I turned a recurring bill-shock problem into a proactive roaming guardrail system
This project started after a customer surfaced a $45,000 international roaming bill. But the real problem was bigger than one case. We had already received similar complaints from customers saying they did not realize charges were rising so quickly until the bill arrived.
That created a damaging cycle: low visibility into roaming costs → bill shock → fraud complaints → waivers and non-payment → bad debt risk. I helped redefine the experience so the system could intervene much earlier.
The final model introduced communications at every $100 accrual, auto-charge at $250, and suspension at $1000 if payment failed. I translated that strategy into journey mapping, communications planning, wireframes, and recovery states customers could actually understand.
Overview
International roaming is a classic high-stakes experience: customers often do not understand the cost until the damage is already done. That makes it a strong example of strategic UX work because the problem spans customer behavior, system rules, billing timing, communications, and trust.
- CX Analyst / UX Designer shaping the end-to-end experience
- Mapped the customer journey and intervention points
- Aligned communications to threshold-based system behavior
- Created wireframes and recovery states for implementation
- Customers needed earlier awareness and clearer action paths
- The business needed fewer waivers and lower bad debt risk
- Support needed a predictable intervention model instead of reactive escalation handling
The customer problem
Customers were not intentionally overspending. The system simply failed to make risk visible while charges were accumulating. By the time many customers understood what had happened, the bill was already large enough to trigger panic and distrust.
- “I didn’t know it was adding up this fast.”
- “I would have stopped if I had known sooner.”
- “These charges look fraudulent.”
- Weak in-the-moment visibility into rising charges
- No strong threshold-based intervention path
- No clear recovery flow once payment became urgent
- Fraud complaints and waiver requests
- Higher support cost
- Bad debt when customers refused or could not pay
The intervention model
I helped shape a progressive intervention model rather than a single late-stage warning. Each threshold had a different role in reducing uncertainty and limiting damage.
Repeated communication at every $100 gave customers multiple chances to understand that charges were rising and that action might be needed.
Instead of waiting for the full bill cycle, the system attempted payment much earlier. This reduced lag between charge accumulation and business intervention.
If payment did not succeed and charges kept growing, suspension at $1000 prevented another extreme runaway-cost scenario.
Each communication and screen needed to answer: what happened, what happens next, and what the customer can do right now.
Process
This work required connecting escalations, customer behavior, threshold logic, communications, and recovery experience into a single system that teams could align around.
Artifacts
These artifacts prove the work moved beyond abstract policy. They show how I turned system logic into a customer experience that is legible, actionable, and recoverable.
Edge cases I accounted for
High-risk billing experiences do not fail on the happy path. They fail when timing, payment, and customer understanding stop lining up.
- Customers continue using service while charges are growing
- Charge awareness still lags behind behavior, even after initial warnings
- Thresholds may be crossed quickly during heavy roaming usage
- Customers may only react once a payment event becomes real
- $250 auto-charge fails due to decline or insufficient funds
- The customer needs a clear route to update or complete payment
- Suspension at $1000 must feel understandable, not arbitrary
- Reinstatement must be obvious and low-friction once resolved
Impact
The real value of this work was shifting the roaming experience from reactive to proactive. Instead of waiting until the bill cycle exposed the problem, the system now intervenes while the customer still has a chance to understand and act.
- Customers were warned repeatedly instead of once
- The business collected earlier with the $250 checkpoint
- The $1000 suspension prevented further catastrophic growth after failed payment
- The UI made the consequence and next action clearer
- I used a real escalation to frame a broader opportunity
- I mapped uncertainty across policy, system behavior, and customer understanding
- I translated ambiguous business logic into a coherent customer experience
- I focused on decision-making, not just deliverables
Reflection
- One severe complaint gets attention, but repeated complaints reveal the system issue
- Customers need intervention points, not just information
- In high-stakes domains, trust depends on making invisible system behavior understandable
- Measure which $100 thresholds most effectively change behavior
- Test comprehension directly: “What happens next?” and “What should I do now?”
- Instrument the recovery funnel from failed payment to successful reinstatement