All case studies

Field Research · Personas · Journey Mapping · Usability

Mastercard Farmers Network — research that changed the product roadmap

ROLE — Senior Product Researcher
TEAM — Mastercard Labs × Digital Disruptions
TIMELINE — Apr 2019 – May 2020
MARKETS — Tanzania · India · Kenya

MFN: Mastercard Farmers Network

Agricultural payments platform · Tanzania & India deployments

Fieldwork interviews conducted
27+
agents · aggregators · buyers · farmers across 2 continents
Research phases
2
14-week programme
Problem statements
3
surfaced for ideation
Key personas built
3
agent archetypes
User adoption rate
<20%
pre-research baseline

The challenge

Mastercard Labs had been piloting MFN — a platform connecting smallholder farmers, field agents, and agricultural buyers — for nearly two years across Tanzania and India. Despite strong institutional backing (including the Bill & Melinda Gates Foundation), fewer than 20% of eligible farmers were engaged with any activity on the platform. Adoption was stalling, and the team needed to understand why before committing to the next product phase.

I was brought in as Senior Product Researcher to lead a 14-week research programme spanning two phases: first diagnosing the right problems, then designing tested solutions. This case covers Phase 1 — the deep-dive research that produced the problem statements guiding all subsequent design work.

Tanzania

e-Kilimo deployment

PCT (aggregator) piloting with 20 of 300 agents in Mbeya. Transaction spikes driven by crop seasonality. NMB as financial partner creating confusion among farmers.

India

e-Rythu deployment

Araku Valley, Andhra Pradesh. Tanager NGO conducting transactions on behalf of FPOs (farmer producer organisations). Adoption masked by proxy usage.

Shared challenge

Structural low adoption

Both markets shared low literacy, poor mobile infrastructure, weak brand recognition, and an unclear value proposition for the agents most responsible for driving farmer uptake.

Business stakes

Labs incubation decision

Mastercard needed to determine whether MFN had product-market fit before scaling. Research findings would directly shape the go/no-go for Phase 2 prototype investment.

Research approach

The programme ran two parallel tracks — strategy and design — triangulated against each other to cut through noise and arrive at grounded problem statements rather than opinions.

01 — STRATEGY

Data analysis

Transactional database review. Identified adoption rates, seasonal patterns, geographic concentration, and which platform modules were actually in use.

02 — DESIGN

Field interviews

27+ contextual, in-depth interviews (IDIs) across Tanzania and India — agents, aggregators, FPO CEOs, lead farmers, buyers, and channel partners.

03 — BEHAVIORAL

Design audit

Audited MFN's value proposition, GTM approach, user experience, and pricing against a behavioral design framework for low-adoption digital financial services.

04 — SYNTHESIS

Insight clustering

Triangulated data and qualitative findings into clustered themes, segmented user needs, personas, journey maps, and prioritised pains and gains.

Key research insights

Six interconnected insights emerged from the fieldwork. Each one challenged a working assumption the product team held going into the research.

INSIGHT 01

Value proposition gap

The concept resonated with organisations, but delivery was falling short. After two years, PCT's pilot was still limited to 20 of 350 agents — and the second payment transaction feature that aggregators needed hadn't been built.

INSIGHT 02

Agents as critical — and overlooked — intermediaries

Agents were the primary touch point between farmers and MFN, yet they lacked English literacy (the app's only language), owned only feature phones, and had never received a clear articulation of why MFN benefited them personally.

INSIGHT 03

Channel partner brand transfer

In Tanzania, NMB's involvement created confusion: farmers associated e-Kilimo with unfulfilled promises of loans. In India, Tanager's strong community goodwill was the main driver of the limited adoption that existed.

INSIGHT 04

Brand recognition near zero

Most farmers in both markets could not name "e-Kilimo" or "e-Rythu" despite being registered. The digital scale — not the app — was the visible artifact they associated with the system. Brand exposure had never reached the 5–7 touchpoints needed for recognition.

INSIGHT 05

UI changes cause drop-out

For low-literacy users who automate app workflows through repetition, any UI update — even an improved one — forces a relearn that risks drop-out. No usability testing had been conducted before shipping updates to agents in the field.

INSIGHT 06

Mobile payment benefit was stakeholder-specific

Digital payments meaningfully benefit buyers and aggregators (cash security, speed). For farmers, mobile money was impractical — most lacked phones, and the nearest agent required a trip to town.

