Improving user workflows through mixed methods Research
Glossary
BEES: a business-to-business e-commerce platform built to empower small and medium-sized retail partners and help grow their businesses. It mainly caters to CPG (Consumer Product Goods), such as beverages, packaged food, etc.
BEES Deliver: a suite comprising a mobile application for truck drivers and delivery personnel, along with a web application for back-office employees at DDCs (Distribution and Delivery Centers) to facilitate delivery fulfillment.
DDC: Direct Distribution Center
POC: Point of Consumption (a.k.a. Point of Sales)
📝 The context
BEES offers a comprehensive end-to-end solution through its applications for e-commerce sales, sales force management, customer support, and logistics fulfillment.
The BEES app is designed for customers—specifically, individuals at Points of Consumption responsible for restocking inventory at restaurants, bars, grocery stores, and similar businesses.
BEES Deliver is a suite comprising a mobile application for truck drivers and delivery personnel, along with a web application for back-office employees at DDCs (Distribution and Delivery Centers) to facilitate delivery fulfillment.
Each day, trucks depart from DDCs carrying cargo for delivery to POCs, where they also collect payments and proof of visit. However, the BEES Deliver suite was created based only on technical requirements to facilitate last-mile delivery logistics of a B2B e-commerce company.
Since its inception, there has been no comprehensive mapping of foundational user experience data, such as user journeys and user personas.
So, to continue expanding and improving the product, it was crucial to understand users' journeys, personas, and pain points to make informed decisions on enhancing their experience.
🎯 The Challenge
BEES Deliver is a logistics app used by drivers, monitors, and supervisors across BEES markets. However, it faced major friction points in performance, usability, and trust in data.
The team needed to:
Understand pain points with real-world evidence
Identify usability blockers in proposed updated screens
Prioritize product decisions based on qualitative + quantitative insights
🔬 Research Breakdown
📊 1. Internal NPS Survey
2,047 responses from 8 countries
Users reported poor app performance, too many clicks, and outdated information
Key complaints: incorrect POC locations, bad route suggestions, issues with app performance
🧭 2. Field Research – Paraguay & Peru
Delivery crew
Command Center - Monitoring (backoffice) team
Delivery crew - payment collection
Contextual inquiry and interviews with delivery crews, route monitors, and supervisors (backoffice team)
Insights from real delivery environments and DDCs
Mobile App Insights
Drivers wanted more autonomy (e.g., solve simple issues on route)
Critical need for accurate delivery windows (times where POCs can receive deliveries) and map navigation
Desire to chat with POCs directly via the app
Issues with digital payments and excessive paperwork (in Peru)
Difficulty with the map view, where it would randomly zoom out, frustrating navigation
Web Portal Insights
Teams rely on several external tools (WhatsApp, Foxtrot, Roadmap, etc) due to feature gaps in Deliver
KPIs are often untrustworthy or outdated
No easy way to analyze historical vs. real-time data
Need to centralize communications and optimize route suggestions
🧪 3. Usability Testing
13 participants from 🇵🇾 Paraguay, 🇵🇪 Peru, and 🇧🇷 Brazil
Validated new flows for the main tasks performed daily on the app:
Safety checklist
Inventory checklist
Visit/task screens
Task Performance Highlights
✅ High success rate in basic flows (e.g., finishing visits, editing items)
❌ Poor discoverability of “Postpone Visit” feature (35/100)
⚠ Confusion between tabs (e.g., users expected to edit products quantities under “Products” tab instead of “Documents”)
👥 Users & Context
Delivery team, monitors, DDC supervisors
Daily workflows often involved multiple apps, paper, and improvisation
Each country had unique logistics routines and compliance needs
✅ Key Recommendations
Allow drivers to self-correct simple issues on route
Improve app performance
Redesign low-discoverability flows like “Postpone Visit”
Reduce tool-switching with centralized comms and dashboards