Back to All Products

Groupon Filters in Discovery

Developing the right tools for a better discovery experience in a growing marketplace platform
Client: Groupon
Role: Lead Designer


We developed and tested two efforts to improve users' refinement abilities within result sets. Both projects focussed on discoverability of the filter controls, making it easier to refine as well as discover new types of inventory.
Delivered: Problem & Opportunity Analysis - UI & UX - User Testing

Giving Users Control

Result sets display a subset of inventory relevant to a users expressed intent through searching or browsing. There is often a need however to narrow down offers even further. Giving the users the right controls to do so helps in uncovering exactly what they are looking for.

To help users understand what is available, we display our filters exposed within the page, rather than hidden within a filter sheet. From this baseline I worked on two projects that addressed existing performance as well as new initiatives. I've outlined each in more detail below.

Exposed Filters in Discovery

Dynamic Filters

From the data of existing performance we were able to draw a few conclusions. From these we developed goals for improvements.
In broader contexts like browsing we saw filters helping users to refine and make a purchase. In these scenarios we wanted to increase engagement with the controls. In more specific contexts like search, we saw filters distract from the already relevant result sets and decrease overall orders. In these scenarios, getting out of the way would help users evaluate the available offers.

Our resulting objective was to propose a single filter system that could adapt to accomplish both goals based on the context that they appeared in. A single system would help maintain familiarity in interaction, but could be customized to be effective in a variety of use cases.

To increase engagement, we knew we could be even smarter about the controls we offered. By understanding the relevance of specific filter choices based on the context, we could make better recommendations of controls to use. Exposing facets created easy and low commitment actions users could take to narrow down their results, while still allowing further customization in the filter menu for users desiring more control. These suggestions then would change based on what the user was looking for, but behave in a familiar way across contexts.

In situations where we wanted to limit distractions, the filter recommendations could be removed, while still leaving behind familiar access to filter controls.

While developing models for this approach, we evaluated usability through 12 user study lab sessions, using the RITE method to adapt the design & test plan after every 3 sessions. This helped us to get feedback across multiple models and ultimately align on one model that exceeded the others in usability.

Extending Controls

With a focus on becoming a daily habit in local commerce, we started integrating new inventory types into Groupon beyond our traditional vouchers. Some of our first integrations included bookable beauty appointments, restaurant delivery and pickup and even golf tee time bookings. With an eye toward the future of our broadening marketplace - we explored ways to unify the discovery experience as a more scalable alternative to dedicated experiences for each inventory type.
We needed to introduce entry points for users to engage with these unique offerings, as well as capture additional details that came along with them. Most new offers required some additional values like time, location, service, etc. - specific to only that offer type.

We integrated the entry points into our existing filter controls users were already familiar with. From these entry points, we developed a system that revealed additional controls to capture necessary values and persisted criteria as users scrolled and scanned the page.

The system could be expanded across other inventory types to discover and then express criteria in a consistent and familiar way.

To verify the usability of the controls, we ran a small series of user research lab sessions with our bookable beauty appointments. Users discovered and activated the control, declared their time and service, and then browsed the results for availability. Again, we used a RITE method to iterate on the designs between each day of sessions. We refined details such as the entry point control, the directional content as well as the method of information persistence.