INTERNATIONAL COMMERCE REVIEW Free for personal use only. Permission and reprints can be purchased from ECCH (

Transcription

1 110 INTERNATIONAL COMMERCE REVIEW

2 ICR(2010) 8: DOI /s ICR 2010 Published online: 23 February Are Your Staffing Levels Correct? Marshall L. Fisher Pennsylvania, USA Jayanth Krishnan Washington DC, USA Serguei Netessine Pennsylvania, USA Most retailers could significantly boost their revenue simply by changing one planning metric: by linking planned staffing levels to store traffic rather than revenue; by allocating the right number of staff to the right places at the right times. VOLUME 8 N 2 4 WINTER 2009

3 112 ARE YOUR STAFFING LEVELS CORRECT? After the cost of goods sold, expenses related to hiring, training, and employing store labour constitute by far the largest component of a retailer s costs, often accounting for 10% to 20% of sales and for more than 50% of operating costs. No wonder Walmart founder Sam Walton observed that, No matter how you slice it in the retail business, payroll is one of the most important parts of overhead, and overhead is one of the most crucial things you have to fight to maintain your profit margins. 1 Marshall L. Fisher is the UPS Professor in the Department of Operations and Information Management, The Wharton School, University of Pennsylvania. Jayanth Krishnan is an Economist in the Monetary and Capital Markets Department at the International Monetary Fund. Serguei Netessine is an Associate Professor in the Department of Operations and Information Management, The Wharton School, University of Pennsylvania. Retailers continue to struggle with this staff overhead challenge, tackling it with ever increasing sophistication. Attempts to optimize the planning of store labour are now widespread, and a variety of optimization tools have been developed. For example, Walmart is introducing a new computerized scheduling system that will move many of its 1.3 million workers from predictable shifts to a system based on the number of customers in stores. 1 This move contrasts drastically with the more traditional approach of scheduling employees based only on the previous year s sales figures and store promotions. Nevertheless, in our experience, even best-in-class labour scheduling systems lack one crucial ingredient: the ability to determine the impact of the staffing level on revenue. Even less is known about how well these plans are executed. The question is, how important are these planning and execution effects? It goes without saying that employees perform crucial functions in a store and therefore that their staffing levels should affect sales. For example, employees receive merchandise, restock shelves from the stock room as needed, mark correct prices, conduct store audits, and interact with customers during the checkout process. Moreover, depending on the nature of the merchandise, employees may need to explain product features to customers and possibly advise on substitutions, prices, and cross-selling possibilities. But it is hard to disentangle and quantify the financial benefits of variations in these activities: while store staff costs are easy to measure, store staff benefits are harder to measure, so it is easy to improve results by cutting staff costs. As a result, many stores are chronically understaffed. In previous research, for example, we found that the average revenue lift for every extra dollar in employee payroll averages $14, a hugely profitable proposition (Fisher et al. 2007). But there is a problem with averages. On average allocating total labour hours over a complete year of trading, for example a store might have perfect alignment of staffing levels to commercial needs. But this average might hide numerous mismatches depending on the time of the year (e.g., Christmas), the day of the week, the time of day, and so on. The ideal would be to vary the number of available staff to rise and fall in real time to match the changing numbers of customers in the store: to minimize the number of employees on the sales floor when there are few or no customers, and vice versa. The greater the mismatch between this ideal and what actually happens on the ground, the more adverse the effects on the store s financial performance: paying staff to do nothing on the one hand, and losing sales because of insufficient staff on the other. Many new software products have been developed to help retailers tackle this issue. However, big problems remain. Store labour plans are typically developed at corporate headquarters. The plan is then communicated to the store manager, who receives a budgeted amount that he can spend in a given INTERNATIONAL COMMERCE REVIEW

4 ARE YOUR STAFFING LEVELS CORRECT? 113 On average a store might have perfect alignment of staffing to market needs. But within this average there may be many mismatches depending on the time of year (e.g. Christmas), day of the week, time of day, and so on. The ideal would be to vary the number of staff working to fit exactly the work that needs to be done in store at that particular time. But many practical factors get in the way. So how far are retailers from this ideal? month on store employees. The store manager is faced with a number of uncertainties that he needs to combat when implementing the plan. First, there is absenteeism among employees. Second, there are sudden unforeseen highs and lows in demand due to weather, local conditions, and the actions of competitors. Third, there is employee turnover, both voluntary and involuntary. Finally, there are rigidities in employee schedules: some employees may be unwilling to work short hours or late hours; some time slots are virtually impossible to staff, and others are highly desirable from the employees viewpoint. Additionally, most employees will not be willing to work for only an hour or two per day, and they are often hired for a full 8-hour shift instead. There is another major reason for deviations between employee deployment plans and their implementation. In practice, many store managers do not understand what goes into the plan, and hence they may believe the plan is incorrect. So, when it comes to implementing the plan, they may only comply with the most rigid constraints (e.g., an overall budget) and ignore other details. Thus, once a plan has been created, there is no guarantee that it will actually be executed as specified. In fact, before this research, next to nothing was known about how well store managers execute staffing plans or about the financial consequences of their actions. Assessing the Impact of Staffing on Revenue Our research investigated two simple hypotheses: that the financial performance of a retail store is positively associated with a) the quality of labour planning and with b) how well this plan is executed. We tested these hypotheses in a three-year study of 311 stores in a large retail chain, where we measured the quality of store labour planning using the month-tomonth deviations between forecasts of store traffic and planned labour hours, and then compared actual labour deployment to these planned labour hours. In order to identify which measures to focus on, we had to think through the possible effects of staff shortages on store performance. For example, understaffing might lead to inferior signage and substandard customer service, or to more stockouts due to poor shelf restocking, all of which will result in fewer purchases on a given trip and a lower basket value. Customers may respond by supplementing trips to this store with visits to other stores (and focusing on the store traffic variable would not identify this), or they may start making fewer trips to the same store in the future. Following this logic, we concluded that actions by the store manager should have an immediate effect on basket values with possibly a delayed effect on store traffic. We therefore chose average basket size as our dependent variable. The authors thank the Fishman- Davidson Research Center, Procter & Gamble, and the International Commerce Institute for their generous financial support for this project. We are also grateful to Andy Buteux, Nicole DeHoratius, Colin Peacock, the participants of the 2007 COER conference in Boston, and numerous individuals interviewed for this project for their valuable insights and ideas. We gathered data on the variation of sales and revenue across stores, time periods, the number of transactions recorded at the checkout, the value of shopping baskets, the total number of employee hours budgeted for the store in VOLUME 8 N 2 4 WINTER 2009

