MAY'18-SEP'18

Personalization on Groupon

Team: I was the lead designer for both iOS and Android. Worked with 1 UX researcher, 1 content strategist, 1 product manager, and 4 front-end engineers.














About Groupon

Groupon’s mission is to be the daily habit in local commerce. It provides an online marketplace that connects consumers with local merchants who sell deals on activities, services, travel and goods.






PROBLEM

As a new user, I want to find deals that are of interest to me but I keep seeing random deals in the feed. I feel like the app doesn’t know me.
- Groupon User




Problem Highlights

• Personalization for a Groupon customer was limited to implicit user behaviors which lead to a cold start for new users.

• Purchase frequency of new users significantly lower than active users

• Among new users, 2 in 3 users didn’t make a first mobile purchase in 90 days




HOW MIGHT WE

How might we elevate the experience of new Groupon users to be more personalized to their interests?






In-Feed Personalization

As a designer on the discovery team, I led the end to end design of this personalization feature on both iOS and Android. This card provides users with options that can help Groupon understand their interests and accordingly personalize their deal feed. As a result, improving their discovery experience. I am going to discuss the design details and rationales below.




In-feed card

We designed an in-feed card with Groupon’s Deal Categories - Activities, Services, Food etc. Selection of these categories enabled the user to tell Groupon what they prefer to see more of in their feed. We wanted the card to be placed below a few deals in the feed so that it doesn’t come in the way of the user’s natural browse behavior. When the new user begins browsing after signing up and finds the feed to be random, it immediately make sense to them to personalize their feed.



Progressive Disclosure

In addition to the value prop message, the personalization card shows a few deal categories by default so that the user immediately understands what the card is about. It also uses illustration to visually stand out from other cards. Upon interaction with the category pills, the illustration and the value prop messaging go away revealing more categories. Progressive disclosure of categories reveals the right information at the right time and does not overwhelm the user.















Easy Navigation

The user can easily navigate to other categories using the carousel or the CTA at the bottom. Having the categories spread out doesn’t overwhelm the user with all the options at once.






Instant Gratification

As soon as the user submits their preferences, Groupon’s algorithm gets to working and presents the user with a fresh feed with personalized deals in a matter of seconds. This was challenging, particularly from technical implementation perspective. But the team rallied behind because the users strongly expressed the need to instantly see the result of their action during the usability study.





OTHER EXPLORATIONS





RESULTS AND LEARNINGS

RESULTS

The feature was shipped and is currently experimenting. The primary success criteria for this project is increase in Customer Life Time Value(CLTV). We are measuring conversion and engagement impact over long durations.

Based on some early data analysis we see two big engagement insights:
• The card is seeing slightly lower than expected engagement with a click through rate of 1.8%
• We are seeing a drop off of about 49% from the first category to user completing the flow submitting their preferences.

We are using these two insights to explore and iterate on our designs for the next version.

LEARNINGS

User Research really helps with convincing stakeholders: It is important to explore designs rooted in user research insights and data. User research insights can really help to get commitments and alignment on complex engineering tasks which otherwise is left to subjective opinions.

Explorations never go to waste: During this project, we did more than usual explorations. Not only that made our decision making more in-depth, it also gave us a head start to work on the next iteration of this feature based on data analysis.

PROCESS



Formative Research
Worked with the UX researcher to conduct interviews to understand users' browsing behavior on Groupon. This is when we observed new users experiencing this problem.

Problem Analysis
Having heard about this problem first hand from users we started investigating further by looking into data and purchase behaviors for new users. This helped us gather more evidence and validate the problem.

Iterative Design
Next, there were whiteboarding sessions and design critique at multiple forums with design, product and engineering teams.

Usability Study
Once we had narrowed down to a couple of approaches, I prototyped the designs and worked with the UX researcher to conduct usability testing to get user feedback on the designs./h4>

Shipping
Worked with engineers to build out the feature. I also went through two rounds of design QA to make sure the feature works as expected and looks pixel perfect.

A/B Testing & Analysis
The test is currently running and it will take months to know the impact on revenue and Customer Life Time Value. We are analyzing engagement data to get insights on user behavior.







OTHER PROJECTS


MAY'16-AUG'16

Kindle Author Services

Kindle Author Services helps independant authors find and connect with professionals offering creative services that help them bring a book to life from the manuscript stage.

See Project
FEB'15-MAY'15

BUYER CONNECT VISION

Buyer Connect is a buyer and seller communication feature for Flipkart's Seller Platform. It helps sellers to answer buyer queries in a fast and efficient way and thus help them reduce returns and incease orders.

See Project
JUN'18-OCT'18

Groupon Desktop Homepage

Designed an experience for Groupon’s desktop homepage to make it organized and guide users to find something of interest. In the process also, made the homepage to be full-width and adapt to the users’ screen size.

See Project