Loyalty Marketing

This final chapter focuses on the area of loyalty marketing. The author cites his experience in loyalty programs across various countries, and notes that the one constant is the need for huge amounts of data to accurately analyze the behavior of individual customers.

The gut-feel approach to loyalty marketing has given rise to many myths and misconceptions, and very large-scale mistakes. Because loyalty is a critical issue to sustaining a profitable retail operation, it should be based on sound analysis rather than assumptions about consumers' motivation.

Loyalty Programs

Loyalty programs are intended to increase the frequency and amount of purchases. They don't work for every retailer, and even when they seem to be working, closer investigation may show that they do not build loyalty to the store, but merely loyalty to the program itself.

During the heyday of loyalty programs, the typical consumer carried seven or more loyalty cards in their wallet or purse, and often for competing firms. They would use the card to obtain discounts at merchants, but having a card did not incline them to patronize a specific merchant: presenting the card was an inconvenience used to obtain an occasional discount.

Even if a loyalty program fails to build loyalty, it does at least help retailers to correlate purchases to a specific customer and to monitor their purchases over time, which yields valuable information to a retailer. It enables them to differentiate their best customers (the 30% that provide 70% of revenues) from the occasional shopper who cherry-picks discounts (even to the point of being unprofitable to serve). Those are, in fact, the extremes ends of customer segmentation:

The middle group is often broken into three segments to provide a total of five segments that are considered when marketing.

There has long been the belief that these segments represent an evolution, in that pickers will over time purchase more items and eventually become occasional customers, then move through the ranks to become loyal ones. Data suggests that this does not happen as regularly or dependably as assumed: some pickers remain pickers.

The pickers also represent a danger to retailers: they are often served at a loss, in hope that their future value will compensate for the loss taken during the picking "phase," but what is often found is that the picker can destroy your margin by purchasing discount merchandise and receiving rewards on top, to the point that a retailer is losing money by catering to them.

Naturally, the loyalist segment is most desirable - they are truly loyal to both the store and the program, and you always want more of them. But you should also be aware that your competitors want them as well - and many customers across all segments split their business between you and your competitors, with every retailer trying to gain more of their business.

Retailers are of two minds about rewards programs. Some believe that offering rewards to frequent shoppers is counterproductive - they will likely shop with you anyway and any added incentive is lost profit. These retailers offer discounts and rewards only to new or infrequent customers in hopes of evolving them into loyalists. Other retailers seek to reward the segments according to the profit they contribute, giving the best details to those who are already loyal in order to maintain their loyalty. The author suggests the latter approach is the better, as it is a form of loyalty maintenance to safeguard your best customers against competition.

(EN: This is difficult to gain support for internally. It's easy to measure the additional business you get by pandering to the infrequently shoppers, but impossible to measure the business you might have lost if you neglected customers. Until lit happens, it cannot be measured.)

The author mentions a technique for measuring market share. Largely, it's comparative: when a given store is getting smaller and less frequent sales than others that serve the same demographics and trade area, it's a red flag that the store in question is not getting as much market share. This is a sign, but not sufficient evidence: you also have to consider the impact of competition in the area as well as the economic climate: a store may be losing sales because of an economic depression, or because competition is more fierce.

In those instances, it is important to first reconsider your expectations, as they may be unreasonable, and then retool your tactics to address the specific issues that a given store is facing in a given market. You may even find, as the author has, that you have to use a different approach for each ZIP code a store serves in order to be successful in growing market share. And meanwhile, you must be cautious of taking your existing customers for granted, because the competition is constantly trying to steal them.

The author remarks that retailers often look into loyalty programs because sales are sagging, and they assume a loyalty program is a competitive tactic. (EN: his language suggests that he feels this to be a misconception, but he does not elaborate on why it might not be so.)

In grocery and mass-merchandising, loyalty programs may offer considerable discounts on certain items, on the belief that a customer who is lured in by the special price on some things will also shop the rest of the store and buy other items as well. Cherry-pickers do not do this, but loyal and average customers may. It's especially important to use basket data to identify those products that have a strong affinity to other products so that this tactic works - if you discount all items in a basket-set, you are harming your bottom line and encouraging cherry picking behavior.

Quote: "The benefits of a properly managed program are numerous, but the dangers of building the program incorrectly are enormous."

Loyalty programs enable firms to use the data gathered from shoppers to build programs that are specific to a given household. Marketing without this data is based on the assumption that groups of customers who live in the same ZIP code area are identical - though common sense would indicate that no two are exactly the same, and have different needs and shopping habits, there is no way to accurately or effectively differentiate.

Developing a loyalty program "is one of the most complex and difficult projects" for a firm to undertake. It generally beings by offering discounts to get shoppers to adopt and use a card, and over time gathers data about the shoppers to refine the offers made to individual participants. The basic tactic is to promote items that a household is likely to purchase frequently, and to promote other products for which the targeted product has affinity. Eventually, this enables you to broaden the types of products that you can successfully market to a given household, convincing them to purchase from you items they might be habituated to obtaining from other sources.

