Technological Change and Store-Based RetailRead the full report
A joint report with the UC Berkeley Labor Center, written by Françoise Carré, research director for the Center for Social Policy at the University of Massachusetts Boston J. W. McCormack Graduate School of Policy and Global Studies, and Chris Tilly, professor of urban planning and sociology at UCLA.
The retail sector, like the rest of America’s economy and society, is being subjected to an enormous stress test by the coronavirus pandemic.
Grocers, pharmacies, and such mass marketers as Walmart, Target, and Amazon are adjusting to shifting patterns in demand for goods, a dramatic increase in online ordering, and new social distancing guidelines in stores. Workers at these retailers have documented overcrowding and insufficient safety precautions at stores and online fulfillment warehouses. In contrast, department stores, apparel, and luxury retailers have had to close their (“nonessential”) stores and endure plummeting sales as consumers focus on the basic necessities. This workforce has suffered massive furloughs and layoffs estimated to be in the millions. The economic ripple effects of the lockdowns necessitated by the pandemic already include decreased discretionary spending and economic recession. These challenges almost surely will continue to hammer retail businesses and jobs even after closing orders are lifted and consumer fears of going to crowded places abate.
The week-by-week developments in the pandemic-driven economic shutdown and gradual reopening have been gripping. At the same time, it is important to keep our eye on longer-term industry trends, including taking into account how the 2020 crisis and its aftermath are likely to intensify, blunt, or divert them. In this report, we focus on trends in technology adoption in the retail sector, looking beyond the effects of the current crisis to trace how retailers are using digital technologies in ways that alter the quality and quantity of front-line retail jobs. While we recognize the pandemic’s possible impacts on the retail workplace throughout the report, the bulk of our discussion concerns longstanding trends that appear likely to continue over the next five years or longer.
Even before “coronavirus” became a household word, there was a widespread expectation that digital technology would bring big changes to store-based retail. The expectations of different observers vary considerably, however. Some predict a dark vision of a “retail apocalypse,” in which e-commerce will almost entirely wipe out stores. Others describe an imminent “retail renaissance,” in which technology adoption will free retail workers from repetitive drudgery to instead serve as expert guides about the merchandise, and valued, empowered implementers of a store’s sales strategy. Our mission in this report is to analyze available evidence todetermine how plausible either of these scenarios is, to spell out likely workforce impacts of new technologies, and to examine the choices and tradeoffs facing retailers and policymakers in an era of rapid technological change.
COVID-19 appears likely to accelerate many changes in the retail sector that already were in process. This includes the broader trends reshaping the industry—growing market shareconsolidation by a small number of giant corporations, and a shift from traditional department stores and mall-based apparel sellers to mass marketers. It also includes the longstanding pattern of “low-road,” cost-minimizing business practices that have degraded job quality—and in the context of a pandemic, have jeopardized worker safety. The current crisis also may introduce incentives for accelerated diffusion of new digital technologies that are transforming retail work—including the shift to online sales, the spread of cashier-less checkout, increased utilization of autonomous robots, and heightened digital surveillance of both customers and workers.
The effects of COVID-19 on technological adoption in retail will not be unidirectional, however. On one hand, the need to track pathways of contagion puts a benign face on forms of workplace surveillance that might in ordinary times have faced greater resistance. On the other hand, overall sales declines will deplete or even exhaust retailers’ available funds to invest in tech—and if the virus-triggered recession is long and deep, this situation will persist. The crisis likely will deepen retail enterprises’ “digital divide”: market leaders—especially those with a strong online presence—will be able to undertake robust investments in technology, whereas others will lag in such investments.
In this study, we focus on brick-and-mortar retailers and store-based jobs, particularly in grocery and general merchandise (which includes large discounters), though we also include some examples of apparel and home goods retailers. Our data consists mostly of interviews with key actors and experts in the retail terrain, including consultants, technologists producing and selling digital technology goods and services, retailers themselves, representatives of retail-sector unions and advocacy groups, and an academic. We also attended industry association conferences and draw to some extent on published sources, including consultant white papers, print and online media, and academic literature.
