What Netflix Can Teach Talent Management: 3 Ways HR Leaders and CLOs Can Use Recommender Systems Today
As soon as you open Netflix, you see an incredibly convenient list of recommended movies, BBC mysteries, and documentaries you’re likely to enjoy. The core algorithms behind these recommendations have been around for years, but their impact continues to delight and surprise. They just work.
Another source of surprise? The glaring failure of HR and learning organizations to leverage these same analytic tools.
Netflix uses clustering and other machine learning processes to efficiently match viewers (buyers) and producers (sellers). The tidy lists of product recommendations that emerge lead to longer viewing times and increased product engagement, two critical measures of success in this space.
How can human capital leaders leverage these same kinds of analytics in the real world?
Human capital leaders can use recommender systems to improve the match between employees and costly training and development efforts. Better, more efficient development-employee matching can transform individual careers, improve the bottom line, and unleash a competitive talent advantage.
3 Ways To Use Recommender Systems
Undoubtedly there are many ways to use recommender systems for internal talent development. What follows are just three of most immediate opportunities. I am sure a few moments of reflection will reveal many other unique possibilities that fit your industry and your critical metrics.
Use #1: Performance and Training
An enduring training challenge is providing instruction at the right level to maximize learning and (hopefully) implementation.
This is especially true for high-turnover roles such as customer service representatives or new sales associates forced to digest massive amounts of information early in the job training process before they can competently perform their role.
Recommender systems can help by using “in class” performance data to systematically cluster individuals into similarly performing trainee groups. Groups with similar performance profiles are more likely to have common strengths and weaknesses and require similar instruction. This greatly simplifies an instructor’s job.
Critically, some very basic clustering methods can be profitably applied to groups as small as 12 or 15 people (read: this is not just for “Big Data”). The groupings will reflect the unique performance profile of each specific training cohort and new, cohort-specific clusters can be easily created every time.
This is a major improvement over arbitrary, predetermined cutoffs and big step towards principled tailored training.
Use #2: Prediction
Recommender systems can also aid the early identification of likely top performers.
Consider a group of 120 customer service representatives who have completed their initial training and been on the job for at least three months.
If we bring the training performance data of these 120 people together with their later on-the-job performance metrics, we can identify the training performance profiles typical of our later top performers. We can also identify the performance profile of those who struggled in their new position.
With these early training performance profiles in hand, we can then predict who among our current or recent new hire/ trainees is more likely to excel. We can also predict who is more likely to underperform.
What can you do with these predictions? Different companies will prefer different actions according to their workforce needs.
One possibility is to recommend accelerated training for those with a “top performer” profile. This can help reduce total training time for that select group and get them on the job more quickly.
Alternatively, this same system might instead tag them as suitable candidates for early development opportunities that aggressively extend their skills. Increased skills mean increased company value.
But recommender systems can also help those at the other end of the performance spectrum. Those tagged as potential “strugglers” based on their initial training measures may instead be given extended training or early refreshers before they become frustrated and quite or before their subpar performance hurts the company.
The cost of some additional training to solidify core competencies is less than the cost of abruptly losing new hires and initiating a yet another round of recruiting, hiring, and training from scratch.
Note that regardless of the performance level of concern, recommender systems can provide a systematic, repeatable, and improved path to deploying valuable training and development resources.
Use #3: Preferences
Sometimes people just have their own learning preferences. In a very basic application, then, recommender systems can be used to suggest in-house voluntary training and learning opportunities according to what others have previously selected.
Let’s say an employee signs up for a visual design course. It might be useful to know that others opting for visual design have also taken presentation skills training. Here, the recommender system is tapping unused training data to essentially crowdsource suitable recommendations.
This dramatically reduces the search time an employee needs to find a match (again, think matching buyers and sellers) and increases the likelihood of a satisfactory result.
Importantly, we can also extend this basic approach by combining training preferences data with standard HR data already being collected. Recommendations that fold in HR data can yield an even more powerful set of suggestions tailored to job history, desired career paths, relevant workforce demographics, or even organizational changes in mission, culture, and focus.
Over the long haul, this means employees will spend less time hunting and experimenting and more time applying recently acquired skills in their current or near future roles.
From Netflix and Amazon to Google AdWords, recommender systems play a critical economic role by efficiently matching buyer and sellers. By leveraging these same machine learning processes, HR leaders, CLOs, and other learning professionals can take control of the matching problem in training and development.
Moreover, by shifting from the tradition of gut feelings and unexamined organizational processes, recommender systems offer a measureable baseline for principled talent development decision making. This provides grounds for incorporating feedback into our on-going training talent management efforts.
Without that analytics discipline, we simply cannot reconstruct the logic underlying earlier decisions and learn from the feedback the world offers. Such lost organizational learning opportunities are costly. In a tight labor market, those costs accumulate quickly.
Like this post?
Get our FREE Turnover Mini Course!
You’ll get 5 insight-rich daily lessons delivered right to your inbox.
In this series you’ll discover:
- How to calculate this critical HR metric
- How turnover can actually be a GOOD thing for your organization
- How to develop your own LEADING INDICATORS
- Other insightful workforce metrics to use today
There’s a bunch more too. All free. All digestible. Right to your inbox.
Yes! Sign Me Up!
Comments or Questions?
Add your comments OR just send me an email: firstname.lastname@example.org
I would be happy to answer them!
- © 2023 HR Analytics 101