default. That said, I've had two kinds of situations pull me away from the default.

The first is scale. If you're doing an ongoing campaign the default is fine, but short-burst campaigns may need a higher-funnel conversion like an install.

The second is incrementality. I've had situations where optimizing for the business result directly was cannibalizing organic conversions. But when we optimized to a discovery-related event, it wasn't. Examples of discovery related events might be searches in the site/app, or reads of deep informational pages.

So, default to what's closest to your business result unless scale or incrementality suggest otherwise.

Most teams just optimize for installs since it can be hard to get reliable ROAS data in a test setting. CTR seems too up-funnel to me; I'm not sure you'd get meaningful indicators of success in a live setting by just testing for CTR.