The amount of data available to grocery retailers is greater than ever before. POS systems provide store-level data, vendors provide product and promotion data, and third-party suppliers provide regionwide sales data. Countless other sources provide input on dozens of related issues.
This wealth of data can potentially help a retailer sell more product, increase margins, reduce waste, and otherwise maximize profitability. But the amount of data also can overwhelm, and without a carefully considered strategy and the right tools, it can become nothing more than a pile of useless numbers.
"While the sheer quantity of data collected can be helpful, ultimately retailers must be able to drive profitable insights from this data rather than simply collecting and storing it," says Anthony Bruce, CEO, Applied Predictive Technologies. "The fundamental question for all managers is to determine the true cause-and-effect impact of any initiative."
In the past, managers were limited to examining the data collected on numerous Excel spreadsheets, and at best they could make a semi-educated guess based on that information. But this method can be not only overwhelming, depending on the volume of data and number of possible scenarios, but it also lacks the precision that analytic tools offer.
Below are three examples of how data can help retailers make smart, informed decisions. Retail Leader interviewed five data experts to find out how to best use data in each scenario:
- Randy Evins, retail industry principal for food, drug and convenience, SAP
- Michael Havens, president, Havens Associates
- Gary Hawkins, CEO, Hawkins Strategic LLC
- Diana McHenry, director of global retail product marketing, SAS
- Anthony Bruce, CEO, Applied Predictive Technologies
Case 1: Maximizing the Salty-Snack Aisle
In this case, a retailer wants to use data to maximize profit in the salty snacks aisle in the weeks preceding the Super Bowl.
In the past, a store's category manager used historical data – if it was available to him – together with some "gut" knowledge about shopper patterns around the Super Bowl to decide how to stock his aisle that week. Now, the manager – or designated person in the chain of command – can input data from several sources and run a variety of models to see which stocking arrangement works best.
The data considered in this scenario, experts say, answer the following:
- How did the given products – potato chips, dip, peanuts, etc. – sell in the weeks leading up to the previous Super Bowl? The data from the retailer's own POS system would be most pertinent, but a retailer also can incorporate Nielsen and SymphonyIRI data to answer this question.
- How has location historically affected sales? Do your end caps typically produce a 10 percent lift? What about a display set up in an aisle?
- How have a retailer's own promotions, such as circulars, affected sales? If you promote your Super Bowl items in your own circular, what type of lift should you expect?
- What affinities exist among the items, if such data is available? For example, if the retailer previously placed the dry dip mixes in the chip aisle, how did they sell compared with mixes sold in their typical location near the soups? "If I see that [my most valuable] shoppers make a category purchase – salty snacks – but also frequently purchase fresh fruits or vegetables in the same transaction, I would be inclined to cross-merchandise the most popular snack SKUs in the produce department," Hawkins says.
- Sometimes affinity data is available from large manufacturers. "A company like Mars Corporation is going to know which of their products sell well with other products," Havens notes.
- What promotions are available, and how have such promotions previously affected sales? Snack manufacturers are naturally keen to increase promotions during the Super Bowl period, so their offers need to be taken into consideration. Store-level data still is essential here. However, if a previous manufacturers' promotions don't have a history of increasing lift in a given store, they might not help sales during the Super Bowl either.
- What financial incentives from manufacturers exist? Do you have funding to include a particular product in your flyer or newspaper ad? Is a vendor paying to have items put on the end cap? These issues affect overall profit.
- What trends are currently affecting snack sales? For example, if a store's demographics have recently skewed toward the healthy food crowd and data show that sales of "baked" potato chips are outpacing other salty snack sales, the retailer should take that into account.
- What has been the consumer response to external communication, such as advertising. "The fact is advertising works," Havens says. Data about consumer response to advertising is available from Nielsen, Havens says. It's important to consider consumer response, he adds, because not all demographics respond as well to certain ads as others.
president, Havens Associates
But gathering the data is just the first step. Retailers also must analyze it to find meaning from it. Depending on the retailer's system, the store's ability to analyze the data ranges from negligible to game changing. If a retailer asks managers to look over piles of data on spreadsheets and try to make sense of it, the value will be minimal.
On the other hand, a retailer with sophisticated software analytics can run multiple models incorporating several variables and end up with a fair estimate of what should happen in each case. Retailers can use the information to create a planogram for the snack aisle in the weeks preceding the Super Bowl.
