The Solution to Service-Worker Churn

Retailers have long known that high turnover among frontline employees is expensive, draining both time and money as managers constantly recruit and train new staff. Conventional wisdom holds that scheduling is a major driver of turnover. To reduce churn, retailers are urged to post schedules earlier; create consistent, predictable, and fair work schedules; and ban “clopenings”—schedules that call for an employee to work a closing shift and then an opening one the next day. Those steps can help, but a study we conducted of 280 million shifts worked by 1.3 million employees across 20 major U.S. retail chains found that reality is far more nuanced. Different aspects of scheduling affect each store differently, and only an analysis of data can determine which ones are the most important for a given site, and even to what degree scheduling is the culprit fueling turnover.

To date, most efforts to address turnover have been blunt, uniform, and not informed by data on the local workforce in question. With data-rich workforce systems, managers now have the tools to do much better. They can use analytics to design locally tailored schedules that boost both employee satisfaction and staffing efficiency. In this article we show how to identify, prioritize, and act on the scheduling levers that matter most in each operation. Although our data comes from retail, the same dynamics apply across frontline service industries where scheduling instability drives turnover.

The approach we are advocating doesn’t require collecting more data or building new infrastructure. It just entails having the necessary analytical capability. Nearly every retailer already has the raw data needed to understand turnover at specific sites: timestamps (electronic or physical records of when an employee started a shift, took a break, completed a task, or ended the workday), shift patterns, approvals, and absences. However, most companies use the systems that collect this data only for payroll or to prove they are complying with government laws and regulations. To our knowledge, no organization has fully embraced the data-driven customization our research prescribes. Consequently, the approach we outline in this article offers companies with high turnover among frontline workers a remedy that could quickly have a major positive impact on their businesses.

Advertisement

The High Cost of Turnover

Employee retention varied dramatically across the 20 retailers we studied. Annual rates ranged from just 30% to 73%, with an average of 52%, and median tenure stretched from five to 13 months. By comparison, in many white-collar jobs, annual retention rates typically exceed 80%, and even in logistics or manufacturing work, they often remain above 70%. Persistent churn in frontline retail creates a host of problems, such as chronic understaffing, training pipelines that never fill, and service inconsistencies that customers feel immediately.

The direct costs of high turnover include time and money spent recruiting, onboarding, and training. The indirect costs are more subtle but just as damaging: Sales are lost because there aren’t enough sales associates to keep shelves stocked, assist shoppers, and keep stores organized. Meanwhile, supervisors spend more time replacing workers than coaching existing ones. Widely cited estimates by the SHRM Foundation and Gallup peg replacement costs for frontline roles at anywhere from 50% to 200% of annual wages, depending on role complexity and ramp time (the period it takes for a new employee to become fully proficient)—enough to erase the thin profit margins typical in the service sector.

Scheduling data can act as an early warning system, but the signs are not always easy to read. Stores with erratic scheduling (frequent last-minute changes, inconsistent shift patterns, and limited advance notice disrupt employees’ ability to plan their lives) often experience large fluctuations in turnover, absenteeism, and customer service scores. But operational noise—the small, everyday variations in demand, staffing, or logistics—can make schedules appear unstable even when systems are functioning as they should. The challenge for leaders is to distinguish between normal variability and structural problems that undermine retention. Managers should watch out for signs of weak communication, strained middle management, and a culture that prioritizes short-term efficiency over consistency. If firms monitor scheduling patterns just as they monitor customer satisfaction and inventory turnover, they can spot critical morale and retention risks before serious problems arise.

An analysis of scheduling patterns can reveal not only where problems exist but why. And by drilling deeper into shift-level records, managers can identify the specific mix of factors driving turnover in a particular store or region.

Rely on Data, Not Intuition

Each of the retail chains we studied uses a workforce management tool to generate and track the details of the schedules of every shift: when they were posted, whether they changed, and how those changes aligned with employee requests or local conditions. Unlike survey-based studies that rely on what people say about their schedules, these records capture what actually happens. We found that even within the same retail sectors, scheduling practices and turnover rates varied widely—far more than most executives realize.

