top of page
Role:
Date:
UX Research, Design, Testing
2024
What is Flex?
The Amazon Flex application is the main tool depended upon by drivers performing deliveries across the world. To date, 2.9 million people have downloaded this application.
My Motivation
My time as a Lead Delivery Driver for an Amazon DSP led me to suspect the app could have greater value by improving driver workflow efficiency.
RESEARCH
Does the Flex app hinder delivery service?
Who are the People Involved?
Direct Users:
Amazon Delivery Associates (DA)
Delivery drivers working for contracted Delivery Service Providers (DSP) who rely on the Flex app for route management and performance.
Independent Contractors*
Individuals working on a "gig" basis who use their own vehicles for deliveries.
*(outside scope of this project)
Indirect Users (Direct Impact):
Delivery Service Provider (DSP)
Companies contracted by Amazon that employ the DAs who use the Flex app. Their business success is tied to the performance of their drivers.
Customers
People receiving deliveries from Amazon. Their satisfaction is directly impacted by the efficiency and accuracy of the delivery process facilitated by the app.
What are their Goals?
Amazon
”...customer obsession rather than competitor focus, passion for invention, commitment to operational excellence, and long-term thinking...” Amazon.com
Delivery Associates (DA)
Deliver all packages with top performance: no returns, errors, or complaints from customers.
Delivery Service Partner (DSP)
“...efficient, reliable, and always on time...”
Customers
Delivery of products that are on-time, undamaged, accurate to instructions, with excellent service.
Address any Flex app impediments, to thus boost drivers’ productivity (KPI) and ultimately customer satisfaction.
What’s my Value Proposition?
Hypotheses RE: Performance
Drivers would prefer a way to better monitor their performance in real-time.
Feedback
H4
Drivers believe the current process to remedy delivery mishaps arduous.
Error Correction
H5
Drivers would prefer a method to prevent actions that adversely affect their performance score.
Warnings
H6
Hypotheses RE: Speed
Drivers find organizing packages cumbersome, necessitating better in-app organization.
Organization of Stops
H1
Drivers prefer to have a tool to monitor their progress and time in-app, than week-end reports.
Route Progress Monitoring
H2
Drivers prefer to speedily remedy system errors involving GPS Location.
Location Services
H3
What are my Assumptions?
Priority
?
Changes to the app, above other tools and processes, needs improvement to achieve higher performance.
Statistical Significance
?
That metrics are affected significantly (positive or negative) by the Flex app’s design
Trade-off
?
That additional features can be implemented or altered without affecting current uptime and data reliability.
With these assumptions noted, I proceeded to form the following hypotheses:
Feasibility
?
That changes to the “Flex” app are technically and financially feasible.
Research Results
Survey responses from 8* drivers validated all hypotheses.
Pain Points
Stops on the same block are not clearly distinguished. (H1)
Overlapping, Hard-to-Visualize Stops
PP1
Organizing large deliveries and multi-location stops is time-consuming (H1)
Cumbersome Multi-Location Handling
PP2
Excessive taps/screens slow down delivery completion. (H1)
Too Many Steps for Delivery Completion
PP3
Stops aren’t sequenced properly, leading to redundant driving (H2)
Poor Route Optimization
PP4
No clear way to know if drivers are ahead or behind during their route. (H2)
Lack of Real-Time Monitoring
PP5
Drivers want to know how they’re performing relative to set goals. (H2)
No Comparison Metrics
PP6
GPS errors and route inefficiencies causing delays. (H3)
Inaccurate Navigation/Mapping
PP7
Drivers want more specific, real-time insights on their performance. (H4) (H5)
Lack of Detailed, Real-Time Feedback
PP8
Limited ability to easily fix errors in the app. (H5) (H6)
Difficulty Correcting Mistakes
PP9
Drivers often don’t know why their performance score has dropped. (H6)
Unclear Performance Deductions
PP10
Frequent Crashes/Glitches
App crashes or glitches, especially when submitting Picture of Delivery (POD).
PP11
DESIGN
How? ie. Methodology
A short-answer survey* that aimed at understanding direct users’ (drivers) current experience, with permission from management. This approach allowed me to deploy questions in a way that would not interrupt their work.
*Further research using in-depth interviews, and subsequent preference, usability, and learnability testing were not conducted.
Sketches
Therefore, our Problem is:
Design flaws in the Amazon Flex app hinder the speed and quality of package delivery service.
What does the Solution require?