The U.S. retail sector today is in some ways primed for change; in other ways, it is ill-positioned to pursue it. While retail is a heterogeneous sector, certain characterizations apply broadly. The industry as a whole is “overstored,” with more retail space than needed. Profit results are mixed: margins on sales are thin, but return on investment is handsome. Retail has been a favored target of private equity companies, and asset-stripping by such companies lies behind some high-profile retail bankruptcies in recent years. Ownership in the retail sector is highly consolidated and continuing to concentrate, with a handful of companies accounting for a large and growing share of total sales. Employment also is highly concentrated, although less extremely than sales.
1. The employment baseline and recent changes.
The 16 million people in retail (a statistic from “normal” times preceding the pandemic) consist above all of salespersons, cashiers, stockers, and supervisors of these workers. With the exception of managers, supervisors, and health care workers, these are relatively low-paid occupations, with high turnover, low formal credential requirements, and limited presence of unions across the United States. Some broad trends in employment are discernible in the data. E-commerce is of course enjoying rapid growth in employment, but the trajectories of the different retail subsectors have diverged widely. In the years since 2011, as the economy shifted from bust to boom, grocery employment grew briskly (though it has plateaued in recent years, pre-pandemic), clothing stores have stagnated, and general merchandise has had mixed fortunes, with department stores struggling but other sections of general merchandise, such as big box sellers and dollar stores, continuing to grow. The job losses are due in part to e-commerce, but also in important part to the continuing growth in the dominance of discounters, including the big box players and dollar stores, which are growing at the expense of department stores and apparel retailers. Their leaner staffing models lead to overall drops in employment.
These changing fortunes have affected some sociodemographic groups more than others. Cashiers in grocery and salespeople in apparel and general merchandise have taken significant hits—and, of note, these jobs predominantly are held by women. The bleeding of department stores stirs concern because the general merchandise sector employs far higher percentages of women and people of color than retail as a whole, and far more than e-commerce, whose workforce is considerably whiter and more male than retail overall.
2. Stores will survive, but will change.
Despite continuing talk of a “retail apocalypse,” retail stores remain a durable way of selling, and we expect them to remain so. As of late 2019, e-commerce sales still only amounted to less than 10% of retail sales, and there were conflicting data on whether store openings or closings were more numerous that year. An April 2020 survey found that in the midst of widespread shelter-in-place policies and warnings about the dangers of person-to-person contact, 70% of Americans still were buying their groceries in stores. A set of advantages of store-based shopping contributes to the continued dominance of stores in today’s worst-case scenario, and will continue to matter long after. Customers will continue to want to look at products up close, get personalized advice, and have human interactions. Retailers will continue to want stores to communicate with customers in compelling fashion, to observe and learn about those customers (increasingly aided by new technology), and to have dispersed locations for delivery or pickup of online sales. And store-based shopping eliminates the added cost of delivery that otherwise must be shouldered by either merchants or consumers. Importantly for the retail workforce, though stores are here to stay, the mix of job activities and functions taking place within stores is changing and will continue to change—an issue at least as important as potential job destruction.
3. Store objectives and high-level strategies.
Store-based retailers currently are dealing with coronavirus-created crises, ranging from simple survival, to meeting changed consumer demands, to ensuring safety. In the longer run, at least for the large majority of retailers that will survive these challenges, their three central objectives are going to be the same ones we found in our 2018–19 fieldwork: reserve their existing market share in face of the assault from Amazon and other e-commerce sellers, develop new revenue streams, and cut operating costs. Store-based retailers have undertaken a variety of strategies to pursue these objectives. Many of the strategies rely heavily on technological investments, and we focus on these strategies in our report.