"At some stage, there are some modules and features that function well, but mostly, the one which we require at a high level — the 2nd payment registration — they didn't succeed in developing them clearly."

— PCT Aggregator Admin, Mbeya, Tanzania

Agent personas

Agents were MFN's most consequential and least understood user. I developed three distinct personas from fieldwork to give the product team a grounded picture of who was actually trying to use the app — and what was blocking them.

Persona 01 — Tanzania

The Committed Agent

Active user. Uses MFN with help from family members for English navigation. Has accepted technical difficulties as normal. Represents PCT's 20 piloted agents.

Needs: simplified UI, offline functionality, local-language support

Persona 02 — Tanzania

The Struggling Agent

Registered but disengaged. Understands the transaction steps but can't articulate MFN's value. Drops off after each UI update. At risk of permanent churn.

Needs: clear personal value proposition, stable UX, peer training

Persona 03 — India

The Lead Farmer / FPO Agent

High community trust, volunteer capacity. Does not own a smartphone. Tanager conducts transactions on his behalf. The "on-behalf-of" model masks real adoption.

Needs: USSD/feature phone path, harvest-estimation integration

Agent user journey

The journey map traced a PCT agent in Mbeya through a single produce purchase — surfacing where the friction was structural (infrastructure, literacy) versus designed-in (app update flows, registration steps).

AGENT JOURNEY — PRODUCE PURCHASE (e-Kilimo, Tanzania)
Awareness
Training session from PCT / NMB representative Pain: NMB messaging conflates e-Kilimo with loan access; creates expectation that isn't met Opp: single clear message about the app's job during farmer aggregation
Registration
Agent registers farmer on smartphone app Pain: app is English-only; agent uses local language exclusively; family member often needed Opp: Swahili/Hindi localisation; simplified registration with fewer fields
Weighing
Physical scale weighs produce; agent records in app Pain: physical scale overshadows app; farmers associate the "system" with the scale, not e-Kilimo Opp: brand both tools consistently as a unified system
Transaction
Record purchase on app; first cash payment to farmer Pain: 7–10 mins per transaction; phone freezes; cracked screen makes touch targets miss Opp: streamlined transaction flow; usability testing before each release
2nd Payment
Second payment after produce quality confirmed Pain: feature not built — agent must manage outside MFN; undermines platform value Opp: 2nd payment registration is the #1 feature request from aggregators
App Update
MFN releases an improved UI version Pain: agent's automated memory of flow breaks; drops off until PCT re-trains all 20 agents Opp: usability testing before release; staged rollouts with agent-facing changelogs in local language

Problem statements for ideation

Phase 1 closed with three prioritised problem statements — one per user archetype — delivered to the Mastercard Labs team for selection and advancement into the ideation phase.

PS 01 — BUYERS

Go-to-Market segmentation

Large businesses and small buyers have fundamentally different needs and risk profiles. A segmented GTM strategy — with large buyers as adoption stimulants downstream — was missing from the current approach.

PS 02 — AGGREGATORS

Channel partner communication

No channel partner was performing the role of consistent on-the-ground supporter. Clearer division of responsibility and structured agent-to-farmer communication cadences were needed.

PS 03 — AGENTS

Incentive model redesign

Agents bore the operational burden of the entire platform but received no clear answer to "what's in it for me?" A redesigned incentive scheme — time-bound, transparent, tied to real value — was the priority problem to solve.

Outcome & impact

The research programme delivered findings that directly shaped the Mastercard Labs product roadmap. By surfacing that low adoption was a multi-layer problem — not just a UX issue, but a GTM, incentive, and brand problem simultaneously — the team could prioritise with confidence rather than guessing.

+20%
increase in user engagement following usability-led fixes
+15%
faster product development cycle after cross-functional alignment workshops
+10%
new user acquisition uplift from data-led market trend recommendations

Beyond the metrics, the most durable impact was structural: the research established usability testing as a standard gate before any MFN app release — a practice that had been entirely absent before the engagement. The agent personas and journey map became reference artefacts the product team continued to use through Phase 2 prototyping.

Fixing the fundamental baseline issues will not only help drive MFN's adoption, but allow Labs to uncover additional challenges on customer usage, product design, and go-to-market plan.

— MFN Mid-Term Research Report, Digital Disruptions, October 2019
← Impacc Back to all work →