Credit: Nick Sirotich

Shoppers hate waiting. If they’ve made the effort to go shop at a brick-and-mortar store, rather than stay home and order something online, they hate not finding what they’re looking for. They hate not finding assistance. They hate bad service.

Unfortunately, bad service is often the norm. Research from MIT’s Sloan School of Management finds that large shares of shoppers can’t find anyone to help them. They’re put off by jumbled piles of unfolded clothes or misplaced items. Many walk out the door without dropping a dime, costing, in one estimate, between 5 percent and 15 percent in lost sales, potentially more if people are frustrated enough to take to social media.

To understand why stores appear dysfunctional, look at how chaotic life is for their workers. Consider the experience of former Target employee Adrian Ugalde. Sometimes, during the holiday rush of Christmas or at back-to-school time, Ugalde received the full-time hours he was promised when he was hired. But outside of that, his hours bounced around, sometimes wildly. One week he was assigned 20 hours of work. The next week, 12. Sometimes he would be on the closing shift until 12:30 a.m., take an hour bus ride or expensive rideshare home to the one-bedroom apartment he shares with four other people, only to be back on the job the following morning at seven. In order to keep his health insurance, the 37-year-old routinely had to beg coworkers to give up their shifts so he could cobble together enough hours of work each week to stay eligible for full-time status. 

Ugalde’s chaotic schedule is the product of the exploitative use of a sophisticated, “smart software” computer algorithm. For the past several years, these programs have been determining how many retail, food-service, or other hourly service workers are needed per shift; what hours these workers are given each week; and which days and which shifts workers get. The algorithms use big data—previous sales trends, weather patterns, consumer preferences—to predict how many customers are likely to be in a store at a given time, and then assign staffing levels to match that expected demand. More rain predicted? More customers are likely to dash indoors to buy stuff; more last-minute calls are made to weary workers to come in. Or, when the rain doesn’t bring in the expected rush, more workers are called in at the last minute and then sent home, often without pay.

Algorithmic scheduling has wreaked utter havoc in the lives, health, financial stability, and future prospects of millions of service workers. Now, 16 million retail workers (one out of every 10 employees in the United States), 12 million food-service workers, and scores of other hourly workers and their families are trapped in a never-ending cycle of precarious and unpredictable low-wage labor. “A computer decides when I work and how much I work and whether I work,” said Ugalde, who made $10 an hour at Target. “It’s just trying to fit me in like a puzzle piece. A computer doesn’t know your life. It only knows what you input.” 

Scheduling software uses big data—previous sales trends, weather patterns, consumer preferences—to predict how many customers are likely to be in a store at a given time, and then assign staffing levels to match that expected demand.

The rise of scheduling algorithms has been driven by one goal: increasing efficiency to increase profits. In a capitalist economy, that’s hardly surprising. Especially in an era dubbed the “retail apocalypse,” when brick-and-mortar stores are under increasing pressure from online retail, it sounds brilliant. Use big data to better track customer demand, then come up with a smart-software algorithm to figure out how many workers you need and when to most efficiently deploy them. The cost of labor is the single largest controllable expense for retailers and other labor-intensive businesses. So it’s the first place they look to cut.

The push to use new technologies to wring as much value from workers as possible isn’t exclusive to this era, or even to brick-and-mortar businesses. New surveillance technology has turned truckers into America’s most-watched workers. Amazon—the death star for physical stores—uses algorithms in its warehouses to track and time worker movements. 

But it doesn’t have to be this way. In fact, it shouldn’t be this way. The software is easy to program to treat employees like human beings. Some companies already have programmed software to be humane. And they’ve found that when they do, it isn’t just employee health and well-being that improve. Customer satisfaction does, too. Research shows that better, more predictable schedules for staffers means happier workers. That means better customer service and more sales. And that’s better for business.

Nevertheless, U.S. corporations have overwhelmingly chosen to program software to do the opposite. As the data scientist Cathy O’Neil writes in her book, Weapons of Math Destruction, “It’s almost as if the software were designed expressly to punish low-wage workers and to keep them down.” 

Scheduling software debuted in many companies right around the time of the Great Recession, just when a spike in involuntary, poorly paid, part-time hourly work took root across the economy. As retail and service companies began transitioning from employing mostly full-time to mostly part-time staff, they began to demand that workers have “open availability.” Algorithms programmed to closely match predicted customer demand and labor then began assigning these part-time, always-available workers increasingly erratic hours.

For years, this problem was largely invisible, subsumed by research on the impact of post-recession unemployment and pay cuts. Experts stumbled on it almost accidentally. In 2014, Daniel Schneider, a demographer and sociology professor at the University of California, Berkeley, and his colleague Kristen Harknett, a sociologist at the University of California, San Francisco, were studying how the 2008 recession had affected families. Much had been written about plummeting birth rates, higher divorce rates, and lower marital quality. Many attributed that family instability to the loss of a job and the devastation of unemployment.