We expect fairly quick, widespread adoption by retailers of technology solutions that fall into a relatively small number of categories and criteria:
We reject the notion that if an efficiency-enhancing technology is available, it will be adopted quickly. Industry structure, the internal social organization of the firms deciding on technology use, and external social ties and influences matter. This is all the more true in an environment of double uncertainty. First, there is short- to medium-term uncertainty about the course of the pandemic and its impacts on business shutdowns, consumer buying habits, and the economy. Additionally, there is long-term uncertainty regarding which technologies will pay for themselves, and the tradeoffs between moving quickly versus waiting for later, more cost-effective iterations of technological solutions.
The trajectory of technology adoption and implementation will differ by retail subsector, by market segment, by company, and across specific technologies. The pace and extent of adoption also will vary based on shifts in overall economic, social, and policy environments.The accumulation of this set of uneven effects will, in turn, determine the overall deployment of new technologies in retail stores, and their qualitative and quantitative impacts on jobs and consequences for the workforce. We pay particular attention to “job content effects” that result when technology implementation alters workers’ mix of tasks and the nature of work. We also distinguish between “substitution effects,” which occur when machines make possible performing a job better and less expensively with fewer people; and “scale effects” that result when technology makes goods and services cheaper so consumers buy more of them. Scale effects offset substitution effects to varying degrees, sometimes leading to increasing employment in a sector.
We don’t organize our analysis around the set of particular technologies being implemented in stores, but rather break down the activities in stores into “bundles” of major functions and technology combinations. The four core bundles are: inventory management (which particularly involves stockers); checkout (primarily involving cashiers); e-commerce (the main jobs involved are the new jobs in e-commerce fulfillment); and customer interaction (particularly touching on salespeople). An added bundle is worker management, which extends across all the other functions and has impacts not just on the workers being overseen, but also on the supervisors and managers themselves. For each of these bundles, there are multiple, conflicting possible implications for the workforce: job-eliminating automation or changes in the mix of activities that make up a job—changes that are supportive and empowering for workers or shift toward more coercive, highly controlled work.
1. Inventory Management
Technology is changing inventory management in ways that are likely to increase stocker interactions with technology, but that are less likely to reduce headcount in the near term. Perhaps the most dramatic change is the use of cameras and other sensors linked to artificial intelligence systems to instantly analyze visual images and other data. These tools can be used to verify planogram compliance, spot outages or misplaced merchandise on the shelves, flag spills,detect and deter shoplifting, and send alerts about other such problems as freezer or cold cases where the temperature has risen. Other innovative systems include electronic pricing, automated receiving of boxes at the back of the store, and inventory systems tied to e-commerce fulfillment, as well as enhancements to the overall systems predicting inventory needs, and tracking and managing merchandise flow throughout the supply chain. Though some of these innovations are replacing, or will soon replace, front-line labor, their direct impacts are small and appear likely to remain so for a while. The larger impacts on stocker jobs will be a growing need to interact with machines, including being directed and paced by them.
Retailers seem to be gravitating toward self-checkout and decentralized checkout (by mobile employees with tablets) rather than more elaborate technologies; the COVID-19 crisis seems likely to spur the spread of “contactless” checkout. Retailers are investing in cashier-less checkout at three different levels. The first is simply increases in traditional self-checkout, in some cases with enhanced equipment. The second entails “scan and go” systems, in which shoppers use their phone or another device to scan while they are shopping, pay online, and walk out without interacting with a cashier. These first two levels do not actually automate checkout, but simply transfer the work to customers; reduced contact with cashiers, and especially the possibility of checking out with one’s own phone, gain appeal when there is fear of infection. The third level is the one represented by Amazon Go and a host of competing systems: cameras and other sensors track what items a customer takes off the shelves, and automatically charge the customer when she or he leaves the store. Scalability of cashier-less checkout still appears some ways off, but the continuing spread of the first two levels is likely to translate into fewer cashier jobs, and more jobs that combine some cashiering with other duties. Another growing shift is supplementing cash register-based checkout with staff armed with tablets who can check customers out around the store—an option for years in smaller and more sales-intensive retail outlets, but now being pursued (though on hold during the pandemic) by Walmart, Target, and others for whom it is new.
