Selected work

Best Buy · Checkout · Design & Delivery

Building the Future of Checkout

Turning a concept into a scalable checkout — and protecting the experience through technical complexity, edge cases, and experimentation.

Role
Experience Designer
Teams
Commerce View, Payments, Fulfillment
Timeline
6 months
Focus
Interaction design, systems thinking, design QA, experimentation
Outcome
A scalable, highly performing checkout experience

The Ask

“Increase checkout conversion in the Best Buy app and mobile web experiences.”
— Best Buy product leadership

Best Buy's previous checkout redesign had failed its A/B test. This time the direction was validated in discovery, but it still had to hold up through execution.

Discovery aligned the Commerce View, Payments, and Fulfillment teams on a single direction: a one-page accordion built around four customer needs, confidence, reviewability, control, and transparency. That first half of the process is covered in Defining the Future of Checkout.

My role was to protect that direction through execution, and turn it into a checkout that met the four customer needs and supported a business case the organization could commit to as the future of checkout.

Business outcome

A scalable checkout foundation

A single, extensible checkout the organization aligned on as the future of checkout — capable of supporting the complexity of modern commerce.

Customer outcome

A checkout customers can trust

The four needs from discovery — confidence, reviewability, control, and transparency — translated into visible order details, editable steps, and a clear sense of what happens next.

Challenge

Advocating for the Experience

A critical turning point occurred when I proposed interaction changes that better supported customer needs but expanded implementation scope.

The proposal introduced additional engineering complexity and challenged initial delivery assumptions.

This led to a series of discussions across Product, Engineering, Checkout, Payments, and Fulfillment.

After weighing short-term implementation costs against long-term customer value, the organization aligned around pursuing the stronger experience.

Old checkout interaction: a full-page “Select a payment method” spoke page that takes the customer out of the checkout context

Before Inconsistent Spoke page interaction model

New checkout interaction: a payment method drawer overlay that keeps the dimmed checkout page visible behind itCheckout stays in view

After Refined drawer & side sheet interaction model

Fig 1The interaction model, before and after. The spoke page takes over the full screen and breaks context; the drawer keeps checkout in view.
Option A

Original Interaction Model

Option B · Recommended

Proposed Interaction Model

Customer Benefits

The full page takeover gives each step the entire screen, with the most room to work with.

Keeps customers in the context of checkout. It feels like a lighter, quicker interaction, and creates one consistent pattern across every step.

Engineering Considerations

Little to no dev lift, so we could test the experience sooner.

Heavy dev lift, requiring a full rebuild of interactions across checkout.

Business Impact

The cheapest, fastest path to a live test, favoring short term speed to market.

Higher upfront investment for a scalable, consistent checkout foundation.

Final Recommendation

Pursue the drawer model. The heavier engineering lift was outweighed by a consistent, in context experience that scales across all of checkout.

Fig 2Decision framework. A side by side read of the original and proposed interaction models, weighed to a clear recommendation.

The Complexity

Designing at Scale

Designing the primary flow was only part of the challenge.

The experience also needed to support dozens of customer scenarios and edge cases.

I partnered closely with the Commerce View (Checkout), Payments and Fulfillment teams to ensure consistency across every variation and ensure scenarios across every dimension were supported.

Guest checkout: an empty Ship To form the customer fills in
Guest & authenticated usersMultiple component states for different types of users (pre-filled, empty, etc.)
Store pickup with In-Store, Locker, and Curbside options
Three methods of fulfillmentShipping, store pickup, digital delivery
Payment screen showing card entry plus Zip, Apple Pay, and PayPal methods
Four methods of paymentCredit card, Apple pay, PayPal, Zip
Cart with a protection plan and a trade-in credit applied
“Add-on” servicesProtection plans, Trade-ins, Gifting, etc.
Terms and conditions with auto-renewal subscription disclosures
Terms & conditions updatesThe entire experience stayed compliant.
Ship To form showing an ‘unable to ship to this address’ error
Dozens of error scenariosProviding clarity when things go wrong.
Fig 3Edge case gallery. A sample of the many scenarios the one flow had to support.

