AI Image Generator
AI Image Generator enables marketers to create highly engaging ads within clicks. My responsibility is to scale it by working backward from customers—staying aligned with the long-term vision while delivering on business goals.
Role
UX lead
Areas
Strategy, Vision
Responsive web design
UI & Interaction patterns
Year
Feb 2024 - Feb 2025
(11 UX improvements / new features launched)
What’s Launched
Modern Simplified Experience: To optimize workspace while preserving context, I led the shift from a modal to a side sheet interface. I also modernized and simplified the layout to ensure a smoother user experience.​
Migration from modal to side sheet, completed Apr. 2024
AI-Generated Editing: To address the top pain point, I created Amazon Ads' first AI-powered editing feature, enabling marketers to adjust product scale and position in generated images.
Product scale and position editing, launched April 2024
1-Click Image Selection​​: To streamline image selection, I designed a 1-click experience, making the process faster and easier for marketers.
1-click image selection, launched Aug 2024
Within one year of launching Image Generator, as the sole UX lead, I partnered with Product, Engineering and Science on launching a series of new features and improvements, we’ve seen
Active advertiser adoption
+1,880%
Campaign submissions
+39%
From 2.5k to 46k
Product sales
+5%
AI-GEN ads revenue
+745%
From $828k to $7M monthly
“My category has high costs with setting up a shooting environment to showcase my products. So I prefer to use this tool to generate backgrounds and reduce my budget expenses.”
– Seller, small business
“When I need a large volume of image assets, such as for Posts and Stores, this tool can help solve my lack of inspiration.”
– Seller, large brand
Customer feedback, Nov. 2024
Background
The Audience​​
​
Amazon Ads users include small brands (80%), large brands, and agencies—each facing different challenges, from limited resources to slow turnarounds.
The Biggest Pain Point​​
​
For many years, the lack of creative resources is one of the biggest constraints for brands to grow their businesses through advertising. Without engaging ad experiences, businesses face difficulties attracting shoppers and expressing their brand identity.
Ads examples 2023
The Solution: Image Generator MVP Launch​
​
To solve this biggest pain point, we launched Image Generator at end of 2023. With image generator, advertisers no longer face the problem of resource constraints, any brands can easily create images featuring their product for free and use those images in their ads.
AI Image Generator launch video, Oct 2023
Launched experience within campaign creation flow
Objectives
The Vision​
Empower any advertisers to create engaging ad creatives within minutes.
The Business Goal​
Increase adoption from 2.5K (Q4 2023) to 50K (Q4 2024).
Working Backward from Customers
Problem Statement
​
How might we enable marketers to easily create quality AI-gen images, so that they can launch engaging ads quickly?
Design thinking workshop led by me, Jan 2024
User interviews and survey, Q1 2024
Top Insights​
​
Advertisers want more control.
The generated images are “so close”—but ultimately do not meet our customers' quality bar. This phrase was mentioned in every interview, even by customers who decided to use our AI images in live campaigns.
​
Advertisers want more guidance.
Advertisers are often not creative specialists. They don’t know what to do when it comes to an empty prompt box. They want to create nice images, but they don’t know how to get there.
“So I don't wanna make it look like, oh, this product is only this size when actually, it's larger or smaller. That's a bit like false advertisement. So that's probably why I was a bit hesitant to just go live with them because of the sizing issue.” – Seller
“The cutout effect is not very precise, and my product is in a floating state. This image is supposed to have a vanity as the background, but the generated image has a dinning table in the background.”
– Seller
“You guys have data on what the product is, maybe even looking at the title like a suggestions or like a generated recommendation so that you can like try this product on bathroom counter, that could be helpful...At least just to start and then people could always modify from there.” – Seller
“I have difficulty in how to write an effective prompt to generate the images I need.” – Seller
Quotes from user interviews, 2024
Design for Co-Creation
Focus​: Facilitating a co-creation relationship between users and the tool.
