Controlling Self-Scan Checkout Losses

Self-checkout and pay (SCO) technologies are not necessarily a recent development within retailing. Early forays into enabling consumers to scan and pay for their items without recourse to a retail staff member can be seen in the late 1980s and early 1990s. But in the last ten years, the pace of development and adoption has quickened considerably, especially in the grocery sector. In addition, far more types of SCO are in use.

While fixed SCO machines remain the most dominant type of system currently in operation (where consumers go to a fixed checkout and pay station/robot), retailers are now offering shoppers alternatives. These alternatives include:

  • Scan-and-go, where the consumer is provided with a device to scan items they wish to purchase,
  • Mobile SCO, where the consumer uses their own handheld device to scan and, in some cases, pay for their items,
  • Smart carts, some of which can automatically detect items placed within them, and
  • Whole-store SCO systems, where once a shopper has registered and scanned themselves into a store, they can pick up items and leave without any need for further scanning or interaction with a payment point, such as Amazon Go, Tesco’s GetGo, and ALDI’s SHOP&GO.

The considerable growth in the use of SCO systems has been fueled by three major factors: many retailers’ desire to reduce their labor costs by getting consumers to undertake tasks previously done by employees; brick-and-mortar retailers’ need to better compete with online shopping by reducing some of the key friction points in their shopper journeys; and the growing capability of SCO technologies, telecommunication networks, and computing to work more reliably, across more platforms and locations, and at relatively lower costs. Taken together, in many parts of the world, SCO is now spreading across various retail sectors, including grocery, DIY, fashion, convenience, beauty, and pharmacy, to name but a few.

- Sponsors -

Despite the claims of some providers, SCO introduction has not always been an unbridled win for the retailer, nor indeed the shopper, some of whom regard SCO as an unwarranted irritant in the shopping journey. In addition, many of these systems have been found to generate an increase in retail losses. The opportunities SCOs present for both malicious and non-malicious activity is now relatively well documented, including non-scanning items, misrepresenting products, switching barcodes, walking away, and so on.

An ECR Retail Loss Group report published in 2018 estimated that for each 1 percent of retail sales processed through fixed SCOs, a retailer will suffer an increase in unknown loss of one basis point. This means that if 50 percent of sales go through this type of system, then a retailer could see an additional loss of 0.5 percent of their retail sales. With some estimates suggesting that unknown losses (shrinkage) for a grocery retailer might be in the region of 1.5 percent of retail sales per year, this would represent a 33 percent increase in losses. More significantly, estimates for other types of SCO system losses were even more concerning. Scan-and-go and mobile could see the rate of losses rise higher, possibly as much as seven to ten basis points of loss for each 1 percent of transaction value.

Despite these considerable losses, for many retailers, the proverbial genie is now out of the bottle, and few SCO adopters are realistically envisaging a future without it as a major part of their checkout environment. The question, therefore, is no longer will SCO be a significant part of the retail landscape, but how can it be effectively managed and controlled to ensure it does not become an unacceptably large drain on business profitability?

In this respect, there has been growing interest in, and development of, a wide range of interventions and approaches designed to try to mitigate the risks posed by SCO systems. These have largely been focused on key points within the shopper journey—registration, the store entrance, store aisles, the checkout area, and the store exit. In addition, they have coalesced around four broad themes—the application of various technologies, the way in which guardianship is delivered, adjustments to store processes, and changes to the design and layout of stores and SCO-specific environments.

Survey of Retailers Using SCO

As part of a larger study looking at the impact of SCO systems on retail losses, the ECR Retail Loss Group commissioned a global survey of retailers to better understand what SCO technologies are being used, the extent to which retailers are concerned about the losses associated with them, and the various ways in which they are trying to control them.

In total, ninety-three retailers took part in the survey drawn from twenty-five countries. Most responses came from European retailers (51), followed by North American (29), and Australasian (11). Nearly three-quarters were grocery retailers. The rest were a mixture of non-grocery, including pharmacy, DIY, department stores, and apparel retailers.

The total sales for all respondents in 2019 and 2020 were more than €2.237 trillion ($2.44 trillion USD), accounting for approximately 13 percent of global retail sales. In terms of overall sales value, eight of the top ten and fourteen of the top twenty largest retailers in the world took part in this study.

It should be noted that the survey does not claim to be a representative sample of all retailing. Africa, Asia, and Latin America are heavily underrepresented, and of course the survey is only focused on those retailers using some form of SCO, and who are prepared to complete the survey. As such, the results must be interpreted within that context.

Survey Findings

While the survey covered a range of topics, only the results relating to the use of a range of interventions focused on fixed, scan-and-go, and mobile SCO are presented here. A free report that summarizes all results from the study is available at ecrloss.com.