E-commerce both erodes store headcount and shifts store staffing to online order fulfillment. Growth of e-commerce has two kinds of impacts on stores. Most obviously, it displaces some store-based selling. In addition, it feeds the expansion of store-based fulfillment of online orders, both for delivery and for customer pickup. The latter trend involves a number of new technologies; one steadily diffusing use of new technology is the creation of highly automated micro-fulfillment centers located at stores. Order fulfillment creates a set of new worker roles, including picking orders, tending micro-fulfillment centers, and handing off orders to customers at curbside.
4. Customer Interaction
New customer interaction technologies can both replace store visits and make store visits more personalized by giving salespeople access to much more customer information. Technological innovation in customer interaction is evolving in several directions with distinct implications for the workforce. As with e-commerce, one possible outcome of chatbots (artificial intelligence (AI)-enhanced automated interaction), virtual reality try-ons of clothes or furnishings, and so on may be to make store visits unnecessary. On the other hand, store-based retailers are anxious to use these contacts to drive customers to stores, or in some cases to put them in remote communication with store-based salespeople to drive additional sales. And growing capacity to store and rapidly access data about a customer’s past online and store-based purchases and preferences means that store-based staff increasingly have access to “clienteling” tools that allow them to interact with customers in personalized and customized ways.
5. Worker Management
Worker management tools are varied, and create the potential for either greater empowerment and autonomy or increased surveillance and control of workers. The leading edge of technologically driven changes in worker management consists of ever-more-sophisticated scheduling systems. These increasingly are cell phone-based, and allow store-based workers to express preferences and swap shifts without having to speak to a supervisor. Newer and less widespread are task management systems that range from simple to-do lists to software that gives detailed direction and simultaneously tracks workers’ performance. Electronic communication channels can facilitate top-down communication from management to workers and even convey training, as well as potentially bottom-up and horizontal (worker-to-worker) communication. The proliferation of computer vision-equipped cameras and other sensors, including wearables and sensors that track workers’ phones, bring with them the possibility of far more pervasive surveillance of workers. Finally, large-scale decision systems can automate some kinds of routine decisions, increasing efficiency but removing managerial discretion and potentially reducing the need for multiple managers or supervisors in some store-based settings. One indication of potential change on this front is Amazon’s use of automated systems to make decisions about discipline and even termination of warehouse workers; the diversity of store tasks makes this less feasible in retail work so far.
Prediction is always risky, generalization even more so, and the current pandemic conditions make prognostication even trickier. In projecting possible store-based workplace futures, we note that those futures are contingent on decisions by retailers, consumers, and policymakers. They also are contingent on workers’ responses. All that said, some predictions seem fairly safe: almost surely e-commerce will continue to grow, traditional department stores and apparel retailers will continue to wane (partly due to e-commerce, but also to displacement by big-box and small-box discounters), and the ranks of cashiers will be further thinned. In terms of the nature of the jobs, the default option for most retailers in recent decades has been “low-road” strategies that keep labor relatively cheap and the formal credential requirements low, while intensifying monitoring and control of the workforce along with increasing work tasks. Reports of employers’ inadequate practices regarding worker safety during the pandemic underscore this longstanding pattern. That history suggests that, to the extent store-based retailers choose their technology future without new regulatory guidance, they are likely to continue to make low-road choices.
Heightened use of new digital technologies is likely to change both the nature and the number of jobs in store-based retail. The pandemic brings the possibility of speeding changes like the shift to e-commerce, but the accompanying recession also brings the possibility of slowing them down. In terms of changes in the nature of jobs, as we have emphasized, new technologies can facilitate both supportive and coercive changes for workers. For example, untethering cashiers from the cash register by giving them tablets can reduce the rote nature of the job, but it also can lead to increased pressures to produce under circumstances less under their control. The same is true for salespeople: instant access to background information on a particular customer can deepen their discretion on how to interact, or can be channeled in a way that is more scripted.
