Demand-Based Decision Making for Labor
An important emerging application of Practical AI is improving decisions related to staffing levels. There has always been a drive for operational efficiency and effectiveness in the labor domain, as labor costs comprise about 60% of all corporate expenses. This pursuit is especially challenging in today’s tight labor market — one in which employees prefer to know their timetables in advance, scheduling rules are more restrictive, and the cost of hiring new people is higher than ever.
While the value of good staffing decisions has always been significant, it has become more difficult to predict demand in today’s environment and staff accordingly.
Altered shopping habits, new business models and competitor types, and different influence drivers have made many legacy approaches to predicting customer behavior (and therefore labor needs) less effective.
Additionally, we often seen this topic considered as just a labor scheduling problem. While building shift schedules, assigning people to shifts and managing shift swaps are all indeed processes that can be automated, the highest point of leverage is the decision regarding how many people (and of what position) should be scheduled where and when.
This is true because labor is not just a cost. There can be a direct relationship between staffing levels and the realized demand. Consider a coffee retail store at the busy morning hour. If the counter is under-staffed, potential customers might be turned off by lengthy lines and opt to find coffee elsewhere. Conversely, extra staff in the more quiet afternoon hours can be a dead-weight cost, representing a direct hit to the bottom line.
To be most effective, AI applied to labor decision-making needs to:
In working with many large global enterprises, we have observed that analyzing the economic trade-off (point 2 above) is often not considered as deeply as the forecasting step, but it is critical. Examples of typical trade-offs are:
- A restaurant needs to balance the cost risk of having more staff than necessary with the revenue impact of not having enough staff to service demand or upsell
- A retailer wants to operate efficiently but not lose sales opportunities at peak demand periods due to long lines or an inability to help customers
- A service business wants to minimize staff downtime but make enough appointment slots available to customers to capture potential demand
CONCLUSION
Improving decisions related to labor levels is a ripe opportunity for the application of artificial intelligence. Further, recent technological advances offer the potential for material performance improvement that was not possible even a few years
ago. We have seen executives who apply the guidelines in this paper able to capture significant financial value in short time lines. We hope this stimulates your interest as you consider how to deploy AI pragmatically within your organization.
~ the Predion team
