Over 2,000 mentors available, including leaders at Amazon, Airbnb, Netflix, and more. Check it out
Published

Alternatives to take-home assignments

Hiring data professionals? Take-home assignments may not always be beneficial.
Leoson Hoay

Data Scientist, Learning Collider

As a job seeker, you have probably encountered and worked on several take-home assignments from your would-be employers. If you are in charge of hiring at your company, you may have dished out a few of your own. 

At first glance, the take-home assignment has its merits. It is often difficult to tell if someone has the skills that they claim to have just by examining their resume. Take-home assignments provide a skills-based scenario that engages the candidate on their expertise, which employers can use as a confirmation of the candidate's competencies. As a quote from the longest tenured CEO of Starbucks goes:

Hiring people is an art, not a science, and resumes can't tell you whether someone will fit into a company's culture.
- Howard Schultz

Schultz speaks of "culture", which relates to another proposed strength of the take-home assignment: the ability to tailor the problem set to something related to the company's mission or day-to-day tasks - which may reveal more about the candidate's ability to perform on the job, as compared to one-off SQL questions during technical interviews or regurgitating specific algorithms in live-coding interviews. In addition, the take-home assignment can purportedly increase equity by shifting the focus off paper qualifications and on to practical skills. 

Yet, do these merits truly translate to real-world benefits in the hiring process? My experience as a data scientist and data science leader tells me that this is not always the case.

Problems with take-home assignments

More often than not, the take-home assignment is only handed out after initial screening rounds, which means that the resume (and other associated paper qualifications) are still looked at first. Regardless of how egalitarian this initial screening is, the role of the take-home assignment in increasing equity is thus diminished in this sense. Moreover, take-home assignments have an accessibility problem - individuals with access to better technology, better working environments, and better connections are in a better position to excel at these assignments and complete them more efficiently. 

Speaking of efficiency, a real problem that plagues take-home assignments is the time commitment required. Articles from Slate and CNBC have cited worries from applicants, with some portraying take-home assignments as potential 'free labor'.

I’ve been selected to move on to the second stage interview process for a really interesting position. It’s an events position and they mentioned in the first interview that we would be asked to complete a project if selected for the second phase. I’ve just gotten the outline of the project for the second phase, and they are requesting that we submit three different proposals for three separate events, including budgets, marketing plans, staffing, logistics, income statements, as well as concept/design/themes. They have given us seven days to complete these assignments, and we must drop off hard copies of each assignment on the deadline date. Am I wrong to think that this is incredibly excessive?
- Quote from Slate article, March 18 2019 Image

I am an optimistic person, and I would like to think that most companies are not intentionally taking advantage of job candidates. However, even just the perception of take-home assignments as 'free labor' and time-consuming can hurt the hiring process - with strong candidates possibly dropping out of the pipeline if presented with one. In my experience managing analytic teams, I have had chances to participate in hiring sprints where a take-home assignment was involved. I can say that it definitely created significant tedium and delay for the hiring process, and have witnessed good candidates drop out or accept an offer elsewhere.

Alternatives

Here are some other ways to go about assessing a candidate's abilities:

  1. An alternative strategy I have found to be very effective is getting the candidate to present a project that they completed in the past at the technical interview. For experienced candidates, this can be something that they have done professionally; for entry-level candidates, it could be something from school or their personal portfolio that relates to the position's responsibilities. As they say, the ultimate test of your knowledge is your capacity to convey it to another. There are few better ways of assessing someone's understanding and process than getting them to teach it to you. And using something they have already done eliminates the time wasted on completing an arbitrary take-home assignment.
  2. Related to the above: Ask specific questions about the projects in their portfolio and the things already listed on their resume. These questions should beg explanation. For example - if a data engineering candidate has a project showcasing an ETL (Extract, Transform, Load) pipeline, you could ask the candidate to explain how they arrived at the architecture they chose, and why they chose it.
  3. If having a specific scenario is desirable, then we can revert to the tried-and-true role-play interview. The scenario should be posed during the interview itself, and should avoid focusing on testing for rote memorization ("What Python packages can you use to create a softmax layer?"), and instead focus on knowledge fundamentals and decision making ("Why would you use softmax in this case instead of argmax?"). The former is easily searchable online or found through interacting with ChatGPT anyway, and not being able to remember specific pieces of code at all times is a non-factor in the real world.

In today's context, where powerful tools such as ChatGPT are transforming the way individuals work, consume knowledge, and create value, the take-home assignment just obscures too much of the candidate's process, and is in most cases disrespectful of the candidate's time. In data roles, the assignment often involves processing and analyzing some data with respect to a research question or scenario. Those with unrestricted access to fast internet, fast computers, and tools such as GPT will certainly succeed in completing it, and completing it quicker than those who do not - and it is often still necessary during the face-to-face interview for the candidates to explain their process of completing the assignment.

Ask them instead to explain their process for something that they have already completed in their professional/academic journey. Or assess their performance on a role-related scenario that can be done on the spot during the face-to-face interview. Either of these options will be a reliable way of assessing the candidate's competency while still showing respect for their time - in turn saving the organization time on the hiring process, and minimizing the risk of losing good candidates to the ether.

Find an expert mentor

Get the career advice you need to succeed. Find a mentor who can help you with your career goals, on the leading mentorship marketplace.