Rethinking Recruiting

The problem

Applying for jobs sucks.

We’ve all been there. You spend hours upon hours, typically across multiple months, applying to a seemingly endless stream of jobs - writing cover letters, modifying your resume, tweaking your LinkedIn profile, filling out online forms, searching for referrals - only to get rejected from almost every one that you apply to. It’s disheartening.

Websites like LinkedIn and Indeed are supposed to help, giving you personalized recommendations and allowing you to set up alerts on relevant postings, but oftentimes these aren’t accurate enough. I can’t tell you how many times I got sent job recommendations for things like “head of social media marketing” and “senior analyst” when I was looking for early-level career postings pertaining to market research. And even the suggestions that were a good fit for me often lead to complete dead ends. It seemed hopeless. And it wasn’t just me. Even my aunt, who has 30 years of professional experience, including management, said she had spent several months applying to nearly 100 jobs before she found the one.

That got me thinking - why is the job hunt such a shot in the dark? Of the many aspects of human life that technology has improved, it seems ridiculous that we still spend so much time and energy on a process that almost everyone must go through, with a very low success rate - according to the site Hire Lehigh, it takes a professional between 100 and 200 job applications to get one offer. This is a huge waste of time and resource - in the time that professional spent applying to 99 of those jobs, he/she could have already been contributing to the role they would ultimately end up in.

Knowing how many applications one professional must send in to land a gig, there must also be a large influx of applications for each job posting that recruiters ultimately have to weave through. Some advances in automated resume readers have helped this, but these tools end up widening the gap for people from diverse backgrounds, who these systems commonly discriminate against.

There had to be a way of improving and simplifying the job process, for both the applicant and the recruiter. That’s what led me to my concept redesign of the online job application portal, which I am calling Rethinking Recruiting - or “RR” for short.

How it works

Uses machine learning to understand who is most likely to get the job
Curates applicants’ suggestions to show jobs where, based on their information, they have the highest chances of success
Decreases the amount of applicants recruiters must review for each listing
Calls out potential biases in a company’s recruitment against people from diverse backgrounds

The thought behind RR is that it will use machine learning to better curate the job application process, for both the job seeker and the recruiter. RR will take user data - people’s educational background, years of experience, career field, certifications, demographics, etc - along with their current roles and hopeful roles and “learn” who is likely to get what jobs. When people apply and get accepted to or rejected from jobs, RR will start compiling any commonalities among these applicants to create profile information on who each role or each company is hiring.

For example, RR learns that company X rejects 95% of people who have less than a bachelor’s degree, so RR will no longer suggest company X roles to those who have not graduated from college (and are not about to graduate). Or, RR learns that 88% of people who got hired for the role “senior business analyst” have at least 5 years of experience in a business analyst role, and 98% have a bachelor’s degree. RR will now suggest Company Y’s senior business analyst posting to those with a college degree and 5+ years of experience. Perhaps RR learns that Company Z’s frequently posted “entry-level software developer” job almost always hires people 2 or less years out of college with a degree in either computer science, computer engineering, or IT. RR now knows to only show this listing to people who fit those two criteria. Maybe RR notices that Company Z also hires ex-convicts while Company Y never does - that information would change who is shown listings from each company, as well.

The goal of RR is to save time for both parties involved in the process. With this process, applicants get fewer recommendations, but each one is tailored with all of their background credentials and demographics in mind, and the likelihood that they will get an offer is much higher. Recruiters who use RR will receive less applicants, but each one has a high likelihood of fitting their typical hire demographics.

Of course, RR isn’t for everyone. It isn’t for the “diamond in the rough” individual who, without a college degree, was able to climb the ladder of a small startup-turned-corporation as a sales leader, and is now searching for an equivalent sales manager position at a new company. It isn’t for those niche cases where someone may not fit the typical criteria, but has such a compelling experience that they get the job anyway.

But RR is for the typical professional who is tired of wasting time on fruitless job applications, or the overworked recruiter who doesn’t want to read through 150 resumes only to toss out two-thirds of them for not meeting basic conditions.

