UX/UI DESIGNER at Jobscan
Company size:
10 - 15 employees
Responsibilities
Visual Design, UX Design, Front End Development
Website
How can we help people looking for jobs get past Applicant Tracking Systems large companies use to filter out candidates?
Jobscan was created to address this problem by analyzing candidates resumes, comparing them against job descriptions, and providing a detailed report showing hard skills, soft skills, keywords, and other findings. This puts candidates in a better position to land interviews, and get the job they're looking for. Jobscan is a self-sustaining startup not backed by investors. Currently they're located in Pioneer Square.
The Problem
Applying for jobs is hard. It can feel like when you’re clicking “apply” your resume goes into a black hole, never to be seen again. That black hole is called an Applicant Tracking System.
Companies use Applicant Tracking Systems to handle their recruiting needs. These systems can filter out candidates by searching their resume for keywords, education, years of experience, among many other data points.
Each ATS has its own way of managing candidates, so unless someone knows how a specific ATS works, knowing what to change on a resume for a job is impossible. In general, people don't know enough about applicant tracking systems when applying for jobs online. If they do, there's uncertainty around what to change in their resume to give them the best chance at landing a job.
Background
Jobscan has researched the top ATS systems used by thousands of companies, and built an algorithm based on common patterns among them. Using this algorithm, Jobscan recommends users to make changes to their resume/cover letter based on a provided job description.
How Jobscan works
- Upload your Resume
- Paste a Job Description
- Scan to see a detailed Report of how your resume compares
Design Process
RESEARCH
Career Coach Interviews
Career Coaches assist jobseekers during their job application journey. They help with anything from formatting their resume to deciding on which industry they should focus on in their job search. Anything from career exploration, to career changes, to career development.
Jobscan has features built specifically for career coaches. Coaches are given the ability to manage clients, track progress, and scan their clients resumes. There were new Career Coach features to roll out, so reaching out to existing Jobscan Coaches was the first step in the design process. I started by emailing current career coaches, creating an interview guide, and scheduling interviews.
The interview guide was structured around understanding Career Coaches attitudes towards the current experience and features, pain points they faced, and functionality they would like to see in future versions of the product. I documented findings from each interview, analyzed and categorized the raw data, then turned the insights into design and engineering work items in github.
Target User Behavior
I analyzed user behavior by studying recordings and heatmaps of users on Hotjar, an excellent user research tool. By watching recordings of users accomplishing tasks, I was able to compare the recording findings against my UX Audit findings, and validate usability bugs in the product.
Hotjar's allows designers to view which pages users visit, where their mouse is, and highlights moments where users click to interact with elements all in a timeline. I also used scroll heatmaps set up on specific pages to determine what elements on a page were getting higher visibility than others.
I used Google analytics to study demographic information about the users of the site. I was able to determine the mobile platform wasn't where a large number of users were interacting with the site, so we prioritized the desktop experience. Google Analytics was especially helpful in seeing what pages were the most highly trafficked, and how those users were initially coming to Jobscan. Analytics helped see what pages and flows had the highest conversion rates, and how long it took users to convert.
Mixpanel was one of my favorite tools I used. It was especially helpful in identifying conversion rates for specific flows, so the team could focus on prioritizing specific features. I looked at numbers for different funnels, and measured against my UX Audit findings to address the most pressing usability issues.
Focus Group with Recruiters
The first step of analyzing the data I did by organizing all user feedback into a document, and categorizing by naturally occuring categories. For example, say a high number of recruiters talked about wanting a tool that allowed them to analyze a person's past experience that isn't directly related to their company. I would add all relevant data from the focus group into this category. Then I would name it "Valuable Candidate Experience" or something similar.
Synthesis is a bit more difficult. After synthesizing data, a team should emerge with design requirements. Changes to the taxonomy or visual design of the site could follow. Synthesizing the data should be done as a team, or at the least validated by the team. Going off the above example, first addressing whether "Valuable Candidate Experience" is a problem that warrants solving is the first question. A designer and the team should have a list of questions they evaluate research findings against to determine which they address. Does it fit in the product? How high priority is this change? Etc.
Regarding design requirements, it can be a simple color change in a button to change heirarchy. Or it could be an entirely new feature that warrants its own . Research findings can manifest themselves in features and changes that can take a variety of forms with respect to the constraints of that product. Its important to discuss how and what the feature should accomplish in a product before a designer comes up with mockups or prototypes.
A/B Testing
Jobscan is big on having data inform product decisions, and using A/B testing was a great way of quickly testing different versions of a design or text to determine how other variables like conversion rate are affected.
I assisted planning a focus group where we invited recruiters in to talk about their experiences, needs, and wants in the industry. I also documented insights during the focus group, and afterwards turned those quotes from the recruiters into design requirements. Turning raw research data into requirements is a two step process of first analyzing, then synthesizing.
Many recruiters wanted more information about who their candidate was. To have a more humanized approach
to recruiting that didn't feel so robotic.
DESIGN FEATURES
Variations
Problem
When users view a report after scanning their resume, some variations of words aren’t recognized.
For example "design" may be recognized as a skill, but "designers” and "designed" isn’t. This leads to people questioning the credibility of Jobscan. Based on the feedback we were receiving, it was in users mental models that variations of words should automatically be recognized. Implementing this feature would drastically improve the credibility of Jobscan, and also decrease the dropoff rate of new users.