To identify which aspects of scheduling truly predict employee turnover, we used LASSO regression, an advanced statistical method designed to cut through hundreds of potential variables and isolate the few that matter most. Drawing on prior research in operations, labor economics, and organizational behavior, we built a comprehensive set of metrics spanning five dimensions of scheduling quality:

• week-to-week consistency (or stability) of work routines: whether an employee worked the same days, started and ended at similar times, and received hours comparable to previous weeks

• predictability: the amount of advance notice employees received

• control: the degree to which employees could influence their schedules through requests for time off and changes in their availability—measured by how often managers accommodated such requests

• physical fatigue: the strain created when shifts are sequenced poorly, such as when a worker has short rest periods between shifts, is assigned clopenings, and works long strings of consecutive days

• fairness: whether employees received equitable treatment compared with their peers in the same store—measured by whether they received shorter notice of their schedules, less-favorable shifts, or fewer approved requests for schedule changes than colleagues did

Together these measures provide a multidimensional view of how scheduling affects workers’ attitudes at particular locations and impacts turnover at them. In simple terms, the statistical analysis functions as a truth detector: It distills 166 scheduling variables down to the smallest set that best explain which workers stay and which leave.

Among the retailers in our study, scheduling practices varied a great deal, and so did their impact. Take predictability. In some firms, shifts were posted nearly three weeks ahead; in others, workers got less than a week’s notice. Overall, we noted a 12-day divide between the most and least predictable workplaces. Broadly, longer notice periods aligned with lower turnover: Retailers offering two to three weeks’ notice averaged monthly attrition around 5%, compared with 7% to 8% for those giving less than a week’s notice. Yet the relationship wasn’t absolute. Another retailer kept monthly turnover below 4% with a 12-day notice window, while one that offered a similar lead time lost nearly twice as many employees. Predictability helped, but it wasn’t the whole story.

An additional aspect that varied a great deal was managerial flexibility: how readily supervisors accommodated employees’ requests to change their work schedules. (This is the practice we used to measure the dimension of control.) Across the 20 retailers, approval rates for such requests ranged from below 50% to nearly 100%, reflecting two fundamentally different management philosophies.

Companies whose managers routinely approved scheduling changes tended to retain staff longer than those whose managers didn’t. Retailers that both approved a high proportion of requests for schedule changes and provided frontline workers with ample advance notice of their schedules experienced a turnover rate that was almost half that of their less-flexible peers. Still, a company that approved only two-thirds of requests achieved the lowest attrition rate in our sample.

These findings highlight a central lesson: Data, not intuition, should guide scheduling practices. A data-driven examination of turnover drivers will help managers of individual local operations move beyond simple rules of thumb (such as “more notice is good” or “denying change requests is bad”) to understand the real trade-offs that shape both operations and employees’ lives. Only by pinpointing which factors matter most to workers in their specific context can organizations design schedules that are not just fair but also effective.

A Playbook to Customize Scheduling

Here is a playbook for understanding which aspects of scheduling are fueling your turnover in local operations so that you can create a tailored approach to curbing it. This endeavor is especially worthwhile because compared with remedies for other drivers of turnover—such as increasing compensation and hiring more people to reduce workloads—better scheduling practices don’t add to costs.

1. Identify the factors driving local turnover. To determine which aspects of scheduling are contributing to turnover, start by mining your workforce data and zeroing in on different employee segments and locations. Let your data reveal what really predicts turnover. Many firms already apply advanced analytics to pricing, assortment, and logistics; it’s time to bring the same analytical discipline to the human side of operations.

We conducted our LASSO analysis separately for each company, its operations in each state, and each worker group (part-time, full-time, newer, and longer-tenured) to see when and where specific practices mattered most. We found that each retailer had its own distinct pattern. In some organizations, even individual regions or store formats showed unique drivers—clear evidence that the forces behind retention are deeply dependent on the local context.

The variation isn’t random; it reflects how different operational models and workforce realities interact. For example, in high-volume grocery or convenience formats, physical fatigue and lack of rest between shifts drive turnover; in fashion and cosmetics retail, where employees rely on commissions and client relationships, fairness and consistency weigh more heavily. Even within the same company, stores serving different neighborhoods can experience different dynamics: Those in lower-income areas tend to see stronger effects from fatigue-related variables like short rest windows, while those in higher-income markets respond more to fairness and predictability.

We found the same variety of factors when we examined worker segments. Part-time and newer employees were most affected by having short rests between shifts, a long string of consecutive workdays, or unstable start times. Full-time and longer-tenured employees, by contrast, responded more to fairness and consistency: whether their schedules were equitable relative to those of peers and whether changes were communicated routinely. In short, employees leave for different reasons across retailers. Treating the entire workforce as if everyone values the same schedule features leaves major potential retention gains untapped.

