Indeed case study — Charles David Olivar

01Overview

I interned at Indeed as a UX design intern. I worked with my mentor Michael Swingler on designing user input screens for Indeed’s mobile app, and I got to watch Indeed’s design system grow over the summer.

Goal: increase the number of “applies” (when someone taps Apply) on mobile.

The bet behind my project had two parts:

Hypothesis: more data about what a job seeker wants leads to more relevant jobs, and more relevant jobs lead to more applies.
Idea: a short series of screens that collect that data and help Indeed match job seekers to jobs they will be interested in.

Exploring that idea was my project for the summer.

02Research

A whiteboard of handwritten notes listing goals, success measures, opportunities and a six-week timeline with usability tests — Planning on the whiteboard with my mentor and PM: goals, what success looks like, and a week-by-week plan with testing in it.

I started by learning Indeed’s products, talking with my mentor and PM, and reading existing research, including Indeed’s personas. I can’t share Indeed’s personas or research here, but with them in mind I focused on three cases:

New users
Currently active users
Users coming back after a while away

Competitive analysis

A phone folder holding seven job search apps: Monster, Snag, Glassdoor, LinkedIn Jobs, Job Search, Indeed Jobs and Job Spotter — The job search apps I compared.

I went through related job search apps and paid attention to the questions and inputs they used across the job seeker’s journey. I no longer have the write-up of that research, since it stayed with Indeed after my internship.

Synthesis

From Indeed’s data and personas plus what competitors do, these looked like the right topics to ask job seekers about during onboarding:

Job type (full time, part time and so on)
Salary
Location
Company
Experience level
Where they are in their job search
Urgency
Lifestyle preferences

Some of these were already just query filters on Indeed’s products. The new idea was to store them on a job seeker’s profile to improve future results and suggestions.

03Ideation

With those topics in mind, I sketched screens for each kind of input.

Pages of pencil sketches of onboarding flows: account creation, job title, location and distance questions, and a recommendations list, with margin notes and crossed-out ideas — My sketches, crossed-out ideas and notes to self included.

Initial mockups

I digitized the sketches and met regularly with my mentor and PM to get feedback on the screens.

Grayscale mid-fidelity mockups: a recommendations screen with categories to unlock, and question screens for skills, ranking what matters in a job, location, commute time, job categories and job titles — Mid-fi mockups. The recommendations screen unlocks more categories as you answer more questions.

User flow

I used Whimsical to map the paths a job seeker could take through the onboarding screens to a page of relevant job recommendations. You can look at the flow in Whimsical.

A large flowchart of onboarding screens branching by location, culture fit, room to grow and fit for the job, all ending at recommendations — The flow. Lots of ways in, and they all end at recommendations.

Interaction prototypes

I also explored the interactions for a couple of the screens.

Choosing what matters in a job.

Selecting up to five choices.

Upping the fidelity

Then I added color and visual design to the mockups, and we narrowed the set to the screens we wanted to prototype and test.

High-fidelity mockups with color and icons: where you would like to work on a map, sliders for what matters most at work, skills, job titles and job types, each with a Save button — Hi-fi screens, cut down to the ones we’d test.

04Prototype and testing

The Framer prototype

I learned Framer for this project and it became my favorite prototyping tool: more interactive than InVision, more flexible and scalable than Principle, and faster than building it in HTML, CSS and JavaScript.

The Framer editor showing prototype code that uses Mapbox, next to a phone preview with Skills, Lifestyle, Job Type, Location and Titles tiles — The prototype in Framer. The home screen is just a placeholder so testers could get to each screen.

The location screen was the most fun to build. I used the Mapbox API for an interactive map and added autocomplete to the input, so testers used a real interaction instead of a fixed answer.

The prototype I tested was hosted on Framer’s cloud, and it is no longer online.

User testing

I worked with UX researcher Georgia Gallavin to test the prototype through UserTesting.com with 8 people on their own phones.

What we wanted to learn

Do job seekers see the value of the onboarding, and is it clear how it helps their search?
Are the categories relevant and desirable? What is missing or confusing?
Can job seekers finish the exercises easily? Where does it hurt?

7 of 8 testers overlooked the control for reviewing their selected job titles.

Top takeaways

Overall, job seekers understood the exercises as the first step in matching them with relevant jobs.
They expected to enter salary and experience level during onboarding.
Creating a skill from scratch, instead of picking from the list, was unclear. Most did not notice the show/hide control for their selected job titles.

More detail

One tester preferred to search for jobs first instead of entering preferences before seeing recommendations.
Lifestyle criteria felt novel, but testers struggled to see how they would map to relevant jobs.
Testers treated Skills as a place to pick from an existing list using the search box.
They added a location several ways: typing a zip code, using the autosuggest dropdown, or choosing their current location.
They often looked for a way to cover a broader area on the map, to show how far they would travel.
Most typed an industry or job title straight into the search box instead of browsing categories.

05Moving forward

Iterate

I took the feedback from testing into the final designs, and worked with content strategist Michael Lawrence on the copy for each screen.

Eight final screens: job types, commute time on a map, location, skills, skill search with the keyboard, and job titles with Show selected and Browse jobs toggles — Final designs after testing and copy edits.

Next steps

The user flow for the whole mobile app was still being worked out by my mentor’s team, so it was not certain where my screens would end up. Indeed was moving toward a native version of the app, and I hoped some of these ideas would show up in it.

One idea I had: let a call to action for each screen float among the job feed, so job seekers can fill in their information when they have a minute instead of all in one sitting.

Wireframes of a generic page and search results with onboarding prompt cards among gray job cards, such as Where do you want to work, with progress like 4/5 questions answered — Prompt cards in the feed with a progress count, so people can answer one question at a time.

06Takeaways

Think ahead, then scale back

We created and tested ideas that might not be possible with Indeed’s current data and technology. That was deliberate: it let us explore, so Indeed could later apply the insights, while engineers took what was practical now.

Mentorship matters

My mentor has 20 years in industry, and his art background helped me, as a student whose main education was in computer science. He sat across from me every day and was glad to answer my questions, which made the internship a good one and pushed me to grow as a designer.

Eight interns sitting in a row on the floor in front of a painted Indeed mural, smiling — Interns at the Indeed San Francisco office.

Want to read another? All design work →