US fintech appSubscription flows

Monet Cancellation

Building a split offboarding funnel to ensure users can get the most before they decide to cancel.

The redesigned cancellation flow in four screens: Don't lose the tools improving your credit, a Before you go sheet, What's making you cancel, and Review canceling Premium.
Role
Product Designer, Project Management & Handoff
Timeline
9 months across phases
Tools
Figma
Status
In progress
$5New Starter tier
2Flow variants designed
8User states mapped
9 moAcross phases

Overview

Giving user more options than before covering all open ends of the current membership.

Problem

Bright ran two tiers: Premium or nothing. A user who cancelled went to $0, with no middle.

My part

Product design, project management and handoff for the cancellation flow, from the offer timing to the seven account states behind it.

Where it landed

The Premium cancellation flow is closed and signed off. The Starter flow had feedback implemented and a final review pending.

Context

Monet is where every subscription decision happens.

Monet is Bright's monetization surface. It covers every screen a user touches when they subscribe, manage their plan or cancel. I was part of this project across its phases, taking the cancellation flow through design, project management and handoff.

Problem

Cancelling meant losing the user completely.

Bright had Premium or nothing. The new Starter tier was a way to stay in the product at a price a leaving user might accept.

  • Premium · $14 / moFull subscription. All features unlocked.
  • Starter (new) · $5 / moCash advance, credit features, and Rent Reporting as an add-on. A way to stay in the product.
  • Basic · $0Cancelled. Limited access.

Two user groups needed designing for: existing users migrated to Starter automatically with a 30-day grace period, and Premium users actively cancelling who needed to see the $5 offer.

The older 2.0 cancellation flow: a reasons survey, a Don't stop your progress screen, and a 70% off special offer.

Older 2.0: the flow did not use the feedback from the user survey to retain users who were leaving.

The older 2.0 UI: Don't stop your progress now, and two Use your full 12 months membership screens with an hourglass and a clock.

Older 2.0 UI.

Redesign

The 3.0 flow shows what leaving costs.

The 2.0 flow asked why the user wanted to leave, then offered a discount, without using the survey answers. The 3.0 flow opens on the premium tools the user is about to lose, adds a confirmation sheet, asks what is making them cancel, and ends on a review of the benefits and the date access ends.

The improved 3.0 UI: Don't lose the tools improving your credit, a Before you go sheet with Keep my Progress and Continue to Cancel, and a Before you cancel, please note screen.

Improved 3.0 UI.

The 3.0 cancellation flow in four screens: the tools you will lose, a Before you go sheet, What is making you cancel, and Review canceling Premium with the plan end date.

The 3.0 flow, end to end.

Decision

When to show the offer.

Netflix shows you a cheaper plan before you cancel. Audible does the same. The mid-flow downsell is a proven pattern, but it only works if the offer is genuinely strong.

I designed multiple flow variants. Variant 1 had stronger visuals. Variant 2 put the $5 offer at the right moment in the flow. We aligned on Variant 2 for the first test, not because it looked better on screen, but because it gave users the information at the moment it was actually useful.

First testMid-flow downsellShow the $5 Starter offer before the user confirms cancellation. Right information at the right moment.
FallbackPost-cancellation offerConfirm the cancel, then pitch Starter. Designed and ready. Some users respond better when the product honours their decision first.
Two phone screens of the Starter offer: a bottom sheet reading Enjoy Starter free for 30 days, listing four Starter benefits with Continue and Save and Lose all benefits buttons, beside the post-cancellation screen, Your Premium membership is now canceled, which keeps the benefits with Bright Starter at $5 a month.

The Starter offer. Left: the offer sheet, Enjoy Starter free for 30 days, with Continue and Save. Right: the post-cancellation screen, which keeps Starter at $5/mo.

Constraints

The compliance layer.

Every screen in Monet touches compliance, and cancellation screens even more so. These were not copy notes. Getting them wrong on a cancellation screen is a legal exposure.

Flag 1"Free" on a final screenThe word appeared where Starter costs $5 a month. It came out.
Flag 2"You are eligible" above a loan offerIt implied an approval the user did not have. Rewritten to match the compliance-reviewed copy on the Bright website exactly.

Complexity

The hardest part was the number of states, not the flow.

A single user could be in seven distinct states, each with different copy, different calls to action and different settings screens. I mapped every one and designed them all.

  • MigratedIn the notice period.
  • MigratedIn the grace period.
  • MigratedOnce billing started.
  • PremiumCancelled.
  • StarterForce opt-in.
  • StarterExplicit opt-in.
  • BasicLimited access.
Dashboard 2.0 on the left, annotated with no delight and repeating information, beside the new Premium membership dashboard on the right.

Dashboard 2.0 was confusing and inconvenient to understand and act on. Annotated: no delight, repeating information.

The Bright Membership dashboard in three states: Premium, Starter with a 30 days free banner, and Starter.

All membership dashboard states: Premium, Starter with 30 days free, Starter.

The Starter dashboard in the middle, with the Premium upsell card on the left and the Starter benefits list on the right.

Starter dashboard, with the Premium upsell and the Starter benefits.

Two close views: the Starter dashboard, and the Get Premium for $14 upsell.

Starter dashboard and the Premium upsell, up close.

Where it landed

One flow signed off, one in final review.

The Premium cancellation flow is closed and signed off. The Starter cancellation flow had feedback implemented, with final review pending at handoff. The full settings states and wireframe set were delivered and are with engineering.

No results are reported here yet, because the flow had not been through a live test at handoff.

Takeaways

What I would do differently.

  1. Map all user states before touching the design. Seven states across three user types is a known complexity for a tiered subscription product. I found the full scope through iteration instead of up front.
  2. Get data before locking the offer timing. The mid-flow versus post-cancellation decision needed data I did not have. I would push for prior A/B test results from a comparable product first. We chose right based on reasoning, but data would have made it faster.