The characteristics of regional labor markets add yet another layer. We mapped results across all 50 U.S. states to see how local labor markets shape the importance of scheduling factors. In the Midwest and the South, turnover was most strongly linked to irregular schedules—for example, when an employee was scheduled on Monday and Thursday one week but on Saturday and Sunday the next. On the coasts, what mattered most was how fair schedules felt—whether some employees consistently received more advance notice of their shifts or more desirable assignments, such as preferred hours or days, than others. These regional contrasts reflect not just economic conditions, such as tighter labor markets or higher costs of living, but also cultural expectations about work-life balance and the influence of local labor policies. In many coastal cities, “fair workweek” laws require employers to give employees at least two weeks’ notice of their schedules and compensate them for last-minute changes, creating both higher expectations of fairness and different operational constraints. For multistate employers, the lesson is clear: Uniform scheduling policies rarely deliver uniform results.

What’s more, as everyone knows, scheduling is not always the factor causing churn. Our analysis of two retailers showed almost no scheduling effect on turnover. If your data suggests that’s true for your organization, you’ll need to explore other factors, such as compensation, advancement opportunities, the design of jobs, leadership, and culture.

2. Prioritize, test, and scale. Once you’ve diagnosed the key drivers, focus on the scheduling factors that matter most to employees and that are operationally feasible to improve. Concentrating on a few high-impact changes can help achieve early wins and build momentum for broader adoption.

Pilot targeted changes in a select group of sites. Use A/B testing or phased rollouts to observe how schedule adjustments influence retention, performance, and morale. Treat testing as a learning lab: Measure what works, learn why, and refine before scaling up.

Once results are validated, expand proven practices first to the stores or employee groups where they’ll have the greatest impact. Communicate the rationale behind the changes so that employees understand the tailored, evidence-based approach.

3. Empower frontline managers. Organizations cannot implement localized scheduling without empowered, capable frontline managers. These managers are the translators of strategy; they turn the lessons drawn from analytics into the daily reality for their teams.

Algorithms suggest patterns, but humans must determine whether those patterns make sense in practice. The most effective store managers use data not as a mandate but as a guide, balancing individual workers’ preferences with operational needs to make scheduling reforms succeed on the ground. Only store managers can understand who is juggling childcare, who has a two-hour commute, or who thrives on extra shifts. A model might flag “rest between shifts” as a key predictor of attrition, yet only a local manager knows which employees are volunteering for extra hours and which are being overscheduled. Translating data-driven insights into daily scheduling decisions requires judgment, empathy, and trust—qualities that no algorithm can replace.

4. Continuously improve. Organizations should turn scheduling into a learning system. They should monitor patterns, build feedback loops between analytics teams and store managers, review retention metrics quarterly, and refine scheduling rules accordingly. In effect, scheduling should be a living experiment rather than a static policy.

Data-Driven Leadership

Research on fair and stable scheduling shows that predictable hours improve frontline workers’ morale, their performance, and sales. Our findings don’t overturn that wisdom; they sharpen it. Stability and fairness matter, but not equally and not everywhere.

Although our data comes from retail, our lessons extend to any setting that depends on coordinated, shift-based work: hospital wards, hotel front desks, airport ground crews, call centers, factory floors, and many others. In all these environments, small changes in the design or perceived fairness of schedules can translate into large gains in retention, service quality, and productivity.

Retailers have long known that localizing product assortments and tailoring merchandise to neighborhood tastes can dramatically improve sales. Our research shows the same principle applies to scheduling. Just as customers in different markets want different products, employees in different locations value different aspects of their schedules.

The broad message is clear: Scheduling is not a universal formula; it’s a portfolio of tailored practices. The best schedules, like the best operations, are engineered locally, tested continuously, and refined using evidence, not assumptions. With today’s workforce analytics, managers can design locally relevant schedules that reflect operational realities, employee preferences, and regional labor dynamics. It’s imperative for senior leaders to use their company’s data to pinpoint which levers truly move the needle in local settings. The companies of leaders who do so will gain a decisive advantage in the form of higher employee satisfaction, retention, and productivity; improved service quality; and lower operational costs.

Copyright 2026 Harvard Business School Publishing Corporation. Distributed by The New York Times Syndicate.

Explore AAPL Membership benefits.