But Schneider and Harknett began to find lasting negative effects beyond job loss. Even among those who kept their jobs, or found new ones, they found an enormous and enormously debilitating sense of everyday uncertainty. Many workers reported that they didn’t know when or how much they would work, if they’d be asked to stay late, or if they’d be called in at all. Despite the scope of the problem, Schneider and Harknett quickly realized that there was no data about what was going on. “We’re so focused on wages in our labor-force surveys in the United States,” Schneider said. “This is different. This is about time.”

In 2016, what began as Schneider and Harknett’s informal interviews with hourly retail and service-sector workers in the San Francisco Bay area grew into the Shift Project, which is now the largest source of data on work scheduling for hourly service workers, with reports from 84,000 workers in the retail and fast-food sectors from across the country. The data includes worker schedules, economic security, and the health and well-being of workers and families. 

Half of the workers in the Shift Project’s database say their shifts change from day to day and week to week. Since the recession, working hours have begun to swing wildly from one week to the next for low-wage workers with less than a college degree. Two-thirds of the workers reported that they have to keep their schedule open and be available to work day or night at the drop of a hat. Two-thirds of workers in the survey get their schedules less than two weeks in advance. Sixteen percent said they have less than 72 hours’ notice. Eighty percent said they either have no input or limited input in making their schedules. 

These workers tend to be among the most invisible and powerless. Previous research has found that the workers who are disproportionately more likely to have nonstandard hours are women, people of color, young people, and people with children. A September 2019 analysis of General Social Survey data on schedule volatility found that workers who are “black, young, and without a college degree appear to be at highest risk.” Research released in October shows that nonwhite workers are as much as 30 percent more likely to have precarious work schedules than white workers.

When Schneider asked one young single mother what kind of schedule she’d prefer, she sighed, and said it would be “amazing” if she could work a predictable night shift, and have someone she trusted sleep with her child, instead of having to frantically call around for help or park her child in front of the TV of whichever relative, friend, or neighbor is available at the last minute any time of the day or night. “I thought it was really sad that a stable night shift was her big dream. It shows how limited workers see the realm of possibility,” Schneider said. “But she’s right to dream for a stable shift, because it’s pretty uncommon. Variability is the rule.”

That variability in schedules, the Shift Project and other research has found, is associated with higher levels of depression and psychological distress—feelings of nervousness, hopelessness, worthlessness, and being overwhelmed—than other low-wage workers with stable schedules. Nearly three-fourths of the workers with erratic schedules reported poor sleep. It’s not hard to imagine the stress, anxiety, and sleepless nights—not to mention financial hardship—that result when your income swings more than 30 percent each week, while your bills don’t. 

These workers are already poorly paid, and the researchers tested whether higher wages or more predictable schedules would make a bigger difference in their lives. Time strain, they found, had twice the impact. “It’s really stressful,” Schneider said, “to have no control over disorganized time.” Try making appointments for the doctor or dentist or parent-teacher conference when you get your work schedule with only a day or two of notice. 

As Sam Hughes discovered, attempting to make time for medical necessities can lead to retribution. When the 26-year-old Hughes (who identifies as nonbinary and prefers a gender-neutral pronoun) got a job at a Fred Meyer deli in Salem, Oregon, in 2018, they made clear they needed four hours a week for doctors’ appointments and wouldn’t be able to have open availability. But instead of the 30 predictable hours a week Hughes asked for, they got 20 erratic hours. When Hughes asked for two additional hours off one week to make it to an unexpected doctor’s appointment, their weekly work hours were cut to 14. 

Like many hourly and service workers, Hughes made so little and had such an unstable work schedule that they qualified for public benefits like food stamps and Medicaid. They wound up working odd jobs for the landlord to pay rent and make ends meet. When we met, Hughes held out their arm to show me scars in the hollow of their elbow. “I was selling my plasma,” Hughes told me. 

Researchers tested whether higher wages or more predictable schedules would make a bigger difference in the lives of low-wage workers. Time strain, they found, had twice the impact.

The stress of disorganized time carries into the next generation. In another new paper released in October, Schneider and his coauthors found higher levels of anxiety, stress, and depression in children whose parents work unpredictable hours compared to those whose parents have set schedules. It’s easy to see why. Research shows that at a time of rapid brain, cognitive, and social-emotional development, children need stability and attachment to a warm, responsive caregiver. Parents’ unpredictable schedules disrupt that. The study found that these parents’ children have multiple caregivers, are constantly changing schedules, and are more often in informal care settings. That puts them at a further disadvantage to the children of better-educated parents with more control over their time, reinforcing inequality.

