Amazon — Case study
Amazon Experiments
30% experiments, 70% delivery.
At Amazon, design roles often come with well-defined deliverables. But as a manager, I believe it’s just as important to make space for curiosity and invention.
That’s why I encouraged my teams to adopt a 30/70 mindset: dedicating 30% of our time to experimentation, as long as it could be tied back to improving the Amazon customer experience.
The experiments shown here reflect that mindset. Some evolved into full products. Others sparked ideas that shaped future initiatives, and a few are still waiting for the right moment.
Some were part of broader Amazon programs, where I contributed but don’t claim full credit.
Experiment 01
An AR experience with your Amazon box
Inspired by the launch of our newly designed Amazon boxes (see Experiment 03), my team and I explored how to turn everyday packaging into a portal for entertainment. We prototyped an augmented reality experience that brought the box to life, bridging physical delivery and digital storytelling.
A) Box arrival
Customers receive a specially printed Amazon box with a call to action to scan it in the Amazon app.
B) Activation
A simple scan triggers the AR experience directly in the app.
C) Immersive reveal
3D objects emerge from the box in augmented space, from characters to scene elements related to the campaign.
D) Interactive journey
Users interact with the AR elements to unlock deeper content: movie trailers, product deep links or themed Easter eggs.
The approach
We pitched the idea to our creative technology team, who quickly built a prototype. Thanks to its cubic shape, the physical box worked perfectly as a marker, so the entire box became an AR object to interact with.
Evaluate and learn
We started with an internal MVP rollout and encouraged as many colleagues as possible to test it. Office hours and input from across teams gave us valuable insights (and a healthy dose of suggestions).
The key learning: object interaction fell short. Many testers struggled to move elements smoothly in AR, so early iterations pivoted to a more passive, viewable experience, focused on visual delight over control mechanics.
To gauge market potential, we partnered with the advertising sales team and showed the prototype in client meetings: a lightweight, real-world test of interest.
We handed our documentation and demo to the product team to build on. After our prototype, Amazon began launching official AR experiences tied to shipment boxes, a clear sign that our early exploration helped shape future product directions.
Experiment 02
Gamifying the Alexa experience
When Amazon Echo launched in 2014, my team in Germany was asked to prepare a local campaign introducing this new voice technology to a broader audience. Around the same time, Amazon Prime Video partnered with Michael “Bully” Herbig, one of Germany’s most beloved comedians, on an exclusive comedy format.
We saw an opportunity to combine the two: a custom Alexa Skill that let users interact with Bully’s voice through jokes, sketches and character responses. Part campaign, part experiment, it showcased entertainment and technology in one experience.
We built the experience around one of Bully’s most iconic sketches, the “Kastagnetten” bit: three secret agents who constantly finish each other’s sentences in absurd, synchronized rhythm. Its timing, interruptions and quick dialogue were a natural fit for Alexa’s interaction model.
Our goal was to let users trigger punchlines, complete sentences or get hilariously interrupted by Alexa, blurring the line between scripted comedy and responsive voice UX.
A three-agent rhythm
Bully Herbig voiced the first agent, Alexa became the second and the user played the third. A predefined dialogue between Bully and Alexa set the stage, and the user completed the punchline or triggered the next beat.
Once users completed the sequence, they entered a contest to win premiere tickets for Bully’s new movie.
Voice flow
User: Alexa, play <GAME>.
Alexa: Short introduction.
Actor: Short introduction.
Alexa: Are you ready to play?
User: Yes.
Alexa: Block A
Actor: Block A
User: <Respond>
Alexa: Block B
Actor: Block B
User: <Respond>
Alexa: Block C
Actor: Block C
User: <Respond>
Alexa: Enter-a-chance-to-win message.
Testing with people
Our copywriters wrote the first script based on research into video footage of the show from the late ’90s and early 2000s. We wanted to learn whether users could complete the sentences without knowing the series.
For phase 1, we blocked out three meeting rooms. Each team member played the Actor, Alexa or the user, communicating by audio only, to test the script and how natural it felt. We then invited colleagues to test it in the same rooms. Each test took only 5–10 minutes, which gave us good qualitative input.
As we refined the script, we learned how to guide testers through the interaction. But as with any voice-only experience, disorientation was a real risk when timing or input didn’t go as planned.
Our answer was a fallback inspired by the iPhone’s home button: users could say “Help” at any point to reset, and the system triggered the fallback automatically if no response came within 5 seconds. The experience stayed safe and recoverable, even when a joke didn’t land the first time.
We built a working MVP and even demoed it live with Michael “Bully” Herbig himself, an unforgettable moment for the team.
The full campaign was later cancelled and never tested at scale, but we shared our prototypes and learnings with the Alexa product team. Since then, a wave of gamified voice experiences has emerged on Alexa, and it’s rewarding to know our early exploration helped pave the way.
Experiment 03
Branded Amazon boxes
In 2015, I joined a cross-functional design initiative exploring a bold idea: what if the Amazon box itself became a branded storytelling surface?
Led by the entertainment teams in Los Angeles, our global design group prototyped a series of Amazon boxes, each turning packaging into a new customer touchpoint. My role focused on ideation and prototyping, working with our EU teams on how print, layout and visual narratives could be embedded directly on the boxes.
Before production, when fulfillment centers would eventually get on-demand printing, we had to define the creative and logistical constraints. What kind of branding works on cardboard? How would customers react? Could it scale across regions and campaigns?
Our fulfillment centers provided test-printed boxes based on our early designs, so we could evaluate print quality, layout scaling and how branding held up in real shipping conditions. It was a hands-on, iterative process that bridged creative vision and operational feasibility.
For early prototypes, we developed a quick, scrappy method to test ideas ourselves: we laser-printed designs in reverse on standard paper, then transferred them onto cardboard with turpentine and a sponge.
It was fast and surprisingly effective, letting us assess layout, contrast and storytelling potential long before industrial printing was in reach.
We also used the boxes in client-facing storytelling. Each prototype was packed with the full pitch materials (concept decks, campaign previews and mockups), delivered in our own branded Amazon boxes. The box became the first impression, helping clients grasp the concept immediately and making it feel tangible.
Closer to rollout, we hit a key constraint: German fulfillment centers didn’t yet have the on-demand printing capabilities of those in the U.S. We switched to printed wraps, which could be produced, distributed and applied across multiple centers, preserving the creative impact within regional production capabilities.
The wider team’s efforts contributed to the launch of multiple branded box campaigns for Amazon clients, turning everyday packaging into a moment of delight for millions of customers worldwide.
