We’re excited to announce that pymetrics has been acquired by Harver!
Read MoreFor the latest blogs, please visit Harver.
%402x.jpg)
Over the years, Netflix has put significant thought and energy into fine-tuning its recommendation system, which saves users significant time and brain-power. Unlike Blockbuster, which encouraged shoppers to rent the latest and most popular films of the month, Netflix is able to intelligently fast-track the route to whatever film or TV show is most likely to keep you engaged and happy, largely irrespective of its success among other viewers.
If the stats are anything to go by, this methodology has been strikingly successful. According to some sources, around 80% of viewers discover their next Netflix binge through recommendation, as opposed to organically searching the site themselves. Often, it’s one of the first programs at their fingertips on their highly-personalized home page.
Netflix rigorously evaluates feedback from every visit to their service and continually re-trains algorithms with those signals to improve the accuracy of their predictions of what you will enjoy and engage with most. Their data, algorithms, and computation systems continue to feed into each other to produce fresh recommendations and ultimately provide you with a product that can continuously learn about you and, like a close friend, make accurate recommendations about what you’ll enjoy binging next.
So, what does this have to do with talent?
A talent matching engine uses similar concepts to recommend individuals who are best suited for a role within your organization. A talent matching engine like pymetrics’ uses specialized data and algorithms to marry what’s required for the role with an individual’s inherent traits. A high degree of overlap = a recommended match.
The data pymetrics uses are objective measures of real-time behavior collected via gamified assessments. Through recording an individual’s gameplay, pymetrics collects measures at a millisecond time scale. This allows for a highly precise and dynamic reading of the individual’s behavior, akin to the information that Netflix can learn about you by which movies you click on, which you actually finish, and which you don’t express any interest in at all.
The games take approximately 25 minutes to complete and provide 72 different behavioral measures from which to extract signal, and we did not invent these assessments. The pymetrics games are construct assessments taken from the peer-reviewed academic literature, with decades worth of research to support their construct validity and global relevance. What pymetrics did invent, however, is the talent matching technology that applies these tasks for use in talent acquisition and management space specifically.
One of our key differentiators, and the cornerstone of the two-way talent matching engine, is that the exact same set of games played by candidates for a role is also played by employees who are deemed successful at a particular company and in the role of interest. Unlike other vendors in the space, pymetrics is able to build data science-driven & bias-controlled success models that are then used to identify candidates who we predict will be successful for the specific target position and company. Job applicants are then categorized for recruiters as Highly Recommend, Recommend, or Do not Recommend based on their inherent similarity to pymetrics bespoke success model. In other words, the level of match or fit is determined by the extent to which each individual’s assessment performance matches the average performance of successful employees.
pymetrics’ results primarily carry job-specific information, but also incorporate a certain degree of company-specific information, as the success model for a particular job will tend to vary from company to company. Finally, pymetrics' results are non-directional, meaning an individual’s behavior is not globally good or bad, it is simply a good fit for the target position and company or not. In the case of Netflix as well, it is often impossible to unanimously agree on which movies are good or bad, as it depends on the viewer’s affinity or cinematographic “match” if you will to particular films.
If you’re interested in learning more about our talent matching engine, and how we uniquely combine behavioral, data, and industrial and organizational science to match people to jobs, head over to our Science page here.