Once that period begins, a retail manager would use real-time data to see if sales in the aisle are performing according to plan, and make adjustments if needed.
Evins notes that available technology allows for such rapid evaluation of data. "In the past, if I put a promotion on an end cap and it failed, there was nothing I could do about it except deal with the excess stock," Evins says. "But in the new world, I'll know several days earlier and will be able to... reduce the price or move the product to a more well-trafficked spot."
Case 2: Evaluating Three Bread Offers
In this scenario, the retailer wants to use data to determine which promotional offers from three bread suppliers he should accept. The answer depends on what effect each of the three promotions would have on sales of these brands; what effect they each would have on sales of other brands in the category; and what effect they would have on the category as a whole. Also, the retailer wants to know what would happen if he were to accept more than one of the offers.
Applied Predictive Technologies
"On the one hand, promotions have the potential to drive incremental profits through an increase in quantity of units sold. On the other, each promotion has the potential for decreased margins and the possibility of cannibalization," Bruce notes.
By running tests, either using software or in stores, retailers can determine the best option. If the retailer can try out each scenario – one bread promotion at a time – in different stores with similar sales patterns, the data can be remarkably accurate and also could be applied to similar promotions in the future.
For example, Bruce says his firm worked with a large retailer that wanted to learn if a buy-one-get-one (BOGO) promotion would work better than a 30-percent-off promotion. "By segmenting the test results based upon a variety of factors, the client determined that the 30-percent-off promotion drove a greater profit lift in markets with fewer competitors, while the BOGO program was most effective in lower income areas with younger populations," he says.
When an actual test isn't feasible, software can simulate such a test to some extent. If enough historical data for various promotions exist, software can help analyze competing offers and make predictions about the results of running the bread promotions alone or in combination.
Regardless of the software used, a wise category manager is still essential. For example, based on the data, the software may recommend a plan to take advantage of the bread offers, but the category manager may factor in the fact that a grand opening in one of her stores is going to change the game.
"No technology eliminates the category manager," Evins says. "Technology doesn't necessarily know all of the nuances going on."
Case 3: Using Data to Reduce Shrinkage
Shrinkage, whether by theft, spoilage or error, is a major issue. "Each year, retailers lose the equivalent of 1 to 2 percent of annual revenue through shrinkage," McHenry says.
But big data can help here as well, experts say.
One shrink-reduction method involves examining data to see variations in typical sales patterns. For example, did razor blade sales suddenly drop in one store? Shoplifting is likely. Is a store out of stock on an item that should be in the back room? Perhaps a paperwork error accounts for it. Does one cashier have an unusually high volume of discounts? Perhaps she is "sweethearting" discounts for friends. In each case, careful tracking of data can provide clues to the source of the problem.
McHenry says a large retailer with 1,200 stores recently used this tactic. "The company had mounting, unexplained losses which spurred them to look into employee theft at the point of sale," she says. "Large volumes of point-of-sale data were combined with human resources data to uncover loss areas and patterns. In this case, through employee training, motivation and monitoring, the retailer reduced shrink."
Data also can be used to help determine the best use of theft-prevention devices and evaluate the impact of those devices after they're installed.
Bruce says his firm worked with a retailer that learned that video cameras were effective at reducing shrinkage, but they also intimidated shoppers in some stores, hurting sales in those locations.
"The reduction in sales widely outpaced their savings from shrink reduction, making this an unprofitable program at a subset of locations," Bruce says. The retailer tailored its program based on each store's profile and increased profits by millions.
Another major cause of shrinkage in grocery stores is spoilage. Evins points out that some spoilage is inevitable – zero spoilage might indicate understocking – but too much is costly. The key is having a process in place that helps managers use data to determine appropriate volumes and monitor the situation.
The variables that might go into the purchasing decisions include historical sales patterns of each item, shelf life, promotions and other cost reductions, special orders, etc.
For example, if a regional manager looked at purchasing data and noticed that a particular store's meat department ordered much more steak than it typically sells, he might question the department manager.
"If the manager told me there is a special order, say a restaurant needs the steaks or a customer is having a barbecue, I would say that's OK," Evins says. "But if he says, 'I had my cutters here today so I wanted to have them cut all this,' that would be a problem."
The key is, data alerted the manager to possible shrinkage looming, and he was able to take steps to address it in advance.
These three cases all count on the strategic use of data to improve lift. By taking the mountain of data available from various sources and analyzing it, retailers can gain tremendous advantages.