Disorganized time can also impact the previous generation. Ashley Worthen, for example, a 30-year-old single mother, would get her schedule for her cashier’s job at an Albertsons grocery store one day before her workweek was to begin. If the computer assigned her weekend hours or early mornings or evening shifts when her son’s daycare was closed, she was forced to rely on her mother for help or to cancel her shift. And because Worthen had to be always available and on call on short notice, Worthen’s mother, Ruth, had to be always available and on call to watch her grandson. That meant Ruth Worthen was held hostage to her daughter’s unpredictable schedule and couldn’t get her own job.

“In order for workers to work these crazy schedules, so many other people have to put their lives on hold,” Schneider said. “It’s a hidden subsidy to companies—all these broader networks it takes to make these unpredictable schedules work.” 

(Target declined to comment for this article. Spokespeople for Fred Meyer, Albertsons, the Chamber of Commerce in San Francisco, and the Oregon Business and Industry did not return several emails and phone calls.)

In just the past five years, even at a time of historically low union membership, stories of millions of workers like Ugalde, Hughes, and Worthen have sparked a newly energized worker movement. Last February, workers won a successful class-action lawsuit in California, arguing that on-call scheduling amounted to abusive wage theft. Organizations like Ugalde’s Los Angeles Alliance for a New Economy and unions, including Worthen and Hughes’s United Food and Commercial Workers (UFCW), have been organizing to pass stable-scheduling legislation. (Research shows that low-income workers have more schedule volatility in states with smaller union membership.) Already, the cities of San Francisco, Emeryville, San Jose, Seattle, New York, Philadelphia, and Chicago, as well as the entire state of Oregon, have passed fair-scheduling laws to give workers more notice and to guarantee some pay if they come in to work and are sent home

After years of worker agitation, in 2018, Walmart began to offer workers the ability to choose “core hours” instead of open availability, and began giving them two and a half weeks of advance notice of their schedules. They now give workers nine hours of rest between shifts and lock down shift changes 24 hours in advance to avoid last-minute disruptions. Walmart also recently began offering the “My Walmart Schedule” mobile app so workers can more easily swap shifts, pick up extra hours, and have more control over their time. The changes have led to reduced absenteeism and turnover rates, said Michelle Malashock, the company’s director of media relations and corporate communications. 

Rather than the cost savings that businesses expect from on-demand scheduling, research shows the opposite. A randomized controlled trial at select Gap stores found that predictable schedules increased productivity and sales.

Walmart isn’t alone in seeing business upsides to employment stability. Research into unpredictable worker schedules shows that such schedules, rather than the cost-saving efficiency boon that many businesses expected, are quite the opposite. A 35-week-long, randomized controlled trial at select Gap Inc. stores, for example, found that the health, sleep, and well-being of workers vastly improved at the stores with more stable schedules. In addition, predictable schedules increased productivity and sales. Labor productivity and sales were both about 5 percent higher in these shops.

And, it turns out, the algorithms are fairly easy to reprogram with a different set of values, beyond cutting labor to the bone. Companies in Australia, the United Kingdom, continental Europe, and other capitalist countries use some of the same scheduling software. They just configure it differently, to comply with more robust labor laws and corporate norms, or to account for the humanity of workers. Some U.S. companies, including Costco, also employ worker-friendly scheduling-software programs. For the Gap study, one key intervention simply entailed asking managers to start with schedules from the prior week.

“Predictable schedules are easy to achieve with today’s scheduling systems. It’s all about how you use them,” said Bob Clements, president of the Axsium Group, a global workforce-management consulting company, who likened U.S. companies’ approach to scheduling workers to the game of 52 Card Pickup. “The retailer is really responsible for the configuration, which is going to refine or drive the software.”

Susan Lambert, a professor at the University of Chicago and one of the Gap study’s coauthors, said that much of what drives schedule instability isn’t just the software per se. Instead, it’s that the software doesn’t account for the unexpected—shipments that come in suddenly, new promotions handed down at the last minute from headquarters, or a surprise visit from an executive, all of which require more staff than the algorithm predicted. That, and what Lambert calls “management by fright.” Managers are driven to both meet strict sales targets, based on big data’s forecasts, and keep to tight labor budgets. “So what ends up happening,” Lambert explained, “is a manager will say, ‘We were supposed to sell 60 sweaters by now, and we’ve only sold 30! Someone has to go home!’ ”

That’s the situation that David Sciaudone, 49, found himself in repeatedly as the former manager of a Sears automotive center in Waterbury, Connecticut. His supervisors told him to tightly schedule his staff with demand, so when the garage got slow, he’d have to send people home. “It was really horrible,” he said. “But if you didn’t get under budget, you’d get a phone call, screaming. . . . They’d say, ‘You have 45 seconds, or you’re going to be on the unemployment line.’ ”

The cut-labor-at-all-costs mind-set, which keeps schedules so erratic, ignores the fact that good workers are retail’s unsung secret weapon. Poor customer service or disorganized merchandise could lead not only to a lost sale but also to the potential loss of a lifetime customer, and, if they’re angry enough, their personal networks. “Understaffing costs are potentially large but can’t be measured easily,” said Saravanan Kesavan, a business professor at the University of North Carolina and coauthor of the Gap study. “On the other hand, overstaffing may be a smaller cost, but it’s highly visible.” 

