Amazon Trade-In helps customers get rid of their old electronics sustainably. With an extensive selection and competitive trade-in value, it eliminated the hassle of determining electronics' value and finding trustworthy buyers.

Since the Trade-In team observed increasing customer complaints, I joined the team to develop a Northstar design to tackle customer pain points and uncover potential opportunities. Over six months, I synthesized customer insights from generative research findings, facilitated brainstorming workshops, conducted rapid prototype testing, brought ideas to high-fidelity designs, and carried out evaluative research. This extensive process led me to create a Northstar design and successfully launch Minimal Lovable Products (MLPs), which garnered strong support and funding for the trade-in team and paved the way for further enhancement of the customer experience.

My Role: UX Design & Evaluative UX Research

Collaborators: UX researcher (0.2), Software Development Manager (1), Business Analyst (1), Product Manager (1), Operation Manager (1), Sales Manager (1)

Project Outcome: our successful MLPs, launched with limited dev resources, convinced the leadership team to shift existing priorities to enhance the customer experience further.

Company: Amazon

Before and After

There are three experience goals for this project. Below is a glimpse of how the redesign accomplished all three experience goals.

Gather Customer Insights

When gathering customer insights at the beginning of a project, I used the framework below to request data and user research support from the data science and the user research team. The framework ensured that I was getting high-quality and actionable customer insights.

For example, after conducting customer interviews and remote usability testing (through the user research team), I discovered numerous customers had difficulty choosing the appropriate electronics model on Amazon and obtaining a precise quote. Upon analyzing the data science team's funnel data, I observed that the most significant drop-off occurred during the model selection process. Triangulation of data provided me the confidence to hypothesize that model selection was a severe customer pain point that impacted many trade-in customers.

DISTILL RESEARCH INSIGHTS

Customer Journey

Based on the framework above, I created a list of customer concerns and opportunities that were both significant and affected a large number of customers. Then, I made a customer journey map to help my team visualize these concerns and opportunities.

DISTILL RESEARCH INSIGHTS

Experience Goals

After discussing the customer journey with my team, we defined three goals to improve the Trade-In customer experience.

1. Increase Trade-In discoverability and improve Trade-In Education.

2. Ensure customers can easily find the right device model and estimated value.

3. Reduce effort in sending Trade-In items to Amazon.

COLLECT VARIED POVS

“How Might We” Team Brainstorm

I held brainstorming sessions with teammates (SDEs, PMs, analysts, and operation managers). We organized opportunities and customer pain points into six HMWs (“How Might We”). Teammates contributed their ideas to each HMW. Then we discussed and refined each idea.

COLLECT VARIED POVS

“2x2 Matrix” Idea Prioritization

To prioritize ideas with all perspectives, I have my team vote on ideas based on their expertise (e.g., engineers voted for highly feasible ideas). Those ideas were then organized into a 2x2 Matrix.

BRING IDEAS TO ACTION

Flow Chart

I held virtual whiteboarding sessions with my team to create a flow chart with our freshly baked ideas. Putting ideas into the context of the trade-in flow also helped us refine our ideas. After multiple whiteboarding sessions, we aligned on the most promising user flow below.

NAVIGATE DEBATES

Wireframe Testing

One major debate arose at this stage: should we surface “upgrade with Promo” first or “send in electronics” first? My business partners pushed to have customers shop immediately after they submitted Trade-In. At the same time, I was concerned that surfacing shopping messages would distract customers from their current trade-in journey.

I created two wireframes (“Option A” and “Option B” below) centered around the debate and ran an unmoderated user test.

Surprisingly, customers reacted negatively to “Option A” and “Option B.” Since the CTA “Submit my Trade-In” created the impression of finality to the trade-in, customers who did not understand how to send in items (or, in general, “how trade-in works”) reported anxiety/hesitation. I improved my design with this new data and resolved the debate with business partners (designs below).

CONNECT DIGITAL AND PHYSICAL

Storyboard the Post-Submission Experience

After customers submitted the Trade-In, their touchpoints with Amazon Trade-In shifted from amazon.com to email and in-person interactions. I used storyboards to communicate my design vision and facilitated cross-organization collaborations.

BIG PICTURE TO DETAILS

Generate High-Fidelity Designs

I have aligned with stakeholders on important product decisions at this design stage and resolved all significant debates. I moved on to create high-fidelity designs for 1) user testing, 2) engineering effort estimates, and 3) leadership presentations.

Turning wireframes into high-fidelity design involves both creative and analytical thinking. Taking the first screen as an example, I started by creating as many options as possible, then condensed to the three most promising options and discussed them with PM & SDE. Referring to “Hick’s Law” (the time it takes to decide increases with the number and complexity of choices) and my PM’s product data that (in the short-term) over half of the customers traded in electronics bought from amazon.com, I settled on two most desirable options to test with customers.

The process described above was repeated for each sub-flows. After hundreds of decisions, I turned all wireframes into High-Fidelity designs and created a prototype for user testing.

DESIGN ITERATIONS

Customer Insights on Designs

I tested my design prototype and started the next round of design iteration. Below are research insights highlights.

💡 Refine design assumptions: Before the user research, I assumed both “decision tree” and “refined search” approaches serve the purpose of helping customers effectively find items that weren’t registered or purchased. Through user research, I found that customers generally take twice as long when using the “decision tree” approach. Customers using “decision tree” also had a harder time recovering from errors than “refined search.” Therefore, I chose “refined search” as the final solution.

🔆 Surface new opportunities: I discovered that customers with expensive electronics that are difficult to identify (e.g., computers) would prefer to see offline trade-ins early, as do customers who live close to an in-store trade-in location. Therefore I surfaced offline trade-in early in the flow.

🛠 Solve UX issues: participants relied heavily on device images to select and confirm their models. Some devices also look similar from one angle and require multiple images to distinguish. I improved the card’s design to feature a large device image with the option to view multiple images.

DESIGN ITERATIONS

Desktop Version

I kept the desktop version most in parity with mobile, besides a few layout and interaction adjustments optimized for desktop.

IMPLEMENTATION

Strategic Prioritization

I worked with PM and Engineers to break the design into four PR/FAQ documents (a written document Amazonians use to propose a product idea in its early stages). Due to limited dev resources for the Amazon Trade-In program, we could not execute all four projects simultaneously. I created a chart below to help my team understand the CX impact of each project. Using the chart and other business data, Trade-In leadership recognized how all four projects work together and abandoned the initial idea of picking only one project to prioritize.

IMPLEMENTATION

Successful MLPs

We needed data to convince the leadership team that those designs would benefit the business. I created Minimal Lovable Products (MLPs) for each PR/FAQ by distilling the most valuable CX of the North Star Design.

To date, we launched two MLPs with existing dev resources. One MLP had +170% improvements in the submission completion rate; another had a +41% lift in upgrades (where customers bought an Amazon device subsequently after their trade-in). Our successful MLPs, launched with limited dev resources, convinced the leadership team to shift existing priorities to enhance the customer experience further.