Arthur Alexander

Maecenas non leo laoreet, condimentum lorem nec, vulputate massa. Fusce at massa nec sapien auctor gravida in in tellus. Mauris id fermentum nulla.

Donna Johnson

Sed a ligula quis sapien lacinia egestas. Fusce at massa nec sapien auctor gravida in in tellus. Vestibulum ante ipsum primis in faucibus orci luctus et ultrices posuere cubilia Curae.

Sara Zhao

Fusce at massa nec sapien auctor gravida in in tellus. Vestibulum ante ipsum primis in faucibus orci luctus et ultrices posuere cubilia Curae. Nulla eu pretium massa.

Adding Search Filters: Increasing Engagements and Conversion

Screen Shot 2021-10-17 at 5.52.14 PM.png — Job Search

Upwork’s work marketplace connects independent talent and clients to get work done. Job search is a core experience for the marketplace, giving talent the controls to find the jobs that are the right fit for them.

Making filters more discoverable and introducing new ones to give users more control over their search results

Overview

Responsive Web Design

Timeline

April - September 2020

Team

1 Product Designer (me)

1 User Researcher

2 Product Managers

4 Front-end Developers

1 Analyst

1 User Problem, 2 Objectives

As a freelancer, I don't trust clients to categorize their jobs correctly or for Upwork to present relevant results in search, so I spend a lot of time searching broadly which makes it hard for me to find jobs to apply to.

In 2018, Upwork started building out a new ontology of work for the platform to create better definitions and relationships on concepts in the world of work. In creating this new data set and driving adoption through job posts and freelancer profiles, we opened up a new world of how clients and freelancers could classify and thus find each other in a more structured way.

When it came to our existing job search experience, we'd heard from freelancers that it was hard to find jobs that were the right fit for them. Clients often did not self-select what freelancers believed were the right categories to classify their work and Upwork’s search algorithms were not always providing relevant results. But with the new ontology and new algorithms powered by it, Upwork was able to help clients select the correct categories and surface the right results.

**Learned behaviors to compensate for a confusing system & UI**

Filters hidden behind a button making them hard to discover
Broad keyword searches with very little filter usage so they don’t exclude a potential job
Refreshing the page often and little pagination because users perceived that most recent is most likely to be won

How might we enable freelancers to more easily narrow down their search results so they can find jobs that meet their skills and needs faster?

Make facet filters more easily discoverable and accessible
Add new Ontology filters to allow freelancers to have more transparency of Upwork’s classification system and enable them to navigate freely through available jobs by category

Exploring Design Treatments

Many search experiences in competitive and comparative products introduced exposed filters to encourage usage, recognition over recall at its finest.

Concept Testing

Users told us that multi-selecting categories at all levels was visually too much, but no multi-selecting didn’t allow for the desired breadth of results.

We conducted a 3 Cell Test to better learn what change is most important

Treatment 1 only tested changing the layout, keeping the same filters

Treatment 2 tested changing the layout & introducing new Category filters

Based on past tests, we didn’t want to make too many changes at once and not be able to identify what change is moving the metrics. With Control in 1 cell, Treatment 1 tested only changing the layout of filters and Treatment 2 tested layout and introduction of new filters.

Impact

Treatment 1 increased filter usage by 22% & allocate-to-hire metrics by 6%

Category filters didn’t have the expected impact but pulling filters out of the drawer opened up the ability to easily test new filters.