Losing an employee can be bad news for organizations: they not only lose a contributor, but must face an extended period of reduced productivity before the lost skill, competencies and knowledge can be replaced. Taking steps to minimize this loss of productivity can lessen the negative impact of attrition and allow the organization to ensure that its structure and processes remain stable despite attrition. One of the easiest ways to find successors for leaving employees is internally from the organization’s own workforce. Internal employees tend to be already familiar with the organization’s systems and processes and can adapt to new roles within the same organization much quicker, thereby reducing the productivity gap caused by attrition to a large extent.
In order to select the best possible fit for a leaving employee, organizations can leverage the extensive data about their own employees available on most modern HR systems. Data related to projects experience, designations, performance appraisals, previous career roles and others can all be used in order to derive insights about possible successors for a leaving employee. A good successor recommendation system should be able to find employees who are the closest possible fit for a leaving employee while also ensuring that the successor can be a strong and productive contributor in his new role. As such, it is necessary to design 2 distinct metrics:
- Similarity metrics: These types of metrics evaluate how closely the available employee pool matches an outgoing employee. E.g., organizations can evaluate how closely a successor’s designation, skill set, training profile, or experience matches with the outgoing employee.
- Evaluation metrics: These types of metrics evaluate employees to ensure they can be a strong contributor to the organization in their new role. E.g., potential successors can be ranked on the basis of their performance appraisal scores, awards won, conferences attended, the size of the team they have worked with etc.
Once these metrics have been selected, their relative importance in the selection process has been defined and the relevant data has been gathered from the HRMS, it is necessary to design an algorithm for evaluating successors. One of the simplest yet robust ways to do this is by using the Vector Distance algorithm. The algorithm creates vectors for all employees based on the relevant attributes, and calculates the distance between these vectors using a mathematical formula. The relative importance of various factors can be accounted for by assigning weights to each variable used in the algorithm. However, this algorithm only works with numeric values, whereas employee attributes can be of the following types:
- Numeric: years spent in the organization, total years of experience, salary
- Categorical: department, designation, competencies
- Boolean: Has the employee been trained? Has the employee worked abroad?
In order to transform all these attributes into numerical values, vectorization techniques like a relational matrix can be used.
Employees who are similar to each other will have a low vector distance compared to employees who are dissimilar. Vectors can be created for all employees in the organization and vector distances for all employees in the organization with respect to an outgoing employee can be calculated. Employees with the least vector distance from an outgoing employee can be identified as potential successors. Organizations can also define a threshold distance to identify all potential successors. The threshold distance can vary based on the specific demand of the role (e.g. technical roles would demand a very close fit in terms of skills and competencies, whereas managerial roles would not need to follow technical competencies very closely). Once a list of potential successors is found, they can be ranked on the basis of their scores on evaluation metrics.
The data-driven approach to successor identification provides numerous benefits to the HR function: it reduces the time spent scanning through employee profiles manually and thus simplifies the process of successor identification. It also provides organizations with an objective basis to identify successors, without the risk of hidden or incorrect information affecting the decision.