Loyalty: Don’t Give Away the Store

Loyalty programs are profitable—if used correctly. HBS Marketing professor Rajiv Lal discusses how grocery stores get it wrong. But you can get it right.
by Manda Salls

Frequent shopping programs that reward customers with discounts or other perks are commonplace in grocery stores, but many are not as effective in influencing buying behavior as they could be, argues HBS professor Rajiv Lal.

"The Impact of Frequent Shopper Programs in Grocery Retailing", written by Lal with HBS colleague David Bell, was published in Quantitative Marketing and Economics last June. Lal discusses his findings with Manda Salls.

Salls: Frequent shopper programs have become part of the shopper's landscape. What practices do you see used most often by retailers? Why did you decide to focus your research on the grocery retail business?

Rajiv Lal: The programs I see most often are the simple programs where you get a card, and use the card to get promotions. You see these in all types of retail business: grocery stores, drug stores, airlines, clothing. What you see, especially with grocers, is that they use these cards to give discounts at the checkout. In the process of doing this, they have accumulated a lot of data. What they do with that data is anybody's guess. Other retailers, such as Talbots or Best Buy, offer a different type of reward program where you spend X amount and receive an offer or discount of some kind.

When you talk to retailers about their frequent shopper programs, they'll say the programs work and are very profitable. Then you ask, "How do you know it's working?" The typical way people measure this is to see if sales went up when they offered the promotion. But it isn't clear that the promotion actually caused this. We wanted to get real data.

It is difficult to get sample data from retailers; it takes a lot on their end to supply samples. We chose grocery retail because it is an industry that typically engages in loyalty programs, and it is possible to get a good sample from them. We sampled one supermarket store and one supermarket chain.

Q: I have at least three loyalty shopper cards in my wallet. What percentage of shoppers participate in these types of programs? Who are retailers targeting?

A: Eighty percent to 90 percent of all sales are on these cards. About 65 percent to 70 percent of the people making purchases (store traffic) are using the card. If a retailer is not getting these kinds of numbers, then the program is not effective. You just don't have enough data. If you are losing track of 20 to 25 percent of your sales, you have a big problem.

By using rewards programs, retailers give away things for free that their best customers would have bought anyway.

With such a large volume of customers purchasing off the card, you have to worry about having a sensitive business that can manage two separate customer pools. One is the pool that is coming in on the card. The other is the pool you don't know much about that is not coming in off the card. Without [data from] the card, you may not treat these customers as well as you would treat them otherwise. A retailer's first effort needs to be to get the right numbers, so they know they are getting enough data.

The targets for loyalty programs vary from retailer to retailer. Some focus on getting more wallet share of customers; others may have an elite circle of customers that they want to treat especially well.

Q: Your research showed a surprising difference in the way rewards programs changed the buying behavior of a retailer's best customers versus lower-spending customers. How does this shift conventional wisdom?

A: Conventional wisdom says that loyalty programs are best utilized to reward your best customers. This is not true in the grocery retail business. Loyalty programs do not affect the behavior of best customers. By using rewards programs, retailers give away things for free that their best customers would have bought anyway. Retailers thereby begin to lose money on the program.

In a grocery store this is particularly bad. The marginal cost is almost 75 percent, so every time you give a dollar to someone in the form of a turkey, or a ham, it's costing you 75 cents. In contrast, if you look at offers from the airlines, the cost of miles rewards programs are virtually zero because airlines can predict the load factor on a flight. If it is too high, they can always give you a different flight.

The behavior you do end up affecting is that of your worst customers. In fact, if you are looking at the profitability of the program, the profitability is coming from the behavior change that you see in the worst customers rather than the best customers.

Q: Should retailers change their approach to big spenders?

A: The first issue here is that you do want to reward your best customers. Rewarding them by giving discounts is very expensive. The best way to reward your best customers is by making them an offer that is highly valued by them, but doesn't cost you a lot of money. For example, allocate more services to good customers. This could mean a designated line at the deli counter, or a ten items or less checkout line. You already have the buying behavior you want. So think about your costs, and make sure you aren't changing good behavior.

The second issue, which I think is even more important, is trying to truly affect customer behavior. This is why the rewards program used at Harrah's Casino is so successful. They ask a) How much is this customer potentially worth? and b) what would it take to make the customer spend more? It isn't about rewarding existing behavior. It is about using the loyalty program—the data you collect—to affect customer behavior in ways that are rewarding for both the customer and the store. Unfortunately, most retailers do not have any sense of wallet share.

