Lessons from an Amplified Mistake
How to automate predictive scheduling to better serve your team, and your bottom line.
The Washington Post recently published an article about Brookdale Senior Living, the leading provider of assisted living, operating over 650 facilities in the United States. The company has come under fire for the use of a scheduling algorithm that resulted in systemic understaffing and brutal working conditions for employees, which severely impacted residents, leading to lawsuits, lost revenue, and a crisis for the company.
For restaurants, Brookdale provides an important case study in the use of data and technology related to labor management, and provides some lessons in how to approach harnessing its power to drive better business outcomes, without precipitating a crisis.
I’ve often described the risks associated with leveraging automated technology as being like running a marathon with poor form. When a single operator makes a bad decision here and there, those headaches generally are not significant at a company-wide level. But when bad decisions are duplicated at scale, the compounded negative impact can damage your people, alienate your customers, and decimate your brand.
In the face of these risks, a common strategy is to avoid using technology to make core operational decisions. However, this leaves a lot of value on the table for restaurants that are struggling with margins. In one analysis we conducted, consistently improving staffing efficiency by just 1% can drive bottom-line profitability of over 10%.
So how can restaurants reap the benefits of predictive automated scheduling without doing harm?
Use your real data.
One of the biggest mistakes in the Brookdale case was that underlying assumptions were based on theoretical studies rather than actual data gleaned from their own performance in real situations. Their senior executives timed staff performing various tasks and extrapolated a model about how much staffing was necessary to operate an entire facility based on those results. Needless to say, the realities of running a facility are so location dependent and nuanced that the models significantly underestimated the actual staffing levels required.
Restaurants are collecting enough data through their various systems that they are now able to get better insights around baseline performance from actual operational data, and don’t need to rely on lab test scenarios. The relationship between actual staffing practices and essential KPIs—like guest satisfaction, employee satisfaction, employee retention, waste management, and long-term sales growth—can be used to calculate optimal staffing levels that support effective, brand-building operations.
Keep managers in the loop.
Brookdale compounded the negative impact of its staffing algorithm by disempowering the people who were seeing the negative impact of the models first-hand. Site managers regularly informed executives that the staffing levels set by the algorithm were insufficient, but their feedback fell on deaf ears. Instead of granting them discretion to adjust, managers were criticized if they deviated from the recommended staffing.
Anyone who has worked consistently with restaurant managers knows they are under a lot of pressure and their view of the big picture can become obscured from the frontlines. Naturally their feedback may contain a fair amount of bias and warrant skepticism. But general managers are typically the most informed when it comes to their location’s particular challenges. Any automated system should not replace managers, but rather, expand their toolset in getting to greater efficiency by empowering more informed decisions and adjustments tailored to their specific situations. Circumventing managers risks muting the very voices you’ve entrusted to warn you of impending disaster.
Focus on the full human.
When data was limited, it was sometimes necessary to view employees as anonymous units of labor. When all you knew was somebody’s wage rate, title, and hours worked, there was a natural tendency to see people as interchangeable.
But that’s not how a good general manager operates. Successful GMs understand and make scheduling decisions based on the idiosyncratic tendencies, strengths, needs, and preferences within their team. Some staff members love an early open, while others have to do school dropoff, and still others are great at holding down a late night rush after other colleagues have been dismissed.
As more functions are delivered through technology, restaurants gain richer insights into what drives—and the impact of—employee performance. By stitching together data from several common systems, it’s possible to analyze everything from employee training history and tenure, to an employee’s correlative impact on guest experience and order accuracy.
When the full employee profile is taken into account, striking a balance between staffing efficiency and increased productivity becomes attainable. And if done in alignment with employee goals, you incentivize shared outcomes that increase employee satisfaction and support restaurant profitability.
As we continue to move towards the use of more advanced data and technology to support restaurants becoming the best versions of themselves, it’s important to understand there are no silver bullets. Simplistic applications of technology that look like shortcuts to better performance are often, instead, express lanes to catastrophic outcomes.
Leveraging data and analysis for better decision-making and automation is a journey. A journey that can benefit every stakeholder, business leader, employee, and guest. A journey that will fundamentally change the way restaurants operate without sacrificing the spirit of hospitality that keeps us showing up every day. A journey that starts by questioning every assumption about how organizations identify value and make decisions.