RR can also help companies see when their recruitment policies and practices are creating undiverse organizations, and work to fix them. For example, RR catches that Company X rejects 25% more women than men, even when their backgrounds are equivalent. Or RR reports that Company Y turns down almost all applicants from historically black colleges and universities, while not turning down as many applicants from an equivalent state school. Having the clear-cut statistics that only a system like RR could find would help companies bridge these previously unseen gaps in their inclusion efforts.

Asking the right questions

The idea seemed like a winner, but I have to be sure that there is a broader use case for RR before I actually implement it. To design a good product, I need to draw from my target audience. These are the types of questions I would ask to get a sense for their wants, needs, and gaps:

How do you feel when applying for a new job?
What are your top pain points when applying for a new job?
When looking for a new job, what resources do you typically turn to?
What are the top websites or online sources you typically use when searching for jobs? Please list up to three.
What do you like about each of those? Why is X your top choice?
Is there anything you dislike about these websites?
How many job applications do you typically submit during your job search?
When attempting to get a new job, how many hours a week do you typically spend searching for jobs to apply to online?
When attempting to get a new job, how many hours a week do you typically spend writing materials for job applications (i.e. resumes, cover letters, etc.) and filling out applications?
Walk me through your job search process. Start at the beginning - deciding to look for a new job - and talk me through what steps you take until you accept an offer.
What are the most important criteria you look for in a job posting?

Personas

In order to better understand how I should design RR, I needed to create personas based off of the research. Here are three examples of the personas I think the design of RR should keep in mind:

Low fidelity designs

I next wanted to sketch out some general ideas of the layout and features would look like on my platform, resulting in some basic low fidelity concept designs:

The Login page

The Match page

The Sign Up page

The Applications page

High fidelity designs

Lastly, I utilized all of the information I had collected and the drafting I had done during the previous stages to build out some of the key pages of the final product in Figma. I am a big fan of unique yet clean fonts and utilizing a bold color balance with white space, so that’s why I designed the general look and feel of the home/login page the way I did.

The Login page

I wanted to make it easy to see exactly where users should go if they wanted to either login or sign up, so I made it easy and clear to navigate between the two (with the default set to the login function). So many times I’ve had to search for the sign-up page - it’s never made sense to me that it wouldn’t be on the first page. I wanted to prioritize ease of sign-up on my home page, since I will rely heavily on lots of people signing up upfront in order for the machine learning aspect of RR to have a high enough sample to accurately predict matches.

I then carried over that simplicity and ease of access into the sign up functionality:

I particularly like using darkened bubbles of the same color to show which option has been selected, as I think that’s a cleaner way of demonstrating selection without removing the other choices (in case of error).

Next, I tackled one of the main pages - the Applications page:

On this page, it was important that the design looked clean, so as not to bury the important information. Yet, I also had a lot of functionality to add, putting the Applications page at risk of clutter. I tried to minimize this by using whitespace, color, larger font sizes, and rectangular-based organization. I also wanted to carry the theme of darkening the selected box that I utilized in the login and sign up pages over into the navigation bar at the top.

Lastly, I designed another important feature, the Matches page:

I attempted to use the same concepts and principles that I did in the Applications page here as well, making use of whitespace, color, font size, and rectangles. This page had even more content, so it was even more critical that I found the right balance so as to not overwhelm the user while still providing all of the relevant information and features.

Conclusion

RR may not exist yet, but hopefully one day it will! Ultimately, taking my project through all of the important UX processes - identifying the problem, outlining the solution, designing the research, building the personas, drafting lo-fi sketches, and creating hi-fi pages - was an important exercise for me in getting the full UX experience. I have learned a lot through this project, which will help me do an even better job on my next one. I am excited to take what I have learned through this personal experiment and build upon it further in graduate school!

What I would do differently next time:

Give myself more time. This project would have turned out better had I not rushed myself.
Better plan the features I would need to include before beginning the design in Figma.
Test the design on real people.
Design a logo.