This process takes a great deal of effort and a great deal of time - but it should not be rushed. Prematurely or inaccurately targeting customers through a loyalty program undermines the value and credibility of the program itself. The author suggests it may take a few years to aggregate data and another few years to produce measurable results. Ramp up too quickly, and you may never see the upside potential.

The author mentions different approaches to loyalty programs:

Also consider that loyalty programs are in broader use. Not only do retailers use loyalty programs, but many credit card companies use them as well, and some manufacturers are experimenting with loyalty programs.

The author provides a list of random topics for loyalty programs:

Types of Loyalty

The author returns to the pareto principle, but does not cite a specific source of his figures, only that it was "determined years ago" that ...

Sales can also be seen as a way to reward good customers and maintain their loyalty. This is where card programs provide the retailer the ability to do so, without making the same deals available to cherry pickers. A deal can be structured or promoted so as to reward only your best customers and be unavailable to the rest.

Retailers seek to foster what the author terms "relationship loyalty," which leads a customer to purchase from them for reasons other than price sensitivity. This often fails, and marketing efforts (advertising and loyalty programs) merely create "deal loyalty," which is strictly driven by price.

A "deal" is defined as any price below the retailer's everyday price for an item - whether it is a sales event, coupon, or other tactic that creates them. The desire of the retailer is to use a deal as an entree to a relationship - to attract new customers or get existing ones to buy something unusual in hope that they will continue to purchase even after the deal has ended.

The author has used the term "cherry pickers" to describe customers who are interested only in deals, purchase only items that are discounted, and do not purchase other items or visit the retailer regularly. The "deal loyal" customer is a habitual cherry picker who damages the store's revenue - sales figures are increased, but profit-per-customer decreases overall due to their influence.

That's not to say that a cherry picker won't become a regular customer, just that this cannot be assumed. Converting such a person into a relationship-loyal customer is a difficult process, and some never will convert.

One tactic the author mentions is offering deals on merchandise needed for an unusual situation. For example, a new parent has never purchased merchandise for a baby - and if you can promote to them at the right time, you can become the merchant they associate to this need, and continue to shop even when they have other children, and when their children grow older.

(EN: "Life event marketing" was in vogue for a time, and it's an interesting concept: there are various incidents in a person's life, such as moving out of their parent's house, graduating from college, moving, changing jobs, getting married, having a child, etc. where their buying patterns are interrupted and they must make a lot of new decisions. This is a very fortuitous time for marketers to capture their patronage. It's gone out of fashion, but it seems to have a great deal of merit - it's just hard to detect and execute upon.)

A more common tactic is to promote high-frequency items to lure in a customer, get them to break from their routine (and their typical suppliers) for the sake of a good price, then attempt to capture more business with them. The immediate benefit is the other non-deal merchandise they purchase, and the longer-term benefit is acquainting them with the retailer's inventory of goods, such that the customer forms a positive impression and returns to purchase regularly.

The author notes that a card program, which identifies the customer at the point of sale, is essential to being able to measure the effect of loyalty programs. He suggests that you must ensure rewards are tied to card usage only, not trip frequency or spend in general (EN: this doesn't make sense to me, as the card is the means to measure frequency and volume of sales - so tying rewards to "card usage" without either of those other metrics seems bizarre. Maybe this is true of a store credit card, as I don't see it being germane to the loyalty card that is not a method of payment.)

Credit and Discount Cards

The author asserts that "many retailers" have store-branded credit cards (EN: I suspect this to be a trend that has largely passed) that have a dual function: not only do they serve as a method of payment, but they also enable the retailer to track card use.

This does add some complexity, as there is interplay between the amount of business a customer gives to a store and the amount of that business they charge to the store card. That is, some of your store's best customers may not use your store card at all, and those that do may not use it for all of their purchases. So it is impossible to accurately determine, based on branded card data alone, the purchasing behavior of an individual customer.

Ideally, analytics are based on 100% of consumer transactions, and the lower percentage of transactions that are tracked, the less accurate the conclusions based on that data.

The "standalone" card that customers use will generally capture more business: the customer may pay with any tender they desire, and use the card to earn discounts and rewards. Where a store offers both a credit and discount card, it seems a good practice to make a single card serve both purposes, but this comes at a loss because a customer will be reluctant provide the card if they intend to use a different method of payment.

Loyalty Segments: Develop Them Early

Segmentation divides a body of customers into groups of people with similar behaviors, for the purpose of more precisely targeting marketing messages. The more granular the segmentation, the more specific your appeal to each customer, and the more you can allocate budget to the most productive segments.

Returning to an earlier notion: not every customer is a good customer. You certainly want to keep the 30% of customers who provide 70% of revenue. You may wish to develop other customers in that direction. But there is invariably a number of customers to whom you should be indifferent (the occasional buyer who is highly unlikely to become a frequent customer), and others whom you should shun (cherry-pickers who damage your profitability)

The author mentions a system of decimal segmentation, which considers the top 10%, next 10%, and so on in terms of the profit (not just revenue) they provide to the store. The top groups are to be maintained, the next group is to be developed, the next to be regarded with indifference, and the last to be regarded as a nuisance. Exactly where the breaks occur will depend on a store's specific numbers.

