I think it’s a mistake to treat this like a conversion experiment. Trialers at the very first possible release of a service, heavily incentivized by a very lengthy free trial, are not similar to a steady state flow of trialers experiencing more transient promos or test experiences, which is a situation where conversion modeling makes sense.
This is not a funnel experiment. This is a huge pile of people specifically seeking out a special event at a launch, so the composition and intent of that is expected to be way, way, way more favorable than general steady state.
It’s like if concert tickets went on sale and everyone is camping in tents to buy them on the first day, and then when the booth opens and people listen to the free sample, 90% of them just pack up and go home.
That’s a much, much worse situation than if only 1/10 people who see an ad for the album choose to buy it on iTunes, whole different ballgame, funnel composition, intent model, everything.
10% conversion on a massive launch promo is drastic failure. This is not at all like getting 10% conversion on a funnel with low intent.
This is not a funnel experiment. This is a huge pile of people specifically seeking out a special event at a launch, so the composition and intent of that is expected to be way, way, way more favorable than general steady state.
It’s like if concert tickets went on sale and everyone is camping in tents to buy them on the first day, and then when the booth opens and people listen to the free sample, 90% of them just pack up and go home.
That’s a much, much worse situation than if only 1/10 people who see an ad for the album choose to buy it on iTunes, whole different ballgame, funnel composition, intent model, everything.
10% conversion on a massive launch promo is drastic failure. This is not at all like getting 10% conversion on a funnel with low intent.