Three key pain points identified by drivers confirmed the Organization Hypothesis, highlighting the need to streamline workflows and redesign the app’s UI elements--essentially an app overhaul. Specifically, the third pain point, “Too Many Steps,” becomes clear when examining the existing user flows:
Drawing inspiration from mapping applications, I recognized that drivers primarily need to know their exact location. Regardless of any actions taken along their route, the primary task in logistics is positional navigation--making sure you are in the correct place. Thus, location information takes precedence over all flows, which necessitated the following redesign:
User Flow Diagram
Process
Decision
PP
Pain Point
Dialog
In similar fashion to the Menu, flows involving dialogs (Customer Contact) ought not obscure navigation information. Dialog boxes as widgets float above the current flow, allowing for fast task switching instead of complete interruption.
Chat Dialog Box
Chat
Menu
Buttons
Title
Menu Box
Dashboard / HUD
Items
Route Box
The main menu similarly ought to provide a way to access sub-workflows / tasks that without full interruption to current stage or the main task of navigation. With the Dashboard and Map remaining in view (albeit shadowed), users can access items such as “Settings” or “News” without losing sight of the task at hand.
In contrast with the current menu, details and actions relevant to an active route are shown in a “Route Box”, a menu widget. This reduces visual clutter and allows for faster access.
Hi-Fi Wireframes
Stop #
Map
Sub-Workflows
Details
Stage
Main Action Button
Menu
Next Stage Button
Selection
Here, I use a widely adapted paradigm (think windowing GUIs) that places the current workflow or “Stage” as a subtask to the main, ongoing task--in this application, map navigation. Additionally, the Stage design acts as a visual framework, allowing any point along the user flow to maintain consistency without sacrificing versatility. This also allows the app to scale for additional workflows without changing familiar navigation elements.
Ultimately, these sketches yielded the following high fidelity wireframes:
Drawing inspiration from mapping applications, I recognized that drivers primarily need to know their exact location. Regardless of any actions taken along their route, the primary task in logistics is positional navigation--making sure you are in the correct place. Thus, location information takes precedence over all flows. The following sketches rendered the first ideas about how to do this:
What design solutions already exist?
Addl’ Q’s : Hardware UX
The 11th pain point mentions “frequent glitches” and “crashes”. Further research into these areas are needed to understand why these might occur:
Device Performance
Does the hardware perform well for other applications? Are there overheating issues? Are there battery performance issues?
OS Performance
Does the OS configuration (memory, storage, updates, etc) create instabilities?
Application Version Management
Are older versions of the flex app in use on certain device setups?
Platform-Specific Bugs
Are there iOS or Android specific issues causing app glitches?
Network Performance
Is the data or Wi-Fi network secure and reliable for all areas of use?
MINIMALLY VIABLE PRODUCT
Route Generation / Sequencing
Are there ways to reduce redundant stop creation?
Location Services/Navigation
Are there ways to provide less erroneous road pathways? Feasible to add a feature that allows reporting road hazards or route errors?
Error Correction
Would the feature to “undo” adversely affect system stability / backend ?
Performance Feedback
How will real-time feedback and performance data be generated and delivered?
Backend
Will the current backend system support these changes / be adversely affected?
What additional infrastructure or APIs might we need to support these features?
Testing
How will you ensure that the redesigned app performs well across different devices and operating systems?
Questions for Engineering
Hours : Minutes
Stops/Locations
Warnings
8:00
200
11
4
2
Packages
Reattemptables
Heads Up Display.
Mike
Leaner.
Seamless
Stages.
The following prototype brings to life the "Stages" redesign of the Amazon Flex App, addressing the key user needs identified in earlier research. To illustrate the impact of these changes, I present side-by-side comparisons with the current application design, highlighting the simplicity and usability of the new approach.
Not TMI.
Simple Nav.
Report Inaccuracies.
Toggled Package Scanner.
One Step
at a Time.
Read Up-To-Date CX Notes.
Prioritized Access Information.
Report Issues.
See & Report Safety Concerns.
Easy Customer Contact.
Timed
Chat Dialogues.
Thwart the “Oops”
Finish Delivery, now with added Confidence.
Delve into Performance
Metrics
anytime.
Chance to Undo.
EVALUATION
Usability
Showing the design prototype to 2 seasoned drivers* with the prompt, “Deliver to 633 Bedford Ave”, they were able to navigate the new design with no interventions.
Did the redesign achieve its goals?
That is, does it add value for everyone?
*More extensive testing utilizing many more drivers, in different geographic settings, is needed to truly determine if the redesign accomplishes the goals set forth in the research.
Yes, feedback demonstrates positive achievement of goals.
bottom of page