The content of stock clerks’ jobs seems destined for more radical change than any of the other major retail job categories. Clerks increasingly will be shifted to picking stock from store shelves to assemble orders, tending micro-fulfillment centers, and handing off orders to customers at curbside. On the store floor, they also will be more frequently prompted by “alerts” to replenish stock. As with cashiers, this could make stocker jobs more varied and interesting, but in combination with new ways of tracking work, it also could result in jobs that are surveilled, closely watched, sped up, and stressed. As tech systems take on some routine management tasks and provide more information and guidance to managers, again their jobs may come to involve more discretion—or they may themselves be increasingly algorithmically managed, more closely monitored, and more tightly controlled.
Regarding the largest occupational groups in retail, will retailers primarily use technology to enhance the supportive features of jobs, or render jobs more coercive? We heard many predictions—by retail executives, tech companies, and consultants—of upgrading of the skill and pay of retail workers as they are freed up from routine tasks, and as stores shift to a more “experiential” focus for shoppers. However, for decades, the bulk of retailers have not diverged from the habit of treating labor as a short-term adjustment variable, and labor costs as a primary cost liability. Technologists have followed suit, pitching to retailers tools that tout labor and cost savings as a primary benefit.
If retailers are left alone with tech providers to decide about technology adoption without input from other stakeholders, particularly store workers and policymakers, we have difficulty envisioning a significant divergence from this longstanding approach. We expect that, at best, there will be limited departures from the status quo of retail jobs with low wages, little requirement of formal credentials, high turnover, and fluctuating hours. And we expect technology’s new capabilities for surveillance and detailed direction of work to be added as an overlay in more and more retail jobs—with a boost coming from the heightened legitimacy of surveillance due to its new applications in reducing the spread of coronavirus.
As for the number of retail jobs, the two largest influences will be the ongoing growth of e-commerce and the newer turn toward novel applications of digital technology. E-commerce will have an important impact on the number of jobs, but a limited one. On the one hand, stores will evolve, not disappear. Most large future stores will have more workers fulfilling click-and- collect and home delivery orders, in addition to continuing to support in-person shopping and, in a more limited subset, varieties of experiential retail. On the other hand, the continuing growth of e-commerce will indeed shift more retail functions away from stores and to logistics (warehouses and delivery services), continuing recent trends. Some retail subsectors will be harder hit than others—for example, those with standardized products, portable goods, high sales volumes, and purchases that attract middle- and higher-income consumers willing to pay a premium for delivery.
In the short to medium term, the largest technology-driven job losses (relative or absolute) are likely to strike where they already have been striking: the spread of various forms of cashier-less checkout will reduce the number of cashier and salesperson jobs through a mix of automation and work-shifting onto customers. Automation will make some small, near-term incursions into the ranks of stockers, but large-scale changes are farther off, in part because the stocker job is intrinsically harder to automate (though backroom box unloading can be automated). Potentially more at risk than stocker jobs are store management and supervision positions. Store managers’ jobs are not going away, but productivity-enhancing technology may thin out the ranks of secondary managers and supervisors in larger stores. We expect the biggest changes in mid-market retail: high-end retailers rely on human interaction to make sales, while the lowest-end retailers have thin staffing and low pay, so tech investments would not bring large economic benefits.
Because cashiers and salespeople are disproportionately women, job losses in retail almost surely will hit women hardest. Women of color will be especially affected by job losses because of their concentration in the vulnerable general merchandise subsector. The loss of cashier jobs, along with such associated jobs as baggers in grocery and big box stores, is of particular concern because this is the most common entry point into retail for women and young men, including those with few or no educational credentials—and a pipeline into higher-paying jobs, including supervisory ones. Reductions in stocker ranks will have an outsized impact on Black men and Latinos. Managers and supervisors disproportionately are white and male, so job reductions in management likely would skew toward that profile as well; importantly, fewer managerial jobs translate into fewer opportunities for upward mobility from front-line work within retail. And as noted above, the growing e-commerce sector, one of the two major forces displacing store-based retail workers, is more white and male than the rest of retail (and far more so than general merchandise).