Q: Many shoppers modify their behavior to take advantage of reward or loyalty deals. How do behaviors such as cherry-picking and stockpiling play into the equation?

A: When we looked at this data, we didn't see much of a stockpiling effect. Worst customers probably have more cherry-picking [picking off the discounted promotional items from different stores] than best customers. Reward programs should lessen the effect of cherry-picking, and thereby increase the profitability of worst customers. That is, shoppers are rewarded for spending more money at one store.

In a store like Talbots, I think the argument is slightly different. What's happening there is you have an infrequently purchased good. My sense is that when Talbots send you a coupon for $25, it is effectively increasing your trip frequency. It gives you a reason to go to the store when you may not have thought about going to that store per se. Reward programs give them a reason to go to a particular store and spend money. Once a customer comes into a store where they have purchased items in the past, chances are, they will spend. And they will spend money that they may have spent in a different store. So your major accomplishment is getting them into the store one more time.

Q: Often you'll see that when one store offers a special, its competitor will come up with a similar deal. Does this copycat strategy work in attracting customers? Is there anything retailers can do to counteract this?

A: If you are the only store in town offering a rewards program, our data shows that you will be quite successful. On the other hand, if the competition is also offering these programs, then it is not surprising that you won't see as much change in customer behavior, and the cost of the program will increase. That is why it is crucial to find ways to reward loyalty through non-price offers. In competitive environments, the cost of affecting customer behavior is much higher.

Retailers should use card data to think of services in a broader context. Look at the times and days your best customers tend to shop. If your best customers shop at the deli counter, then make sure the deli counter is fully staffed at these times. The same goes for checkouts. Make sure plenty of checkout counters are open when best customers are shopping.

It is crucial to find ways to reward loyalty through non-price offers.

Do a better job merchandising. Look at who is buying what product. Make sure that you are well stocked and have given a lot of shelf space to items that your best customers purchase.

Do the same for slow-moving items. For example, prunes are often an unprofitable category, and a grocer may be tempted to take them off their shelves, or give them less space. Yet prunes are often bought by the best customers. So the thought is that if you get rid of prunes, these customers might go somewhere else looking for prunes, and this might take these customers away from the store. By having this data we are able to avert the mistake of getting rid of a slow-moving item that was, in fact, part of the grocery basket of some of the best customers. The same is true for items such as blueberries and cherries. There are some customers who really look for these things. Only by having this data can you really see what is going on.

This same logic applies to other retailers, such as department stores. If you look at your data, the best customers are probably the customers who end up shopping for china or silverware—departments that some of your worst shoppers may not shop in. Service levels in those departments, despite the fact that they are costly, may have a big impact on the shopping behavior of your best customer. These are the customers that, if they don't get the right help buying crystal, they are going to take not only the crystal business out of your store, but their other business too.

Many stores would just look at the crystal business by itself. They might say that at such high service levels, they are not profitable. But you have to ask the question, if the best customers are not served well in this department, what impact might it have on their expenditures in other departments? This is an area people don't think about much, because it is not obvious on the outside. Clearly, stores must learn to use their data in creative ways.

Q: Overall, did you find that these programs were worthwhile? How can retailers make rewards programs more profitable and effective?

A: Many retailers use loyalty or rewards programs as a defense mechanism. They just try to copy what the competition is doing. If that is all that you do, then you are not clear in your own mind about what can be done with the help of these programs. Unless you have given some thought beyond copying, you are probably losing out on the real benefits. I would categorize real benefits of these programs into four different buckets:

  • Reward your existing, loyal customers, without incurring a major cost by doing so.
  • Positively affect the behaviors of not-so-loyal (worst) customers, especially middle-tier customers.
  • Find out who the high potential customers are, the ones whose behavior is worth modifying. That is where I think the biggest bang for the buck is.
  • Effective, targeted merchandising and service.

I think most companies have a hard time making good use of these rewards programs. They do not have the wherewithal to really use their data in profitable ways.

Q: What other research are you working on?

A: My most recent research focused on the international expansion of retailers. In our study, we tried to show that U.S. retailers have more difficulty expanding their markets than their international counterparts. I've just started doing research on the future of department stores. Their future will strongly affect the future of vendors and specialty stores.

About the Author

Manda Salls is a Web editor and content developer at Harvard Business School's Baker Library.