How to be great at people analytics
The article shows what it takes for corporates to build out effective people analytics teams.
I found the McKinsey’s article from October 2, 2020, on “How to be great at people analytics” useful, because it was complementary to a plethora of cases we studied throughout our class in Leading with Peoples’ Analytics (LPA) at HBS.
Although the cases showcased pros and cons of having dedicated teams perform people analytics at large companies such as McKinsey, Google, and Teach for America, they did not address complexities of launching and growing such teams at well established institutions. And the task is by far not trivial!
According to McKinsey’s article “ [there were] different ways that firms have approached building their people analytics functions. Team size, composition, and organization vary widely, and priorities for capability development and maturation differ significantly” (McKinsey, 2020). The authors further described immense challenges in how corporate environments embraced this relatively new function: “Most companies still face critical obstacles in the early stages of building their people analytics capabilities, preventing real progress.” (McKinsey, 2020).
Frankly, my own experience from three Fortune 500 companies corroborates these findings, because my corporate heavyweights lacked sufficient talent and innovative spirit to capitalize on this new function, or simply did not even have it (e.g., the Public Service Enterprise Group and Shell).
As my LPA class wrestled through the cases, which helped form understanding of how companies used people’s analytics to drive business performance, promotion decisions, and employee engagement, I could not stop thinking about the process it took them to get to a clear dataset and R-code. McKinsey, however, helped me think about this in terms of “a stairway to impact,” which often requires arduous effort of cross-functional alignment and sponsorship of senior leadership teams.
The main takeaway from the framework outlined in the article was that a buildup of people analytics team at a company was an iterative process, which often necessitated taking steps back, before progressing further up the “stairway to impact.” (McKinsey, 2020).
Out of 12 companies interviewed by McKinsey for the article, none achieved the holly grail of step 5 i.e., “reliable predictions.” Some companies were still in the process of dealing with poor data quality, from which it was impossible to draw any reliable and actionable conclusions.
In other cases, company’s technology used for data storage and processing had to change, which highlighted the need for agile and adaptive talent on LPA teams. However, whatever the stage of a company in building its LPA function, the most successful companies had one thing in common i.e., rather than relying on generalist data science skillsets, the companies leveraged subspecialities including “natural-language processing, network analytics, and quantitative psychometrics,” and linked their work to strategic corporate objectives. They also embraced culture of “trust, empowerment, and ownership”(McKinsey, 2020).
So, why is this article so relevant to my learning? I see LPA as one of my most important tools in how I plan manage hundreds of people at Amazon Logistics in the next couple of years post-MBA. As I grow my responsibility to thousands of employees and a complex operation that goes beyond one delivery station, I believe that no human being will be capable to fully optimize a workforce without relying on data-driven approaches (such as machine learning, and natural language processing to perform deep network analytics). Hence, understanding how to best build LPA function at a large e-commerce conglomerate or a smaller startup, will make me deliver impact on people and business that I seek.
Source: https://www.mckinsey.com/business-functions/people-and-organizational-performance/our-insights/how-to-be-great-at-people-analytics
Hi Bartosz, Thanks for this. You point out the hard realities that most companies face and that we touched on in some early cases in LPA. You often have to work with the data that is available and it may be inconsistent, incomparable or sometimes even tweaked once employees have an idea of what you’re out there to do. Establishing a strong people analytics function also requires establishing a strong culture of being ok to gather and process data, not to mention monetary investments in infrastructure. And to get reliable results, it may take months of data collection before the actual work even starts. I think the challenge also lies in measuring the ROI from going into the field, which, funnily, would probably require more initial data.
Thanks Georgi for taking the time to read the blog post and share you reflections. I agree that to fully understand value of LPA one would need to measure its impact on the proverbial bottom line. My observation of what has been happening in the wider industry is that companies take a leap of faith and follow the new trend of building their data science muscles before even thinking how they can utilize this new function well. So before they even think about ROI, the companies should think strategically about the “why” and “how” of growing LPA within their own structures.
Bartosz your reflection made me question: what constitutes a strong people analytics team? Aside from the technical skills, what other skills do we need this cross-functional team to have? Would companies have better data if the people analytics team is more collaborative and focuses more on people development? For example, do we want our PA teams to go to other teams to teach them about the importance of PA and the basics for managers to understand and be more invested in data collection efforts? Do we want PA team members to rotate and work in different team for smaller scale team projects in which team members can quickly see the positive results of reliable data? I think this stairway of impact could be focusing too much on technicalities and too little on the right people and the right strategies that will allow later to have strong datasets.
Sofia, I like your suggestion for members of the PA team to rotate and evangelize PA strategies in various parts of the organization. However, this could only be done on a smaller scale. Since the PA teams tend to be small today (at least I assume they are), I don’t think this can be the sole strategy to raise prominence of PA function within more mature corporates. I believe much depends on frequent communication of measurable impact achieved by PA work on different parts of an organization but this is so hard to measure due to lack of counter-factual. I don’t think any corporate will want or can run a perfect experiment in this context and measure incremental impact from PA work, and this reinforces story-telling and conviction around that PA function has indeed a positive impact and is worth retaining and building out.