Retailers were given a list of interventions grouped around four key thematic areas:

  • Technologies, such as video and weight checking,
  • Guardianship, such as training and the use of security teams,
  • Design, such as the use of corrals and exit controls, and
  • Process (such as closing SCO machines in off-peak times).

Retailers were asked to indicate which of the interventions they had either already deployed to some degree or were in the trial or planning phase of use.

Fixed Self-Checkout. For grocery retailers, the most deployed intervention was product weight checking (71%), perhaps not surprising given that it is probably one of the most long-standing and established approaches to try to restrict nonscanning by users. The next most frequently used intervention was some form of enhanced selection and training for SCO supervisors (56%), followed by two forms of design intervention: enclosed SCO areas (46%) and some form of corral system for managing the movement of customers (44%). The use of video monitors was also relatively common: 41 percent of grocery respondents were using personal display monitors at each fixed SCO machine.

A very similar picture was found with the use of interventions by non-grocery retailers, although the rank order was slightly different, and there was a lower overall level of utilization. The most common approach was enhanced selection and training for SCO supervisors (41%), followed by using a corral design (38%), personal display monitors (38%), random checks by SCO supervisors (30%), and product weight checking (30%).

A different picture emerged regarding the interventions in the trial or planning phase. For both grocery and non-grocery retailers, the emphasis is clearly shifting to trialing and testing a range of analytics, particularly those driven by some form of video-based intervention. More than one-half of grocery retailers stated that they were in the process of trialing or planning to use analytics to try to remedy the issue of customers not scanning some or all their products (53%). The next most frequently cited analytic was product recognition, for both grocery (41%) and nongrocery (35%). Grocery retailers were also focused on understanding how analytics could be used to identify incidents of barcode switching—something that many retailers believe is a growing concern. Beyond analytics, both types of retail respondents were taking time to consider how they could improve the selection and training of SCO supervisors.

For the most part, this data points to a growing trend by grocery retailers employing fixed SCO systems to invest in a twin-track approach based on enhanced guardianship and analytics to supplement their existing investments in store design, store processes, and various forms of technological intervention. For non-grocers, a broader-ranging strategy cutting across the four thematic areas seems apparent, although this may be a function of the relative immaturity of SCO use in this part of retailing—a process of catching up with the already established approaches found in grocery.

Scan-and-go. Respondents who had deployed or were trialing scan-and-go systems were asked to indicate the types of interventions they were using to try to manage and control the associated losses. Earlier ECR research had identified four areas where it might be possible to apply interventions to control scan-and-go: at the point of registration/entrance to the store, while the user is in the shopping aisle, at the point of checkout, and when the user is exiting the store.

Currently, the checkout area is where the greatest number of possible interventions have been identified (seven out of twelve covered in the research). Within this area, partial rescans of audited users was the most frequently deployed, with over two-thirds of retailers who use scan-and-go implementing this intervention (69%). In support of this strategy, almost the same number were employing an algorithm to select customers for audit (64%). A significant majority of respondents had also invested in carrying out full audits of selected users (62%). Undertaking these audits and achieving a good outcome for both the retailer and consumer is challenging for the staff employed to perform them, so it was perhaps not surprising that nearly two-thirds stated they had invested in enhanced selection and training for these staff (62%). Approaches focused on customers using the scan-and-go system in the aisle were less common, although one-half said they were using covert staff to monitor customers (50%).

In terms of interventions currently being trialed or planned, the top five were all based around the checkout area. Of those, four related to how the customer audits were delivered: introducing some form of algorithm to select users (20%), enabling staff to have the option to select users they view as potentially suspicious (21%), making use of full rescan audits (18%), and finally bringing in a program to carry out random full rescans (15%). This last option is often viewed as the most reliable way in which retailers can begin to get accurate data on the scale and extent of scan-and-go abuse by users, although it is potentially a driver of consumer inconvenience as well. As with fixed SCO, improving the selection and training of staff with responsibility for managing and controlling scan-and-go was frequently cited by respondents as an approach they were currently piloting or planning to use soon (21%).

Mobile SCO. A very similar picture to scan-and-go was found when it came to the types of interventions that had been deployed for mobile SCO. Once again, those focused on the checkout area were the most adopted: partial rescans (60%), use of algorithms to select users for a check (53%), full rescans (41%), and enhanced selection and training of supervisors with responsibility for these systems (47%), and slightly over one in three (36%) had deployed covert security staff to monitor users.

In terms of interventions that were currently being trialed or were in the planning phase, the most common was the use of an exit-point-validation process (19%). This is perhaps understandable given the nature of a mobile SCO system, especially if users are given the opportunity to pay anywhere in the store. Imposing any form of control in these circumstances is extremely difficult, so the introduction of an exit-validation process would seem an understandable strategy to adopt, although it may prove challenging to deliver in some retail environments.