​​
To achieve that, we will focus on feature expansion and building a scalable foundation.
1. Empowering Creative Control with AI Editing
​
The biggest barrier to image usage was incorrect product size or placement—cited in 38% of downvotes. To solve this, I created Amazon’s first AI-generated image editing capability. Following its launch, adoption reached 30%, and related downvotes dropped to 23%.​
Option 1
Option 2
Iteration
Timeline
1 mo.
Product sizing as downvote reason
-15%
Images saved that are edited
42%
2. Providing Guidance and Inspiration
​
To support advertisers with prompt writing, I partnered with the science team to create a guided prompt feature. It provides tailored suggestions based on their products, reducing effort, improving result quality, and sparking inspiration.​
Guided prompt / Ideas for you, launched Dec. 2024
Timeline
2 wks.
Advertiser engagement
29%
3. Building a Scalable Foundation​
​
In addition to launching features that address customer needs, I focused on ensuring a smooth user experience and platform scalability to support rapid growth. To optimize workspace without losing context, I led the transition from a modal to a side sheet experience.
Side sheet migration from modal to side sheet, launched May 2024
Side sheet behavior, responsiveness and internationalization, launched May 2024
Feedback collected from advertisers around UX, July 2024
The Pivot
Entering Q3, adoption grew from 2.5k to 12k—promising, but still falling short. To understand what was holding us back, I dug into data to uncover patterns that interviews weren’t revealing.
I found that while users who accessed the tool had a strong monthly image adoption rate (rising from 23% to 40%), only 10% of the 200k monthly visitors were actually opening the tool. The issue was clear: discoverability was the primary barrier.
Exploring the Advertiser Journey
New Problem Statement
​
How might we enable marketers to easily access quality AI-gen images, so that they can launch engaging ads in minutes?
​
How can users use AI-gen images without going into Image Generator?
User Research
​​
To validate my hypothesis and evaluate usability, I ran a quick user study with 5 participants. This time, I focused the questions on the overall advertiser experience in achieving their marketing goals.
User testing led by me, Aug 2024
Key Learnings
​
Small brands (80% of users) spend minimal time on ad design since they operate as lean teams and with limited design experience. They rely on Amazon to provide ready-to-use recommendations.
Agencies and big brands update their ads regularly based on performance or seasonality.
For better usability, advertisers want access to more image options and the ability to preview images within the context of the ad.
Design for Easy Access
1. Enabling 1-Click Image Selection​
​
To streamline image selection, we introduced a 1-click experience, making the process faster and easier for marketers.
1-click image selection, launched Aug 2024
Timeline
1.5 mo.
Adoption rate
31%
2. Campaign Refresh with Generated Images
​
I took it a step further by exploring areas beyond the campaign creation flow and identified an opportunity to better support agencies and large brands. By proactively surfacing seasonal or holiday image recommendations, we can prompt advertisers to refresh their creatives—improving relevance for shoppers and reducing visual fatigue.
Campaign refresh, launched Nov 2024
Timeline
1 mo.
Engagement rate
23%
Adoption rate
15%
The Results
By the end of 2024, we achieved our adoption goal. Beyond that, campaign submissions increased by 39%, monthly revenue from AI-generated ads grew from $800K to $7M, and product sales saw a 5% uplift.
Feedback collected from advertisers, Nov. 2024
Quotes from Jay at Amazon year-end all-hands meeting, Dec 2024
Coming next
Our ambitious business goals gave us focus and momentum, but over time, the emphasis shifted toward promoting the tool rather than delivering real value to advertisers.
​
For 2025, I proposed a new direction: How might we help advertisers succeed through generated images—rather than simply pushing a tool or assets?
​​
To realign with user needs, I partnered with a researcher to synthesize two years of insights across image generation, ad creation, and campaign building, and proposed a plan to bring this new direction to life.
New proposals by me, Feb 2025