The Experimentation

Testing Reality

Following implementation, the redesign entered A/B testing against the legacy checkout experience.

The initial results were disappointing.

The redesigned experience showed roughly a 1% decrease in conversion in the web experience, but flat in app.

At first glance, it appeared we had another unsuccessful checkout redesign.

Control (A) Redesign (B) 73.06% 71.01% Desktop −2.0 pts vs control 58.96% 58.32% Mobile web −0.6 pts vs control 57.62% 57.22% App −0.4 pts vs control
Fig 4People CR by surface, redesign against control. App and mobile web held flat, while the desktop view fell about two points, the one surface with a real detriment.

Key metrics

The numbers we watched

To understand the full impact of the redesign, we looked past the headline conversion number to a focused set of metrics across the funnel.

$327

App AOV
−$8 vs $335 control

$311

Mobile web AOV
+$1 vs $310 control

$426

Desktop AOV
−$2 vs $428 control

Looking past the headline

Almost all of the movement was in desktop conversion, down about two points, while app and mobile web stayed flat. The desktop result was the one we investigated next.

Looking deeper

Questioning the Results

Rather than immediately abandoning the direction, the team investigated the results more deeply.

As we analyzed the experiment, we uncovered limitations within the testing setup that had not been fully understood before.

We found that a select few products carried 95% of the conversion detriment. All signs pointing towards reseller activity skewing the numbers towards the control.

With this discovery, along with identification of several bugs, our results proved inconclusive. We needed to iterate the designs, our testing setup, & our measures of success.

IdeaPad 15.3 R7 16GB / 512GB Reseller-targeted SKU
Control (A)
Test (B)
People CR
29.76%
16.38%
Share of people
52.1%
47.9%
Share of orders
66.4%
33.6%
Fig 5One reseller-targeted product. Traffic to this SKU split almost evenly, but control took two thirds of the orders and nearly double the People CR. That points to reseller purchasing concentrated in the control, not a weakness in the redesign.

Iteration

The Iterated Designs

The design advanced through iteration, with every decision backed by data.

Each step stays visible and editable while the flow guides customers forward. The direction from discovery, realized in high fidelity.

I worked in tandem with product and engineering to refine the designs while they resolved bugs, and controlled for resellers.

One-page checkout, iteration 1: full mobile screen with shipping, ship to, contact info, and payment sections under a Back to cart header
Iteration 1
One-page checkout, iteration 2: full mobile screen with all steps completed and checked, shipping details placed first, ending in the order summary and Place order
Iteration 2
One-page checkout, iteration 3: full mobile screen with contact, shipping, and payment steps completed and checked, ending in an order summary and Place order
Iteration 3

The results

Outcome

The redesign established a new foundation for checkout at Best Buy.

More importantly, it translated customer insights into a scalable experience capable of supporting the complexity of Best Buy's offerings.

Although validation continues, the organization aligned around this direction as the future of checkout.

The final one-page checkout, complete: contact info, shipping address, and shipping details each collapsed to a completed checkmark, payment method entered, and an itemized summary with delivery date, savings, total, and a Place order button
Fig 6The final experience, mapped back to the four customer needs from discovery.
01

Confidence

Each finished step collapses to a completed checkmark, while the delivery date, savings, and running total stay in view. Customers can see the order is right before they commit.

02

Reviewability

The whole order lives on one page, backed by a persistent summary and a dedicated review sheet. Nothing hides behind a final confirmation screen.

03

Control

Every completed section keeps a Change action that reopens that step in a focused sheet. Customers can edit anything, in any order, without losing their place.

04

Transparency

Subtotal, shipping, and tax are itemized up front, delivery dates are explicit, and added costs like financing are labeled. No surprises at the end.

Learnings

What I Learned

Execution required protecting that direction through technical complexity, organizational constraints, and uncertainty.

The challenge was creating a simple, elegant experience on top of an incredibly complex system.

This project reinforced that great product design requires more than interaction design. It requires systems thinking, alignment, and the ability to make thoughtful tradeoffs while staying focused on customer needs.

More case studies