Like with the scan-and-go systems, the use of audits was also dominant, such as using an algorithm to select users for a check (12%) and bringing in some form of partial rescan process (12%). In addition, evidence also showed that some respondents were investigating use of their mobile app interface to message users to reinforce appropriate behavior, such as reminding them of the importance of scanning all items (12%). Finally, as with all the other forms of SCO systems in this report, respondents were keen to prioritize improvements in the selection and training of those tasked with managing and controlling the systems (16%).

Concerns about SCO

Few significant retail business choices have no downsides. Moving to open product displays in the early twentieth century brought participating retailers a significant growth in retail sales, but it also spawned a dramatic increase in shop theft. Getting rid of routine till/cash register audits saved money on staff costs, but it also facilitated an increase in cash theft by staff. The same is true for the introduction and use of SCO.

In the early days of the latest wave of SCO use, many of the technology providers tried to argue that it was a win, win, win for the retailer—lower labor costs, happier customers, and lower levels of unknown loss (shrinkage)—the proverbial “no brainer” when it came to business choices. Of course, the reality has proved to be somewhat different. While the labor savings for many has been profound, the road to customer happiness has been somewhat rockier. And as the ECR research paper on the scale and extent of losses associated with SCO found, retail losses have gone in the opposite direction to that claimed by the SCO providers.

For those tasked with managing retail losses, early concerns about the extent to which SCO systems provide increased opportunities for both malicious and non-malicious forms of loss—and that these opportunities are now available to a much larger proportion of the shopper population—were well-founded. The current research captured this ongoing concern about the impact of SCO systems on retail losses. Two-thirds of respondents thought it was becoming more of a problem to their businesses, while only one in nine thought it was becoming less of a problem.

This growing concern is likely driven by the ongoing growth in SCO system use. Numerous respondents participating in this research commented on their organizations’ aspirations to drive 80 percent or more of their transactions through SCO systems in the next few years. In addition, a generation of shoppers have now become extremely familiar with this technology, likely moving from being wary and risk-averse to seasoned and opportunity-aware users, with the archetypal self-scan defense of “I thought I had scanned it” now becoming engrained in the shopper psyche.

This concern was reflected in estimated amounts of store losses that may be accounted for by SCO systems. The survey found that between one-fifth and one-quarter of all store losses may now be due to SCO, with some respondents believing it could be even higher. Of course, many of the organizations investing in this technology may well have enjoyed a labor cost-reduction dividend to offset this increase in retail losses. But past savings are soon forgotten, and inflated unknown losses are likely to draw increased attention toward how these systems can be better managed and controlled to ensure that they remain a business choice that contributes positively to profitability.

Investing in Interventions

Earlier ECR research reviewed the various ways in which losses from SCO systems might be better managed and controlled. This study provides a more detailed review of the different approaches that have been deployed or are in the trial/planning phase. The research clearly highlights the need for significantly different approaches depending on the type of SCO in operation. The control toolbox for fixed SCO is quite different from the one used for scan-and-go and mobile SCO systems. It also shows how approaches are likely to change in the future.

For fixed SCO now, a picture emerges of a multi-layered strategy using elements of guardianship (capable SCO supervisors), technologies (weight checking, public and personal display screens), design (use of corrals, entrance and exit controls), and process (controlling the number of open checkouts). For scan-and-go and mobile SCO, the emphasis is much more on using controls at different stages in the shopper journey, primarily at the point of payment, but also focusing on the point of registration/entry to the store, while the user is selecting products, and when they are exiting the store. Presently, using some form of audit check at the point of payment, often driven by a user- selection algorithm, is the dominant strategy for these types of SCO.

Looking into the near future, the data clearly points to supplementing the multi-layered strategy for fixed SCO with a twin-track approach: growing the investment in a range of analytic-based technologies capable of identifying malicious and non-malicious behavior (such as not scanning products, misscanning or switching products, switching barcodes, and repetitively scanning the same, low-value item), and investment in selecting and training SCO supervisors. For scan-and-go and mobile SCO, the data points largely toward more of the same (using audits and algorithms) and an emphasis on better guardianship to deliver this strategy.

As with much in retailing, managing SCO effectively will be about balancing often competing priorities—improving customer service and convenience while limiting retail losses. Certainly, some of the emerging approaches can address this issue. Improving guardianship is a good example of something that can combine the two, as can some types of analytics that can reduce the rate of false positives, improving the checkout experience. Achieving this, however, will require both a cross-functional approach and a strong dose of organizational reality. Examining the impact of any business choice needs to account for not only the benefits but also the negatives that may arise. Otherwise, like many retailers currently using SCO, those tasked with managing retail losses will be constantly playing catch up.