Reskilling is going to define the return to work. There is untapped value in building an internal pipeline of talent. Emerging roles like Machine Learning Engineer and Security Analyst showcase the high ROI of reskilling vs external hiring.
Internal promotions and referrals are the preferred source to fill open jobs. The cost to hire, time to hire, and performance benefits are well documented. Their domain knowledge comes with them into the new role. Their relationships with other teams come with them.
Reskilling is a natural evolution of training. Training is a legacy construct designed to maintain skills and/or provide new skills for basic career moves. There is a lot more each employee is capable of. We have known that, from a commonsense standpoint, for a very long time. During every layoff, there are people the business wants to keep. It is their role which is no longer needed, not their talent.
Complexity prevented businesses from getting the full value from their workforce. A lot of exceptionally talented people were lost as a result. Institutional knowledge was lost. We have data and analysis capabilities to model more advanced career paths. Reskilling is an emerging machine learning use case with high ROI.
Reskilling vs Training
I started my career as an installation and network technician. I ran a small business to pay my way through college. My first corporate job was as a midlevel installer. Twelve years after starting that job, I was one of the earliest Data Scientists.
Along my journey, four different companies had the chance to keep me. I was recognized at all four as a standout, winning awards along the way. It was not a lack of desire. Those companies did not have the infrastructure to reskill me or move me through a complex career path.
That is not a paradigm traditional organizational development and training is built to handle. There is no roadmap to a job which does not exist yet. There is no curriculum adaptive enough to achieve this level of ROI.
Training opportunities were offered at each company. My last company had an excellent training program. Training is designed around a very narrow set of opportunities. Moves into leadership or increased technical capabilities are available in most companies.
Reskilling is different. Many employees have some of the skills necessary for a dramatic career shift. Becoming a better Software Developer or becoming a Manager has value. However, there are less obvious career moves, even moves into roles that do not exist yet, which provide greater ROI.
In many cases, talented employees are stuck in roles that will eventually be downsized. Especially now, with the push towards automation, companies will lose talent they would rather keep. The business does not have a way to move those employees into a role that has longer term value.
Employee Lifetime Value
I left my last corporate job, I was laid off, in early 2012. I had tried to move into a Product Manager role in 2011 which would have kept me at the company. There was no framework to support that move. By 2014, I was recognized as one of the top experts in Big Data, Data Science, and Competitive Strategy. I had published on all three. The business had no way to measure my value except with respect to my current role.
My situation points to the concept of Employee Lifetime Value. There are solutions that support evaluating performance. Performance reviews are descriptive analytics. It is a checkpoint explaining what happened over the last quarter or year. Training fits the performance evaluation framework. Descriptive analytics do not allow for a more comprehensive solution.
Employee Lifetime Value (ELV) is a predictive measure. ELV models explain the ways an employee can contribute in the future. Most training is shaped by the concept of career path or progression. These are the familiar straight lines most employees follow from role to role. Career progression focuses on the most common paths forward. If the business provides training, the employee will be able to follow a small set of predefined career moves.
ELV Models for Individual Value Optimization
ELV models use a different set of data. Capabilities can be tied to business value. That is what ELV models take as raw inputs. Some capabilities decrease in value over time. Others increase or hold their value. This is an optimization problem with a lot of moving parts. Machine learning models can provide decision support to help the business understand how reskilling moves employees from low to high value capabilities.
ELV models are prescriptive and provide insights which are not obvious. That is the hallmark of any good machine learning use case. One example is someone with a current, high value capability. Let’s use Data Science as an example.
A Data Scientist has a high-performance value. Their capabilities are in demand and produce higher than average business value. In a training paradigm, their skills would be kept current with annual training. The goal is to keep that employee performing incrementally better. In a reskilling paradigm, the model would optimize their ELV. The goal is to find the maximum value the employee can provide to the company.
A performance review based on an ELV model is dynamic. Evaluation points are forward looking as well as role based. They are designed to rate the employee not only on their performance in their current role but also rate emerging, higher value capabilities. Mentorship is an emerging, high value capability for a Data Scientist. The ELV model can prescribe reskilling the employee towards mentorship.
A business planning to hire more Data Scientists over the next year or two, will benefit from someone who can reskill other employees who rate above average on their emerging Data Science skills. This is why the ELV model is important. Those relationships are complex. So is building the graph of evaluation criteria and reskilling plans.
A Data Scientist who can reskill five others is more valuable to the business. The hiring plan no longer involves bringing in external talent. The Data Science Mentor’s value is optimized. Five employees’ values are optimized.
Measuring Capabilities as Part of Business Valuations
ELV models allow a company to put a tangible value on their workforce. The business can justify spend based on objective ROI. Google, among others, are acquiring companies for their talent. They can calculate the cost of buying the company is lower than the cost to hire that talent in a traditional way. Essentially, the company’s valuation is capability based as well as revenue/IP/customer/etc. based.
ELV models create a new asset to be added to the balance sheet, talent value. Labor costs are their only representation right now. There should also be a line for workforce value. Part of the guidance for share price needs to include the trajectory of that value.
Decisions about employees are strategic. That is obvious while also difficult to quantify and support. The data is there but it is not actionable in its raw form. The ELV model creates a roadmap to align reskilling with the strategy roadmap.
Force reductions happen organically. Rapid downsizes from eliminating business units, product lines, or staff due to automation can become controlled draw downs. As those capabilities lose business value, the ELV model will recommend reskilling well in advance. The model does not wait until the capability’s value becomes negative. It begins recommending reskilling at the beginning of the value slide. It also reskills downstream talent to manage that process.
From an employee decision making standpoint, they can watch the value of their capabilities increase or decrease over time. There is a perception that becoming better in your current role creates job security. Many employees do not see a layoff coming and an ELV model can change that.
Model Driven Evaluation and Career Cycles
ELV models create a career cycle and cadence. Learn, apply, train. Career performance metrics are supported by annual performance reviews. Time (duration) to reskill. Quality of application. Quality of mentorship.
The second point, quality of application is the traditional performance evaluation. It measures an individual’s ability to reskill. The performance evaluations for employees they mentor, measures their ability to mentor. The performance review is a predictor of future success in the other two phases. Again, it is layered and needs to be an input for the ELV model.
I cannot underemphasize the changes this shift will bring. A business can use ELV to migrate from a legacy to a modern business model. The concepts of dynamic business models and transient competitive advantage have been talked about academically. Machine learning can apply those concepts in a tangible way. ELV models are one part of that transformation.