Process
I started by meeting with team members who could give more context around the problem, people from engineering to management. We discussed user feedback, and framed the current problem users were facing, and discussed ideal solutions. It was important that we understand the problem from users perspectives, so we could understand what their idea of a solution would be. The issue is complex, because the number of variations of a word changes with every word. How do we display word variations logically to a user within the constraints of the current system?
I brainstormed and sketched low fidelity ideas of what different interactions could look like. From these ideas, I worked my way up to higher fidelity mockups made in Adobe Illustrator. I used the approach of having multiple alternatives. I like the ability to evaluate on prototype against the other, it creates a spectrum which makes it easier for a designer and the team to compare pros and cons of each design. It also surfaces more insights from brainstorming, and doesnt force a designer to put all their ideas into one solution.
After rounds of feedback, comparing against initial goals, and refining pixels, I presented the final design and we decided to move forward with engineering. Everything from the icon, to the way words appear under the icon was discussed at length with the team. The engineering team did an excellent job building the logic behind recognizing variations of different words. The current design is live on all reports.
Recruiter Findings
Problem
After the focus group with recruiters, we had lots of insights we wanted to act on.
For jobseekers, having some insight as to what recruiters are looking for in their candidates gives them a big edge on other jobseekers. The team decided we needed some way of making jobseekers feel like they were getting instant feedback from a recruiter. What type of data would be most valuable to show jobseekers, and how should we contextualize it so they understand?
Process
I started by talking with my coworkers who’d started this project to get context. Education and industry breadth stood out as two areas members of the focus group highlighted. From a technical standpoint, we could add these two to the scan report, so we chose to run with those.
By highlighting good experience jobseekers have on their resume, it gives users a pat on the back. The "Industry Breadth" section highlights potentially applicable skills a jobseeker has on their resume. Although the jobseeker may not consider them relevant, they may give recruiter a better sense of what that persons brings to the table. Recruiter Findings adds a personal touch to the job application process, where its easy to feel like your resume is lost in a black hole.
Live Score
Problem
Currently, users are shown a report where they see a list of words and phrases Jobscan recommends they change on their resume.
How can we improve the resume editing process, and make it easier for users to increase their score?
Process
Working with engineering, I took an iterative approach to designing what a live resume-editing interface would look like. Users needed to understand they could edit their resume here, and see skills they needed to add to their resume. The Match Rate Score is updated in the header section. As users made changes, the score increased or decreased accordingly. Users could save their progress in Live Score, so they can come back later. Formatting options were also available. I created high fidelity mockups at different stages of the process after receiving feedback.
Notifications
Problem
Users have no way to see updates to their profile, or changes to the system in general. Jobscan also doesn't have a way of notifying users of new updates, features, and changes to the system. Perviously, users were getting updates through email, or just seeing changes on the site without any warning.
How should we design a useful notifications system within the constraints of the current platform?
Process
Notifications had been a long time coming. It was a large feature that required rounds of feedback and design changes. The project started with a mockup I created, and showed to the team as a proof of concept. From there, I began working with engineering on the specifics of the design.
I had to considering all states of the notifications bar
- What does the UI look like when there are no notifications and a user clicks the bell icon?
- What should the maximum number of unread notifications be?
- What should the maximum height of the notification card be?
- What constitutes a "read" notification?
- Should notifications have their own page?
These are a few of the questions we asked when ensuring we were designing for all possible use cases.
FRONT END DEVELOPMENT
A portion of my work was making changes to the front end code of Jobscan's platform. I'm skilled with HTML, JavaScript, and CSS so this was a fairly easy task for me. I would often file work items that needed to be taken care of, and if engineering was busy on larger work items, I could help take some work off of their plate.
UX AUDIT
LEARNINGS
One of my other first tasks was to do a UX audit on Jobscan's website. My process was to first create a page in a document for each page on the website. I then measured each page against Nielsen's 10 usability heuristics, making a note when a convention or interaction violated one of the principles. I also documented my findings studying recordings in Hotjar or users performing specific tasks like scanning a resume or connecting their LinkedIn profile to their account. All this data I documented, and once the audit was complete, I referenced the audit findings with user research findings from interviews, survey data, and previous research done to surface the most important issues.
Findings in the UX Audit involved visual design changes, as well as UX issues. I attempted to focus less on visual design issues, because this can be an area thats more subjective. Usability issues were the primary focus. In the process I also looked at successful websites like AirBnB, Sketch (sketch.app), and Dropbox for examples of what good design looks like.
I learned how to approach creating new features, or additions to preexisting features. I naturally wanted to create something new, dynamic, and better than the existing version of Jobscan. But usually, a company doesn't have the time or resources to do so. I believe Designing within the constraints of an existing feature is an important skill for a good designer to have.
This means understanding the structure, heirarchy, and functionality of a feature or interface.
Its important to be careful when adding layers of complexity to existing features. Its very easy to turn something users understand into a confusing workflow, just by changing one small thing. Anytime a systems UI is updated, a company risks confusing users, and even losing some. This leads to my last learning,
Validating a design when you can, even with a few target users, can save you hours of time, and help you see a perspective you may never have seen otherwise.
