Spark migration campaign
PRODUCT DESIGN
GROWTH DESIGN
MIGRATION
ABOUT THE PROJECT
Spark 3 shipped in 2022, and the large population still using Spark 2 was never really told why they should follow. Some of them tried the new version on their own, hit missing features and rearranged settings, and went back. That left a harder problem than an ordinary migration: not persuading users to move to something new, but persuading users who had already formed an opinion to reconsider it. The task was to move a very large base of users from a free legacy product to a freemium one, and to do it in a way that converted rather than churned.
RESEARCH & DIRECTION
Trust was the actual constraint. Anyone who had tried Spark 3 once and bounced wasn't going to be moved by a louder announcement. I went through the existing survey feedback and reduced it to three problems worth designing against: - Unclear value. Users couldn't say what the new Spark did better, only what it did differently. - Disruption and time loss. Settings and workflows had to be rebuilt by hand, and nobody wants to spend an afternoon rebuilding something that already worked. - Learning effort. The interface had changed enough that familiarity counted for nothing. Each of those is a reason to postpone. Postponing indefinitely is what churn looks like from the inside.
DESIGN PROCESS
Say what's better, repeatedly and in the right places. The value of Spark 3 was made explicit across in-app modals, email, and articles, rather than assumed. Remove the manual work. A Migration Assistant moved workflows across, so the cost of switching stopped being an afternoon. Carry their settings over. Arriving in Spark 3 with your own UI already in place turns a re-learning problem into a gradual discovery one.
TESTING THE FLOW
Before designing screens, I mapped the whole migration as a single user flow — every entry point, every state a user could arrive in, every place they could stall. Seeing it end to end is what surfaces the blockers early, while they're still cheap to remove. I built the scenarios in Lovable and walked stakeholders through them as working prototypes rather than static flows. Disagreements about what should happen at a given step got settled by clicking through it.
OUTCOME
The first communication wave beat its 30-day migration target by a wide margin, and the campaign is still running. What the numbers confirmed: the blocker was never persuasion. Once the manual work of moving settings and workflows was removed, people moved. The Migration Assistant did more for the result than any of the messaging did. Specific figures are covered by my agreement with Readdle. Happy to talk through them in an interview.