Lifting notification setup engagement from 2% to 40%.

JPMorgan Chase / Chase notification settings

Role
Lead Designer
Timeline
May – August 2021
Team
Product, platform engineering, research
Outcome
First-time setup engagement 2% → 40%
Chase notification settings cover
01

The challenge

Chase sends important messages. Customers rely on alerts for account activity, security warnings, and transaction confirmations. But the experience trained customers to ignore us.

Research revealed customers were overwhelmed and disengaged:

Finding 01

Control was buried

Settings were ~5 steps deep, and many customers didn't know where control lived.

Finding 02

Dense and unclear

Too much information and too many options, organized by internal product names.

Finding 03

Control out of reach

Customers wanted a sense of control but couldn't find it.

What leadership cared about: this risked trust erosion, increased servicing cost, and made customers more likely to miss or distrust fraud and security alerts.

02

The pivotal insight

The problem wasn't “too many alerts.” It was “control is hidden until frustration.”

Customers weren't failing to manage preferences because they didn't care. They failed because the app asked them to do hard configuration work at the worst possible time, after they were already frustrated. So instead of improving a giant toggle page, I changed the model: make preference selection easy upfront, and make ongoing management feel familiar and trustworthy.

03

The design shift

I replaced configuration overload with three clear paths and an inbox mental model.

Shift 01

Three paths, one decision

An overwhelming settings task became a single confident choice, with room to refine later.

Shift 02

Inbox mental model

Notifications and secure messages behave like an inbox, so customers recognize what a message is and feel in control.

Shift 03

Trust-critical alerts protected

Fraud, security, and non-user-initiated activity are treated differently from routine alerts. Less noise, same high-signal trust.

04

Two scenarios

The flows we designed and tested.

Scenario 1 · Simple setup
  • Opens a new Chase account → goes through one-time onboarding
  • Customizes alert preferences → sets “more control” over preferences
  • Makes changes to the defaults → keeps editing as needed
Scenario 2 · Too many alerts
  • Opens a new credit card → receives too many alerts
  • Chooses to update alert settings → makes updates by category
  • Quickly toggles off unwanted alerts → sees confirmation of the changes
05

What we shipped

The full native experience required 18+ months. Rather than wait, I got buy-in for a web-first MVP, mobile web within the app, that proved value fast while earning the right to larger platform investment.

Release · Q1 2021

  1. Kept the existing side navigation, adding menus under Alerts so users can view and manage notifications easily
  2. List of alerts and notifications in a vertical table view

Release · Q2 2021

  1. Distinct visual treatments by transaction type: credit card, mortgage, account transfer
  2. Visual indicator to open transaction details
The alerts management page in mobile browser view, also available in the app, for the Q1 and Q2 releases.

This approach let us make significant strides within the limits of our resources and technology, while working toward the long-term vision.

06

Impact

The web-first MVP proved the concept, and earned investment for the broader vision.

2% → 40%
First-time setup engagement
15–20%
Reduction in notification-related complaints
2022 vision
Defined for unified communication across platform teams
07

What I learned

The jump from 2% to 40% didn't come from better toggles. It came from questioning whether notification management should be reactive and buried at all.

The pattern: reduce cognitive load while protecting trust-critical communication, then ship a proof-oriented MVP that earns the right to bigger platform investment.