Pymetrics – Is customized AI solution for perfect match making in hiring?
Is AI the future for hiring and people management? Can people see the value in substituting human beings making hiring decisions and rely on what a machine tells them? Pymetrics certainly believes so!
I discovered this company past summer, when I was recruiting for my summer internship at Boston Consulting Group. There were a few games that I played in rapid succession over 20 -25 mins. I must admit that I was not sure how playing these games would help the firm assess my capabilities.
After some research I read about the methodology that Pymetrics uses and the theory behind BCG using this as a recruiting tool. The company has twelve tests (some involving pressing spacebar key when we saw a red ball instead of a green one). The data collected from these tests is used to provide a score for the candidate on ~90 traits. As part of the onboarding a new company, Pymetrics works on collecting the data for the high performers in each role to benchmark the traits required to hire for the role. Then they play the match maker and help identify the best candidate for hire. Further for companies such as BCG see value in using this technology can be extended to staffing teams where different types of skill sets are needed to build agile teams. Pymetrics explains the product in a simple manner but there are three things about the product that I really liked.
- Finding perfect fit – The algorithm does not just stop at telling us the best person for the job but is also able to provide a detailed report of the skill sets and therefore match them to another role in the organization that the applicant would not have otherwise applied to
- Customizable – The product is customizable to the needs of the company that chooses to employ it. This means that we can use this tool to hire for companies across multiple industries, as highlighted in the list of clienteles for the company
- “Measure Potential and not Pedigree” – Pymetrics suggests that it takes out the bias that people have regarding branding and perceived skills and rather let the test results speak to the capabilities of the candidates
The platform enables companies to use the platform not just for hiring but also for performance management. It recognizes the shift needed in adopting this product would require recruiters to make is significant and therefore can be pitched as an initial screen that could take away the bias for skill and demographics and provide more data to arm managers taking interviews. There is an extension to the product to conduct recorded interviews and have the algorithm analyze the answers of the candidate and further supplement the traits tests rather than in person interviews.
However, like most products that are trying to solve for this problem. The product does have a few problems to solve before it can become a mainstream solution.
- Onboarding – It takes a lot of time and efforts to onboard a new company and collect the data to train the machine so that the results produced are reliable and make sense.
- Perfect solve for bias – There is no perfect solve for the bias as the company works to smoothen this data on the back end and are still working to make making the best system. They have tried to correct for this with high standards of ethical design principles and have open- sourced the algorithm to invite corrections and criticism
- Competition – This is a very crowded space and there are many players who are offering similar products. This shows that there is a lot of potential, but it would make it harder to launch a differentiated product.
I was thinking about the question Professor Polzer asked me in my Leading with People Analytics class – “Will machines make hiring decisions?” I believe that machines certainly would help put some method to the madness and help remove biases that we as humans cannot ignore. However, the idea of machines doing that seems to trigger a very uncomfortable feeling in us. As if the process would be drained of emotions. I am intrigued whether companies such as Pymetrics can help bridge that gap.
Sources:
Youtube Videos
https://www.youtube.com/watch?v=oig5nL9XeDo
https://www.youtube.com/watch?v=tq5IVfgqZxk
Company Website
Articles
https://www.myconsultingoffer.org/case-study-interview-prep/bcg-pymetrics-test/
Thanks Puja! This is such an interesting topic. Not too long ago, I was asked to take a 30-40 minute “cognitive assessment” after submitting a job application to a high-growth social media company. Similar to your experience at BCG, I was not sure what the team sought to understand from my performance, and felt uneasy having my “potential” assessed through this avenue alone. Perhaps if I had the chance to speak with someone from HR before the assessment to (1) understand their goals; and (2) share a bit more about myself, I would have been more willing to participate. Maybe I should get with the times (probably), but I also think it’s important that companies don’t lose sight of the personal touch in hiring.
This platform makes me nervous. The connection between how I respond to red dots on a screen and some 90 traits is far from intuitive. In the end, I’m left with no choice as the applicant but to believe that the algorithm knows what it’s doing. Perhaps this is a good thing, since it disincentives me from trying to game the system.
On the other hand, something that may not get captured by this system is an applicant’s grit and work ethic. Not every one may be able to get through these cognitive tests. I’ve been wowed time and time again by how sheer grit has been able to help such people deliver outstanding results. Hopefully the algorithm has a way of measuring that too… But then again, the company can claim it already does without us ever knowing if it truly does.
That said, I must say that it is a step in the right direction and I agree that the competitive landscape is perhaps the biggest obstacle Pymetrics has to navigate in order to be successful. Good luck to them!
I agree Zubby, that the emotional and stress situation reactions of a candidate is not getting appropriately evaluated by using this test alone. I think their video interview analytics tool and other ways of quantifying those aspects would be welcome. However, assessing the grit of a candidate remains a challenge in today’s hiring process as well. Despite having intensive in person interviews, reference checks and detailed resumes, we are not able to assess those aspects of the candidates being considered for hiring. I share your excitement on this topic and truly believe that this could be a tool in democratizing opportunities.
Thank you for sharing your experience Pallavi! I do agree that there is a degree of discomfort about being evaluated by just algorithms and the tests that one has to take in a defined period of time. I do wonder whether understanding the goals of the test and sharing more information than the resume you had submitted would change the results of the tests because one of the reasons that this test works is that it takes away any bias of perceived brand value and skills you are said to have based on what you have on your resume and tests for just skills needed for the job. I do however agree that there is room for measuring other qualities that are more related to human emotions that could be baked in using other forms of test. I also struggle with completely trusting an algorithm to make the decision on whether I get the job or not. I am however excited to see where this company takes the product!
I agree with these comments, it would take a lot of convincing for me to believe Pymetrics can actually evaluate potential and fit based on 20 minutes of games. My blog post on “AI Snake Oil” touches on this: https://d3.harvard.edu/platform-peopleanalytics/submission/ai-in-people-analytics-rhetoric-and-snake-oil/