How Crunchyroll grows

Sort cancelling users by what they ask for: ignore requests far from the product; for close requests (popular or under-the-radar shows Crunchyroll could reasonably license), go get the content. If acquired, email those users: 'Hey, this is now on the service. Do you want to come back?', sometimes with a discount or a trial period to return.

Done consistently in streaming: take users who cancel during the free trial (or first billing month) and users who make it through, and look for behaviors that differ between the two groups early on. Label risky vs valuable early behaviors and bend the product and onboarding (first day, first session or first month) toward the valuable ones. Pair with surveys: Google Forms or Zoom calls with long-tenured subscribers and fast cancellers.

Apply the increase to existing subscribers, since new-subscriber-only increases leave most of the gain on the table (host caveat: grandfather early supporters when the base is small). Keep increases marginal, e.g. $1-2 like Spotify and Netflix, never doubling. Model impact separately for existing subscribers and the future flow of new ones; track year-over-year growth in paid subscribers afterward. Expect a step down in acquisition (e.g. 100K/month to 80-90K after +10%) but little extra churn if executed well.

Primary metric was paid subscribers, with an assumption that price could rise gradually later. Crunchyroll deliberately did not raise price for a long time, trying to exhaust every way of adding new subscribers first (marketing, partnerships, more shows, side products for people who were cancelling). The first price increase came only after six or seven years of scaling paid subscribers.

Most paying subscribers were the most hardcore anime fans. Crunchyroll adjusted how it welcomed new anime fans: a branding and 'vibe' change that showed up in marketing, and required a cultural shift inside a company made up of diehard fans. Goal: reinforce the value core subscribers get while slowly expanding beyond the core audience.

Built a brand around the shows and the fan identity (the 'T-shirt test': would fans wear your brand's T-shirt). Introduced other revenue streams around subscribers: merch, events, gaming and more. Also tried products solving tangential needs for people who were cancelling.

Ran many pricing surveys, not only to set the core price but to estimate what the most passionate fans (top 1%) would pay relative to core subscribers, since they may pay more for a barely different product. Results treated with caution because stated and actual behavior differ.