Dealing with those pesky no-shows!

This article first appeared in DNA India on December 15, 2016.

All hiring managers dread the no-shows. It is their worst nightmare come true. We have all come across those situations. Haven’t we? Everybody is excited about the new person and we offer him/her the position. The prospective employee then starts to serve out the notice period with the current employer. In industry segments where three month notice periods is the norm, months pass before a sudden bombshell in the form of an email lands on the hiring manager’s desk. The prospective employee politely informs the hiring manager that due to unavoidable circumstances, he/she would be unable to join the firm, leaving the hiring manager fuming and having to start the search for filling the position all over again.

Hiring no-shows cost organisations a lot of money. Studies suggest that in India nearly 37% of candidates who have been offered a position, do not join the organisation. Two thirds of these candidates drop out towards the end of their notice period. Muse on that for a second. More than a third of candidates who accept the offer do not show up and what’s worse, 67% of them communicate their intention to not show up very close to the joining date leaving employers scrambling to fill those positions.

Employers look for different ways to handle such situations including shortlisting multiple candidates, and engaging with candidates throughout their notice period. While these approaches have yielded some results, organisations are looking for ways to decrease their hiring no-show rate more aggressively. Can data analytics offer help in such a scenario?

I recently had a bank reach out to me with this exact problem. With three month notice periods the norm in their industry, they were finding that these no-shows were costing them enormous amounts of money. So how can we use the power of data to help in such a situation? The idea is to use data on the employee, mostly gleaned from resumes and the interviews along with macro-economic factors. These can then be pushed through machine learning approaches to come up with models that can predict with fairly good success rates, the propensity of a candidate to be a hiring no-show. Based on the probability values for each candidate, organisations can then come up with a range of options including regular emails, phone calls and the like to keep in touch with the candidate during her notice period.

I know the obvious question on your minds. How accurate can these models be? My experience working with employee data sets suggests that one can get 60-70% accuracy. This by itself might not sound much but given the fact that currently an organisation has no way to predict which way a candidate would go, the resultant cost savings alone will be more than worth it.