Though we see a discouraging future for retail jobs, we emphasize that the actual future of these jobs will be the outcome of a set of choices: by retailers, by customers, by policymakers, and by worker advocacy organizations and unions where they are present in the retail sector.
Policies aimed at worker and consumer safety rightly have been foremost during the pandemic, but it also is important to think through bigger-picture policies appropriate for regulating retail’s process of adopting new digital technologies over the long term. Policies could be aimed at minimizing the most negative scenarios for job quality, accentuating technology’s supportive potential, taking the edge off job displacement, and in general creating an environment characterized by fair treatment and concern for workers’ as well as managements’ priorities and needs.
One key set of issues involves workers’ privacy rights and ownership of data collected from and about workers, as well as AI algorithm transparency and checks on algorithm bias. Additionally, appropriate policies could curb abuses of employers’ growing ability to monitor and track workers’ behavior and scheduling availability, and to build decision-making systems that minimize human intervention. Policies aimed at helping displaced workers also are important. This may include support for unemployed and displaced workers as they transition to other work, as well as policies to hold large employers like retailers partly responsible for supporting the workers they displace.
Another set of broader policies could affect the retail sector’s technological trajectory. Extending sales taxes to all e-commerce purchases, as some state governments are discussing, would slow, though not stop, the expansion of online sales. And the adoption of a strict standard for considering someone an independent contractor rather than an employee, as California and other states have done, could reshape (or threaten) the business model of such platform-based companies as Instacart and Shipt. It also reduces the options for retailers to hire people as independent contractors themselves for order picking and delivery, but also other store positions. Requiring all retail outlets to take cash, a step taken by New York City and other localities, complicates totally cashier-less store models like that of Amazon Go—but does not foreclose them.
Finally, it is important to ensure there are spaces for retail workers and spokespeople advocating for their interests to have a voice in the company business strategy and technology implementation, as well as in public policy decisions that will shape future retail jobs. This can take the form of unions, other types of worker associations, or advocacy organizations. The retail sector has low rates of collective bargaining coverage, but where unions are active, they have long engaged with issues of workload, compensation, and organization of the work. In unionized grocers and other retailers, collective bargaining over the implementation (possibly even the choice) of technologies would make a significant difference not only to ultimate job quality outcomes, but also to the transparency and accountability of the implementation process itself. The expansion of the scope and strength of unionization, as well as of independent organizations like United for Respect (formerly OUR Walmart), could improve possibilities for negotiation and worker input around the implementation of technologies.
In closing, it is important for retailers, policymakers, and the public to look beyond the extraordinary circumstances of the 2020 pandemic and start setting constructive guidelines for the use of the revolutionary technologies now being rolled out in the retail sector. COVID-19 has spotlighted some of the serious problems present in retail workplaces, and we should think big about policy frameworks that harness technological change to improve jobs, rather than simply eliminate and further degrade them. Multidimensional approaches (organizing, legal, consultation, or decision-sharing processes) in multiple spheres (worker rights, industry operation, taxation) are likely to be required, given the broad range of technologies being considered and the wide array of retail functions toward which they may be deployed. Worker protections require special attention because job quality and compensation have steadily eroded over decades, as the industry has been rocked by the growth of discounting and rapid consolidation. While rapid technological change risks exacerbating job quality issues and inequalities, it also provides an opportunity to restructure retail jobs in ways that are supportive of workers and their capacities.
This report is part of a larger, multi-industry project generously supported by the Ford Foundation, the W.K. Kellogg Foundation and the Open Society Foundations.