Accorn

The Problem on the Ground

I didn't start with desk research. I went to rural communities across eight states and spent three months living alongside the people we were designing for. What I found was more nuanced than any brief could capture:

• No infrastructure

  No banks, no ATMs, no reliable internet. People traveled hours to access basic financial services.

• Trust was local, not institutional

  Villagers trusted the person who ran the local cooperative, not a bank logo on a screen. Past experiences with hidden fees and poor service had destroyed institutional trust.

• Literacy was a real barrier

  Many users couldn't read English or even their local language in written form. Icons, colors, and spatial patterns had to carry the entire interface.

• Informal economies were sophisticated

  The "unbanked" weren't financially inactive — they saved, lent, and transacted billions annually through informal cooperatives. They didn't need to learn banking. Banking needed to learn them.

The Business Case

Nigeria's informal economy represented billions in untapped financial services revenue. With 772 local government areas, even modest penetration meant massive scale.

• Transaction fees

  Revenue from deposits, withdrawals, bill payments, and transfers through agents.

• Microloans

  Credit services for rural entrepreneurs and farmers — interest income while driving economic activity.

• Government partnerships

  Welfare disbursement channels for government-to-person payments.

• Agent incentives

  Revenue-sharing model that made agents financially motivated to grow their customer base.

Projected first-year revenue: $5M from agents in 100 local government areas across eight states.

Research Methodology

Three months of immersive fieldwork. I didn't send surveys — I sat in village squares, watched transactions happen, and asked people to walk me through how they managed money.

Key Findings

The Agent Model

Instead of targeting end-users directly, I designed for community agents — trusted individuals with smartphones and basic digital skills. They would be the interface between the banking system and the community.

This wasn't a compromise. It was the architecture. Agents already handled money in these communities. I was giving them better tools, not replacing them.

What I Tried and Killed

My first design included a user-facing mode — a simplified screen that agents could hand to customers for self-service. In testing, it failed completely. Users were uncomfortable holding the agent's phone. They wanted the agent to do it while they watched.

I killed the customer-facing mode and redesigned the entire flow as agent-operated with visual confirmation screens that customers could see and verify. Trust increased. Errors dropped to near-zero.

User Stories & Journey

Feature Prioritization

Every feature was evaluated against one question: does this work when the internet drops? Offline-first was the architecture, not a feature. I cut real-time notifications, location tracking, and video onboarding — all of which required connectivity.

User Flow

Sketches & Wireframes

Pen-and-paper first, tested with actual agents in the field. Two rounds of iteration before any screen hit Figma.

Final Designs

Three taps for any core action. Color-coded transaction types. Large touch targets. Offline-first with clear sync indicators. Commission dashboard that agents checked hourly — it was their primary motivator.

Validation

Tested with agents and rural users across three states. The results validated the agent-centric model:

• Simplicity

  Intuitive for agents with low digital literacy. Zero training needed for basic flows.

• Offline mode

  The most praised feature. Service continued uninterrupted in low-connectivity areas.

• Commission transparency

  Agents loved seeing their earnings in real-time. It drove adoption faster than any marketing.

• What I fixed post-testing

  Error messages were too generic. I replaced them with specific, action-oriented guidance. Sync indicators were added after users reported anxiety about pending transactions.

Reflection

This project changed how I think about design. The most important decision wasn't any screen or flow — it was choosing to design for agents instead of end-users. That single architectural choice made everything else possible.

What I'd do differently: I'd push for local language support from day one. We launched in English only, and while agents could translate verbally, the app should have spoken the community's language from the start.