From measuring delivery to  

building financial habits

A digital banking app with millions of downloads but no visibility into what customers did next. The solution was a product strategy, a measurement framework, and built-in AI integrations to close that gap.

Downloadable on iOS and Android

____________________________________________________________________________________________________________________________________________________________

____________________________________________________________________________________________________________________________________________________________

CONTEXT

Arab Bank launched Reflect as Jordan's first fully digital neobank. Commission-free, mobile-first, and built on MACH architecture (Microservices, API-first, Cloud Native, Headless). The product suite covered the full banking stack: wallet accounts, virtual and physical debit and credit cards, contactless payments, bill pay, savings accounts, BNPL, a marketplace, and a loyalty and rewards engine that drove daily engagement.

The strategic bet behind Reflect was about who it was built for. Arab Bank already had a strong regional foothold with wealth management clients and a dedicated youth product. Reflect was aimed at a different segment entirely; the mass consumer. People who were largely cash-dependent, working in trades and vocations, with irregular or informal incomes. People who didn't fully trust large financial institutions with their data, and who often couldn't access the main branches concentrated in capital cities.

For this segment, access was the core problem. A digital bank on their phone, requiring minimal documentation and built around a simple onboarding experience, wasn't just a better product, it was the only product that worked for how they actually lived. Getting Reflect right for this user wasn't optional. It was the entire premise.

____________________________________________________________________________________________________________________________________________________________

THE CHALLENGE

Everyone was measuring delivery.

Nobody was measuring habit.

When I joined, the team was building fast. Features were shipping on schedule. Downloads were climbing. Wallets were being created. On the surface, it looked like progress.

But when you looked at actual customer behavior, what users did after they downloaded the app, the picture changed. Marketing was optimizing for installs. Product was optimizing for releases. Leadership was optimizing for revenue. Each team was winning its own metric, and none of those metrics said anything about whether customers were building a banking habit.

The product was also untethered from any real understanding of who it was for. Features were built on assumptions rather than evidence. And with expansion plans accelerating, the cost of getting this wrong was multiplying.

____________________________________________________________________________________________________________________________________________________________

THE REFRAME

A strategy problem with design

implications downstream

The main reframe was structural: we could not fix the product experience until we knew who we were building for, and we could not know if anything was working until we had a shared definition of success. This sequence mattered as much as the work to be done.

____________________________________________________________________________________________________________________________________________________________

THE APPROACH

A four-pillar framework

The problem was structural, so the solution had to be too. I designed a strategic framework with four interlocking pillars. The order mattered because you can't redesign a customer experience before you know who the customer is, and you can't align teams around growth metrics before those teams have a shared vocabulary. Every pillar was intentional and sequential.

____________________________________________________________________________________________________________________________________________________________

DESIGN ARTIFACT 01

User archetype matrix

Four archetypes mapped along two behavioral axes - Financial Intent and Digital Confidence. Built from behavioral and psychographic signals, not demographics alone. Became the shared language for roadmap prioritization and persona-matched onboarding.

____________________________________________________________________________________________________________________________________________________________

DESIGN ARTIFACT 02

One of the first things I did was map the actual customer journey; not the intended one, but the one customers were experiencing. What I found was a series of broken handoffs. The onboarding flow asked for effort before it demonstrated value. The chatbot escalated instead of resolving. And the transition from "new user" to "active user" was invisible; there was no designed moment where a customer felt like they had actually arrived.

The journey map became the organizing artifact for the redesign. Every friction point on the map became a brief. Every emotional low became a design opportunity.

____________________________________________________________________________________________________________________________________________________________

ONBOARDING REDESIGN

with a persona-aware one

The old onboarding flow presented every feature at once. There was no differentiation between a new user who wanted to save money and one who wanted to pay bills. The home screen was built for the product team, not the customer. It showcased everything the app could do rather than guiding the user to the thing they should do first.

The redesign introduced a personalization gate early in the onboarding flow. A simple question that routed users to a persona-matched first experience. The question also seeded our behavioral intelligence layer, giving the data team a first signal to work with from day one.

____________________________________________________________________________________________________________________________________________________________

BEFORE AND AFTER

to intent-driven experience

The before state of the home screen was built around what the product offered. Every feature was surfaced at equal visual weight, leaving users to figure out what to do first. There was no personalization, no designed progression, and no moment of felt value.

The redesigned experience was built around what the user came to do. The persona selection at onboarding directly informed the home screen hierarchy. Surfacing the actions most likely to match intent and routing everything else into a secondary layer.

____________________________________________________________________________________________________________________________________________________________

AI SUPPORT LAYER

Designing for resolution,

not escalation

The support experience was a symptom of the same structural problem: no system had been designed to understand what a customer actually needed before routing them. Every query, regardless of complexity, defaulted to a human agent queue; creating backlogs, slow response times, and a support load that scaled linearly with user growth. A model that would not survive expansion.

I led two parallel AI initiatives to fix this. The first was a WhatsApp-based triage chatbot built on Infobip; designed to classify intent before responding, routing routine queries to instant automated resolution and complex cases to human agents with the full conversation context pre-packaged. WhatsApp was the right channel for these customers because it was the messaging platform they were already in. The second was the integration of an in-app AI assistant, bringing transactional capability and personalized guidance directly into the product. Unlike a traditional chatbot, this AI assistant understood Arabic in colloquial form, not just formal language, through voice or text. Users could ask it to execute transfers, settle bills, or check account details conversationally.

Together, these two layers formed a complete AI support stack: Infobip handling reactive triage on WhatsApp and the AI-assistant enabling proactive, conversational banking inside the product. The design rationale across both was: a system that tries to handle everything fails at everything. Knowing scope, and handing off cleanly when something exceeds it, builds trust even in escalation.

____________________________________________________________________________________________________________________________________________________________

ECOSYSTEM AND ALIGNMENT

Building the intelligence layer

and the shared language

The work was not just about the product. A redesigned app would not hold if the rest of the organization was still optimizing for different things. I ran co-creation workshops with Marketing, Data Science, Operations, and Customer Experience to align around a single shared growth metric - "the North Star" - and establish common definitions for activation, retention, and success.

At the same time, we built out the behavioral intelligence infrastructure: funnel analytics tracking drop-off at every step of the onboarding and activation flow, cohort segmentation mapping user behavior to archetypes over time, and social signal monitoring that surfaced feedback patterns before they became volume issues.

I also advocated strongly for a design system; not only as a design team deliverable, but as organizational infrastructure. A shared component library meant faster builds, fewer production incidents from UI inconsistencies, and a common visual language that aligned design, development, and QA around the same source of truth.

____________________________________________________________________________________________________________________________________________________________

IMPACT