Predictive analytics in human capital management continues its slow but inexorable march out of the sizzle phase and into the steak -- or for my vegan friends quinoa -- phase. As this phenomenon is occurring, a few topics are getting considerable air time.
- How are predictive engines adapted and applied to the unique business context of every organization – and by whom!
- What types of predictive capabilities in HCM solutions (largely algorithms coupled with machine learning and human testing) have the most relevance and value to a particular HR/HCM agenda?
- Will the predictive analytics guide in solving business problems? … and the all-important …
- How much do data scientists earn and can HR afford them?
Forecasting the winners… more to come (winners and research)
An HCM or Talent Management offering that lacks a compelling predictive analytics strategy and capability set, and is competing outside of smaller companies, is akin to the proverbial “dog that won’t hunt.” (Yes I’ve fully acclimated to living in the South). Although from the buyer perspective, trying to unpack a vendor’s people analytics strategy, or just distinguish it from other capabilities out there that sound awfully similar, might keep some dogs hunting for a while. I’ve maintained for years that the HR tech market needs much more clarity around how solutions are different and why the difference really matters, in a language that typical HR professionals relate to. The absence of this makes the landscape more cluttered and more confusing for buyers.
I’ll be covering key operational and technology dependencies that affect the leveraging of people analytics in my upcoming HfS Blueprint Guide entitled “Predictive Analytics in HR Technology.” This will be published in early March 2017, but way before that, my related HfS POV is coming out in the next week or so. Among other things, it will offer-up a new industry metric called “Time to Predictive Value.” For now, here’s a preview.
Assessing a solution’s predictive analytics capabilities – checkmark or not
Here are three lenses to apply when evaluating whether an HR tech product’s predictive analytics will achieve desired outcomes; and by product, I mean HRMS platform, Talent Management Suite or HR Point Solution:
Time-to-Predictive Value (“TtPV”) is my stab (POV forthcoming!) at creating a meaningful guidepost to help judge one aspect of a product’s capabilities in this realm. It will hopefully bring some much needed clarity to a domain where, for example, “retention or flight risk” -– not a very meaningful metric in isolation, as most metrics aren’t –- often gets a vendor a quarter or half-way toward qualifying for a predictive analytics checkmark.
There are various operational dependencies for leveraging predictive analytics in HCM (or within any business discipline), such as having a large enough relevant data set, sufficient analytics and data science competencies and staff, pursuing closed-loop validations with well defined scenarios, applying appropriate calibrations for different data (e.g., job and organizational) contexts either performed by people or machines (via machine learning), etc. These dependencies and conditions typically take time to be addressed –- from weeks to months or longer. Buyers should have a sense of when they will actually see the predictive value manifesting itself, as that influences ROI and is also a major input to my lens #2.
Degree of Predictive Analytics Business Impact: There’s a wide range of potential business impact and value to be derived from these capabilities in HCM. Two factors that seem to correlate with impact beyond TtPV:
- Whether the best actions or decisions are being guided by the predictive information. In other words, is the analytic prescriptive as well as predictive? (A reason why retention risk in isolation probably has less value than what is often hyped.)
- Is the business problem being solved/avoided, or opportunity created, going to deliver noticeable competitive advantage? Examples include knowing the most important predictors of job success in a critical role, or what factors materially drive or impede employee productivity or customer retention, or is the organization truly ready to succeed on a strategic initiative?
Finally, Innovativeness (yes, it’s a word) of the predictive analytics capabilities: The more innovative a set of these capabilities are, particularly if they lead to practical and measurable business value delivered in relatively short order, the more it inspires other creative ways for solution providers to help solve HCM business problems. Data correlations and cause-and-effect relationships that are very intuitive to discern or simply the product of good common sense (e.g., freezing salary increases or cutting back on company-paid benefits will likely result in a spike in employee turnover) earn very low marks on the innovativeness scale.
In contrast, when Walmart years ago determined that putting diapers on sale will often lead to increased beer sales (somewhat logical, but only AFTER the non-intuitive relationship was discovered), now that’s a winner.
People analytics is hot, and predictive capabilities is a major reason why. But in order for customers to derive business value commensurate with what they are paying for the surrounding solution, they must look beyond the sizzle and assess the quality of the steak in meaningful and business-relevant ways.