About this project

Indeed is the world's largest job board with 90% of the world's GDP in job postings. Mobile represents about 60% of all job search traffic on Indeed, but actual usage on mobile is very poor. More than 80% of users abandon the mobile app after installation, and fewer than 2% use the app to apply to a job.

My goal was to redesign the app to increase engagement and retention. I was given 5 weeks to lead a complete redesign.

Javascript, jQuery, HTML5, CSS3, Principal, React Native, iOS, Android, IQL, Indeed API

 

My Role

As the Senior Lead UX Designer, I was responsible for creating bold, new concepts for the future of job hunting and sourcing. After the success of previous projects, I was asked to solve Indeed's mobile experience problem with a redesign.

I led the redesign on-site at our San Francisco tech campus with the mobile native team, our user research team in Austin, and our global UX team. Together, we worked in weekly design sprints: assessing problems, pitching solutions, building prototypes, and testing our ideas.

The Design Sprint

A battle-tested process for rapid innovation

The sprint is a four-part process for answering critical business questions through design, prototyping, and testing ideas with real users. Instead of waiting to launch a minimal product to understand if an idea is any good, we got clear data from a realistic prototype each week and iterated on that. This granted us a preview of our finished product and user reactions, before making any expensive commitments.


1. Idea

Each week started with a structured discussion addressing a single challenge and possible solutions from competitors, best practices, and previous research.

2. Build

After refining our ideas, the team would decide which solutions were viable before building a high-fidelity prototype.

3. Launch

With our idea prototyped and ready for launch, I would hand it off to our user research team to present before real users for reactions and feedback.

4. Learn

The sprint comes to a close as we interview users and learn by watching them react to the prototypes. We keep what works, refine, and repeat.

Everyone starts somewhere

Indeed Mobile v1

This was our mobile app before the redesign:

  • 80% of job seekers abandon after opening the app
  • 48% of remaining job seekers attempt to login
  • 53% of guest job seekers performed blank searches
  • 4% of remaining job seekers make a job alert
  • 2% of remaining job seekers apply to a job

Logged-in job seekers had a higher retention due to a personalized experience, but half of job seekers opted to skip this, and job seekers without Indeed accounts didn't have an onboarding flow to personalize their job hunt.

Externally conducted eNPS surveys and internal user interviews revealed that most job seekers abandoned because they didn't just dislike the visual aesthetic of the mobile app, they didn't trust it. Inconsistent UI elements, broken user flows, and unresponsive job postings hurt confidence in already demoralized job seekers.

I worked closely with the PMs to triage these issues against our goals for retention and engagement before starting our design sprints.

Not just a pretty interface

Reducing abandonment by 60%

Nearly 80% of users would abandon the app after opening. User research and eNPS surveys indicated users lost interest when they felt the mobile app offered less functionality than the desktop site or because it was visually displeasing. This meant first impression was critical. The mobile app had to look good and prove unique value quickly.

After cleaning up site architecture, we tested four different design directions. The desktop site lacked consistent design and a style guide, leaving mobile up to interpretation, but I wanted to keep mobile close enough to feel like it belonged to the Indeed brand.

The winning design reduced abandonment from ~80% to nearly 20%. Users trusted the cleaner interface and spent more time exploring the app, allowing us to customize their experience and setup notifications to improve retention.

A Welcome Experience

Successfully onboarding new users

While most users who downloaded the app already had an Indeed account and could benefit from their saved settings, new job seekers were presented with a blank, intimidating experience.

I designed and tested a few different ways to onboard new users to improve their first time experience. Our winning variation required minimal input from the user, scanning the text from their resume and autofilling the rest. At the end, users were presented with a job alert so they wouldn't miss an opportunity, while boosting reengagement.

Job Hunting for Road Warriors

Mobile constraints became our advantage

Decreasing abandonment was not enough, we also needed to improve retention by providing unique value to mobile users. I did this by simplifying functionality and focusing on the transient state of mobile:

  1. Simplified search that autocompleted job titles, companies, and more.
  2. Search results that highlight what job seekers look for in jobs.
  3. Job detail pages that are summarized to improve scanning.
  4. Revealing company responsiveness so job seekers don't waste their time on "dead" postings.
  5. Easy one-click apply and estimates on how long an external application could take.

Rapid Prototyping

How we validated ideas with real users

The key to the design sprint was testing our ideas often. We couldn't rely on static mocks to validate complex interactions, so I produced a framework for rapid prototyping that used live data.

Designs were responsive and loaded onto phones as mobile web apps using the web standard manifest. When running in full-screen, they felt like native apps complete with transitions and micro interactions.

Job Discovery

Observing user behavior leads to a new path for mobile

After a few design sprints, a clear trend was emerging: usage on mobile was poor because job seekers used it for discovery, not applying to jobs. Most job seekers wanted to tune their resume, craft a cover letter, or research the company later on their desktop computer.

What would the app look like if it was designed purely around discovery?

Recommendations would be presented as cards with requirements highlighted using natural language processing for rapid scanning. Swiping right would save for later review, swiping left would provide negative feedback to help refine our machine learning algorithm.