Yet despite the increase in health, well-being, productivity, and sales, the Gap didn’t change scheduling and management practices once the study ended. (Spokespeople from the Gap did not respond to phone and email requests for comment.) Why? Wharton School professor Marshall Fisher blames businesses’ cognitive bias—a business school mentality that they have to keep labor costs low at all costs, and a belief that this requires unstable schedules. 

Both assumptions are false. But they are pervasive. Rachel Deutsch, who heads the Fair Workweek Initiative at the Center for Popular Democracy, said her organization challenges companies to take a Fair Workweek pledge, and offers to work with them to implement it. Has anyone taken them up on the challenge? “The short answer is . . . not yet.”

Adrian Ugalde has since quit working for Target. “I couldn’t take it anymore,” he said. He’s found a new position making more money working full-time hours at a smaller drugstore chain that programs its scheduling algorithms for humans, not puzzle pieces. He has a set morning shift that he chose. He’s cross-trained to work both on the floor and in the back room to account for fluctuating demand so he won’t be sent home early. “To these big retailers, they never get to know you. You’re just a slot they have to fill,” Ugalde said. “Here, I feel they care about me as an individual. I love my schedule.”

Sam Hughes supported the passage of the first statewide stable-scheduling law in the nation. It passed with bipartisan support. In Oregon, retail, hospitality, and food-service companies with more than 500 workers worldwide are now required to give a “good faith” estimate of monthly work hours to employees at the time of hire. Starting in 2020, they must give workers their schedules two weeks in advance, a right to rest at least 10 hours between shifts, a right to have input into their schedules, and compensation for any last-minute schedule changes. Workers can sign up for a “standby” list and be contacted if extra hours become available.

“It’s incredibly helpful,” said Hughes, who quit the Fred Meyer deli to work full-time as the social-media coordinator for the UFCW union in Portland, Oregon. “Workers can plan, have lives, make doctor appointments. And if they call you at the last minute, you don’t have to come in.”

Ashley Worthen changed jobs to work a regular schedule as a bookkeeper at Albertsons. She tracks employee hours and shifts. What she’s noticed since the law went into effect is that, with more notice and more input into schedules, workers aren’t frantically canceling at the last minute or madly scrambling to swap shifts like before. “Everyone is happier. Everyone sleeps better. The atmosphere at the store is so much better for workers and managers,” she said. With her more predictable hours, her mother has been able to get a job washing dishes in a rehab facility, helping boost the family’s income. 

After years of worker agitation, in 2018, Walmart began to offer workers the ability to choose “core hours” instead of open availability, and began giving workers two and a half weeks of advance notice of their schedules.

Of course, not all companies need changes in law to treat employees like humans. Inspired by MIT research on strategies to make all jobs good jobs—which includes giving workers schedule stability—Mud Bay, an employee-owned natural pet food and supply store with nearly 60 locations in Oregon and Washington State, seeks to give each employee the number of work hours they want, create schedules with their input, and publish these schedules three weeks in advance. To respond to staffers who want more hours than any one store has available, Mud Bay is beginning to create “packs” of nearby stores so they can boost their hours by working at any of them. “We spend a lot of time thinking about different accommodations. For example, if someone is struggling with their mental health or struggling with sleep and a morning shift isn’t best, we’ll set up an afternoon shift for them,” said Michelle Markus, the company’s chief people officer.

Right now, managers painstakingly schedule every “Muddy,” as employees are called, by hand in an Excel spreadsheet. But Mud Bay is now rolling out scheduling software. Markus said they’re hoping it will help stores become more efficient—not by using sales data and customer information to cut labor, as other companies do, but by saving managers’ time, giving employees better visibility to available hours, and ensuring that the company is complying with fair-scheduling legislation in Oregon and Seattle. “We want to stay grounded in our mission, our values, and our beliefs about what’s important for us, our customers, and as business owners contributing to the community,” Markus said. “It’s paramount that the technology isn’t essentially running our business.”

As with any technology, that will depend on the humans who program it.

Brigid Schulte

Brigid Schulte is a journalist, the author of Overwhelmed, and the director of the Better Life Lab at New America.