Reducing churn for a global subscription service
10 weeks on a churn problem, three concept directions, and a printer of our own to walk the customer journey end to end
- Who
- Ink Subscription Service (Anonymous)Mar 2025 – May 2025 (3mo)
- Team
- Lead strategistInternFounder Product Designer
- Scope
- User flowsJourney mappingFigma designPrototypingUser testing
- Tools
Miro
Figma
Context
An ink subscription service was losing customers, with a brief to surface concepts worth testing further
Where the friction sat between signup and cancellation was the open question. The brief asked for ideas worth testing, and we set ourselves a bigger question, what kind of evidence would let the client actually decide. A flat retention concept on a slide is easy to nod at and hard to act on, and the work needed to produce evidence with teeth, behavioural signal from real users that recommendations could be defended on.
The research
We interviewed customers, mapped the journey, and ran ourselves through it
The brief was retention, so we set out to understand what was actually happening between signing up and cancelling. I built the journey map in Miro from synthesis.
We also ordered a printer of our own, set it up, subscribed, ran through onboarding, and sent the first prints. The things a journey map alone won't tell you (how pricing gets revealed across multiple screens, where the product asserts value rather than showing it, what the cancel flow actually looks like) came out of being a customer for ourselves. Some friction only existed at the system level, across multiple touchpoints, where no single screen would have surfaced it in an interview.
The pattern that surfaced: the product worked, but people were leaving because they weren't sure it was working for them, especially the lighter users.
The concepts
Three directions, taken to different depths
We landed on the following ideas worth exploring:
- Badging and milestones to reward usage
- Surprise & Delight, an unexpected physical send to subscribers
- Personalisation, including a plan selection wizard, comms preference centre, and usage-based print summaries
The three directions weren't equally testable in 10 weeks. Badging had clear, prototypable mechanics that could be split-tested cleanly. Surprise & Delight was operationally dependent on a physical send, so the design work needed to be production-ready more than research-ready. Personalisation was the broadest in scope and would have needed its own dedicated round. So we took each to the depth that matched what would actually be useful next: badging to research, Surprise & Delight to logistics handoff, personalisation to a defensible concept.
I designed and built the Figma prototypes for each, plus the print collateral for the Surprise & Delight pieces. On badging, we split the test into three variants designed to isolate what was actually doing the work, whether it was badging itself or the reward layer beneath it. We took it the furthest, with 15 interviews across the US, Canada, UK, and Germany and three variants split evenly: badges only, badges with rewards, and badges with rewards plus a spin-the-wheel mechanic. The sharpest finding: badging only really worked when tied to tangible rewards, and without them, users found it pleasant but not a reason to stay.
We named in the synthesis that our sample skewed toward long-term subscribers with no intent to churn, so a follow-on round would need to reach lapsed subscribers, dissatisfied current users, and competitor users to test the at-risk question properly.
Outcome
Three directions handed off, with one of them shipping on the US side
Badging handed off with the research findings clear, Surprise & Delight went into logistics work and shipped on the US side as a leaflet with free photo paper, and personalisation stayed conceptual. I don't have visibility into what happened with the rest after handoff. The engagement's value sat in giving the client three distinct decisions to make on the same problem, each with the right depth of evidence to make it.