5 114 ARE YOUR STAFFING LEVELS CORRECT? a given month, the employees wages and managers wages. Of course, basket size is influenced by variables other than store labour, ranging from the size of the store to the demographics of customers who shop the store. To single out the effects of store labour planning and execution, we had to exclude the impact of these other variables. We had to take account of differences in store size, product variety, and in stock rate along with differences in customer demographics of the population residing around the store. We took account of 17 demographic variables, such as household size, average household size, the proportion of households with no children, and the proportion of the local population that is Asian or Hispanic. Two skills drive effective staff scheduling: the plan and the detailed execution of the plan. Many staff plans created by the head office fail to fit local conditions. Even when they do fit local conditions, they are not always implemented well. Result: big performance differences between stores. After factoring out the impact of these other variables, our analysis still showed high statistical significance for an association between planning mismatch and full-time employee mismatch with average basket values. However, mismatches for part-time employees and store managers were not statistically significant. The part-time employee finding might be due to differences in the quality of labour between parttime and full-time employees, with the latter being more experienced and usually responsible for more critical aspects of the employee customer interaction process. Stores that were unable to hire full-time employees tended to substitute part-timers for them, possibly resulting in inferior service. We also found that some stores are consistently better at planning staffing levels to meet traffic whereas other stores are consistently better at executing a given plan, and there appears to be no correlation between the ability to plan and the ability to execute well. Managerial Implications So how important are these mismatches? Using our data, we estimate that if managers were able to reduce staff planning mismatches by 50% then the resulting revenue uplift would be 1.8% of the current chain level of revenue. From a practical point of view, it appears that improving labour planning mainly requires switching from forecasted sales to forecasted traffic as a basis for labour planning, which should be relatively straightforward to do. Many retailers use sales rather than traffic to determine staffing levels in its stores. However, staffing policies affect sales, so using sales as the metric to determine staffing levels is a bit circular. Interestingly, a growing number of retailers are installing traffic counters and using traffic as a basis for labour plainning. For example, this has been cited as a key feature of Payless Shoes advanced labour planning system. 2 Our research supports this trend. The effects of better execution are even greater: eliminating 50% of execution mismatches creates an additional revenue uplift of 2.4%. However, it is much harder to advise on specific policies to improve store labour execution because of the many practical issues associated with managing the seasonal hiring of full-time employees. The most compelling of these is compliance with labour union contracts concerning tenure of employment for full-time employees. We also conjecture that labour execution practices are largely driven by the store manager s experience and INTERNATIONAL COMMERCE REVIEW

6 ARE YOUR STAFFING LEVELS CORRECT? 115 biases. Thus, our measure of execution mismatches might be indicative of the general effectiveness of store managers, and a store manager s effectiveness affects many aspects of retail store operations that are not directly related to store labour. To sum up, our analysis suggests that stores that better schedule full-time employees to match traffic patterns in the store can achieve significantly larger basket values in our example, a revenue lift of 1.8%. However, we believe that this analysis is quite conservative and is likely to underestimate potential improvements. While we possess only monthly data, current best practices suggest that traffic should be matched with labour in hourly or even shorter time intervals. This makes great sense to us: depending on store location and product profile, different stores will experience increased traffic at different points of time throughout the day. Also, as the management of the company we studied pointed out, labour scheduling should be different for different departments within the store if they experience different traffic levels. We believe that, if such data becomes available, it is likely that even higher revenue lifts can be realized. Better scheduling of staffing to fit store traffic could achieve a revenue uplift of 1.8%. Better execution of these schedules could deliver a further uplift of 2.4%. A 4.2% revenue increase is a very large number in an industry where margins are so tight. This is a practical, operational opportunity worth seizing. The combined improvement from better planning and better execution is a 4.2% revenue increase, and this is a very large number in an industry where comparable store sales increases usually run in the low single digits. It thus appears that the old adage of having the right product at the right time and place could be true of employees as well: staffing levels should be aligned with store traffic. We also demonstrate the importance of better execution of the planned labour deployment schedule. Our results suggest that a 2.4% revenue lift can be realized by modestly improving execution. Exactly what specific actions store managers can take to improve execution is an area needing further study. Further Reading: - Marshall L. Fisher, Jayanth Krishnan and Serguei Netessine: Retail Store Execution: An empirical Study, Wharton Operations and Information Management, Working Paper, Marshall L. Fisher, Jayanth Krishnan and Serguei Netessine: Labor Planning, Execution and Retail Store Performance, Wharton Operations and Information Management, Working Paper, Endnotes 1 Kris Maher: Wal-Mart seeks new flexibility in worker shifts. The Wall Street Journal, January 3, 2007, A1. 2 See VOLUME 8 N 2 4 WINTER 2009