Project Overview
Project type: Strategy, Framework, Visual Design
Role: Lead Designer, Ads/Recommendations Team.
Timeline: Q4 2019 - Q3 2020
Partners: Engineering, Research, Search Team, Home Page Team and Design Systems
Platforms: All platforms - iOS, Android, desktop web, mobile web.
Status: Complete
Outcome: Implemented a robust item tile framework that has led to higher buyer engagement. Personalized and customized recommendations that improved the relevance and discoverability of items.
~$345M
Ads Revenue in 2021
>$1M
Ads Revenue per day
$7.4M
Inc in Ads Revenue in 2021
Project Goal
Create a Dynamic Item Card (DIC) framework that adapts to buyers' needs and journeys as well as being customized to different purchasing categories.
Surface just the right information at the right time in order to provide the next best item/action for buyers.
Generate and test numerous creative variants automatically and simultaneously; optimize to drive the most engagement and conversion.
Problem Space
The eBay buyer base is diverse - casual shoppers, collectors, investors, bidders, etc. People come to eBay to buy and sell items, access a large selection of products, find unique and rare items, and to get good deals. Creative visual elements such as item cards and ad banners convey key messages and make first impressions of the products and items that are sold on eBay. These creative elements also aid buyers in finding what they are looking for. Back in 2019, eBay had a "one-size-fits-all" creative solution, which did not meet buyers' diversified needs; did not take into account where buyers are in their shopping journey. Additionally, static creatives make it difficult for sellers to communicate with and engage buyers.
The lack of standardized format for item cards on eBay, resulted in varying visual styles, content and data that did not adapt to context/category.
Item cards had
15+ layout / platform combinations
15+ different categories or types of information
60+ distinct fields
The inconsistent language and styling caused confusion and mistrust among buyers.
Item cards across eBay have variations on text color, icons, font size, language, different signals to indicate the same thing. Screenshots from 2019
Discovery
I created a massive inventory of existing signals/data that appeared on item tiles by collaborating with Structured Data (backend engineers). Collaborated with my Product Partner to create a Discovery deck to identify opportunities and strategy the DIC project.
We leveraged numerous qualitative studies shared by eBay’s UX Research Organization to gain insights into buyer needs related to recommendations and shopping journeys by categories.
Spreadsheet of Item tile data, grouped into categories.
Slide from Discovery Deck focussing on competitors’ item tiles such as Amazon, Etsy, Google Shopping.
Research Slide Deck with insights on buyer needs
Creating a new framework
I collaborated with cross-functional partners - such as PMs and engineers, to establish a preliminary framework that incorporated a better hierarchy of signals on an item tile. We created a Product Requirements Document (PRD) to list rules and requirements of signals.
The process also involved looking at item tiles from competitor sites - such as Amazon, Airbnb, Google Shopping, etc.
This building block became the foundation on which different variants were built.
-
Figma Design Mock - Item tile framework & list of potential signals under each slot
-
Screenshot of online working session between Designers and PM, co-creating variants for item tiles
Design Workshop
5 teams met in Portland for a 3 day workshop to align on the new item tile framework. This included Designers, PMs, engineers and content strategists.
We chose 4 major areas to work on - Price, Shipping, Aspect details and signals.
The end result was the Item Tile consistency playbook that illustrated guidance on visual style and language on item tile signals. This formed the bedrock of the revised item tile framework across eBay.
Photo of participants from the design workshop at Portland
Screenshot of Slide illustrating different ways eBay displays discounted price on Search, Item listing page and on Recommendation item tiles
Screenshot of Slide with item tiles on Recommendation Strips after incorporating visual and textual changes.
Design
For Phase 1, I iterated to create multiple variants (MAB/Multi-arm bandit test) for Ads/Recommendations based on the framework agreed upon. Variations were intended to balance a large amount of content with succinct item tiles.
In collaboration with front-end and back-end engineers, we reduced the number of variants from nine to five variants based on their implementationability.
Screenshot of Figma deliverable of 5 variants for iOS.
Screenshot of Figma deliverable of item tiles listing each signal for inspecting.
Research & Multi-Armed Bandit (MAB) Testing
Conducted a unmoderated research study to understand how well the variants tested with users
Goals:
Test how well different designs of item cards enable buyers to quickly scan and understand the content of a card and take action.
Understand buyers’ reactions to and preferences for different item card variants.
Learn about which specific pieces of information buyers’ prefer to be displayed on item cards.
Learnings from Qualitative study:
Majority (63%) of participants preferred the Full design; they liked having all available information
Scannability:
The Full and Middle designs both appear to be easily scannable, on par or better than the Control design.
The Minimal design seems to be easy to scan but made it difficult to complete the task due to an absence of vital information on the item card (color, title).
Item information preferences:
Price, photos, title and condition were viewed as universal must haves, while shipping cost and shipping time were viewed as universal nice to haves.
Learnings from Quantitative Study:
Full variant - winner on dWeb, middle variant - winner on iOS, No title - variant did not perform well.
There was a 7% lift in Recommendations initiated sales across the site attributed to the test.*
*Numbers are modified
Results & Outcomes
~$345M
Ads Revenue
in 2021
>$1M
Ads Revenue
per day
$7.4M
Inc in Ads Revenue in 2021
The Cross-team collaborations lead to DIC Variants successfully launching on multiple pages such Product Listing Page, View Order Details page and Check Out Success Page in all major markets (US,EU,DE,AUS).
We were able to generate and test over 100 creative variants automatically and simultaneously; optimize the designs to drive the most engagement and conversion.
Enabling relevant recommendations at the right place, helped users shop more quickly and efficiently. Helped buyers re-engage with abandoned shopping journey.
This cutting-edge framework laid the groundwork for highly tailored and customized recommendations that will shape the future of verticals like Fashion, Parts & Accessories on eBay.
Anusha’s impact on the project
-
Anusha's main contribution was redesigning our merchandizing strips from the ground up while also considering our new dynamic item card data format. The impact of this initiative is vast in that these designs have a 1:1 relationship with our revenue and span across eBay. Anusha is at the helm of rethinking how users interact with our site which will be highly impactful. Anusha is amazing to work with. Her attention to detail and ability to understand highly technical concepts is second to none. Because she is so technical and detail oriented, her designs really shine in being user friendly, scalable and modular. Anusha does not hesitate to iterate on tough designs or start over and look from a different vantage point to present multiple points of view.
- Engineering Lead
-
Anusha is extremely purposeful in her work. All of designs and proposals have intent to solve particular design challenges. Additionally, she sets the bar high which takes courage especially in a group setting. If she thinks something doesn’t solve a particular problem as well as it could, she vocalizes this well and constructively. This also takes drive and inventiveness. It’s complicated to reimagine how recommendations are presented to users because there is such as pervasive status quo. Cumulatively, these qualities make Anusha our brand and richly diverse.
- Design Manager
-
Anusha is an active and vocal design partner for the scrum team. Her attention to detail, strong UX mindset, responding and adapting to ambiguity, working across different teams, resilience, thought-partnership, energy made a huge impact on DIC. She is one of best partners! passionate about UX, very user centric, detailed oriented, data-driven, and with many great/fresh design ideas. She lead a great presentation for leadership on Dynamic Item Card, handled the day-to-day DIC project with ease.
- Product Manager