Offers are generally based on an "RFM" system, which considers recency (last visit date), frequency (number of visits), and monetary (amount of money spent). Offers can be made to address one or more of these parameters - for example, a deal good for one week only promotes recency more than frequency or monetary measures.

Where loyalty is the goal, your individual marketing efforts are geared to maintain the behavior of loyal shoppers or increase the behavior of less loyal ones. In either case, you must assess the likelihood of success.

Moreover, you must also consider whether the change and behavior will be profitable to you: the last thing you want to do is create strong loyalty among unprofitable customers - the cherry pickers who harm your profitability. You don't want these shoppers at your store - let the competition lose money on them.

You also cannot target everyone. This is not fiscally responsible. As in all business decisions, you need to spend your budget where it will have the most beneficial effect, and carped-bombing wastes a lot of ammunition.

Finally, you should closely track what your competitors are doing. If you look only at your own efforts, you will not realize that the competitor is winning until your customers have already left, but must react defensively to match or defeat their attacks on your market.

Complexity of Rewards

It is possible to implement a loyalty program that differentiates customers by their level of loyalty, but many retailers choose not to pursue tiered-loyalty programs. (EN: The reason is given that tiered loyalty programs are seen as too costly or complex, but I suspect that treating some customers better than others is a touchy legal issue: the customer who seldom visits and buys little feels he has the "right" to get the same deals as your best customer and will hector employees, publicly complain, or even pursue a legal remedy if the reason for paying a different price is not simple and clear.)

There's a bit of fussy detail about the different kinds of rewards that can be offered to loyal customers, but it becomes a bit tangled between the various factors:

Generally, retailers implement loyalty programs in phases: the first phase generally involves discounts on merchandise, with or without promotion. The next might involve a reward points system. The next might offer frequency discounts, etc.

Having multiple rewards available to different customers based on different RFM (recency, frequency, monetary) behaviors can create a rewards system that is highly customized to each shopper but become tangled and complex when considered in aggregate. At this point, it becomes difficult to track what rewards are offered to which customers and to assess which of several concurrent rewards is motivating customer behavior - though strong analytics can analyze basket data to determine which of several rewards is likely to have effected a given purchase.

Except for the use of blind rewards, keeping customers informed of the rewards they have earned is essential: if a customer is unaware of a reward, it has no motivational value, and constitutes a loss to the retailer if the customer who was willing to pay full price for an item receives a surprise discount (though, arguably, it increases loyalty over time).

(EN: A case study in this might be JC Penney, who recently simplified their sales and discounts on items, but before then it was commonly remarked that customers would often get substantial savings at the register because of promotions they were unaware of while shopping. I've not seen any formal studies, but I strongly suspect that this confused some customers and delighted others.)

The author mentions that Kmart printed savings on the receipts as a way to inform the customer of how much they had saved by participating in the loyalty program, and while the program "seemed to work for a while" they could not substantiate a positive effect. (EN: The presumption here is that customers pay close attention to their receipts - but I suspect for most customers, this is not a common practice.)

The "majority of retailers" feel that competitor price-matching is a loyalty program, but the author disagrees. This practice is merely mass discounting without targeting, and while it is a counter-measure against competitors who attempt to lure customers with lower prices, it does not constitute a loyalty program because it is not specific to certain customers nor is it measurable over time.

In the same way, any mass merchandising action is not a loyalty program. A campaign that offers a $10 discount to anyone who spends $100 is a short-term tactic to get customers to increase the total sale (anyone who has $95 worth of merchandise sees the logic of grabbing an extra item to get the discount), it does not target specific customers over a longer period of time.

A side-note: a large retailer can engage manufacturers in a loyalty program. Where discounts or bonuses are tied to specific items, a manufacturer may be approached about offering the retailer a discount on the wholesale cost to offer his brand rather than a competitor's item. This can help defray the cost of loyalty rewards.

The ultimate goal of any loyalty program is to increase the profit made from a given customer by capturing greater share-of-wallet. This is measured by RFM scoring: how recently they have shopped, how often they shop, and how much they spend per visit.

(EN: I find recency to be overrated. Volume and frequency contribute directly to the profit, but recency is entirely incidental. Retailers get nervous if a customer hasn't shown up in a while, fearing they might be going to a competitor, but unless there has been a change in frequency or volume, this is a baseless fear. A customer who spends $250 every three months is a better customer than one who spends $75 each month - and there's no cause for panic because he hasn't dropped by in two and a half months.)

The collection of customer contact information enables loyalty marketers to target offers to an individual customer - the value of which is that the promotion has much higher relevancy to the customer. The author remarks, as an observation, that customers dislike junk mail but appreciate receiving offers that are specific to their tastes and needs. (EN: Other sources provide more formal research and observation studies that support this casual observation.)

A side-benefit of this is that it often provides data to inform more general advertising efforts: where unknown members of the general public have similar traits to a known customer (income, household size, etc.), the insights gleaned from participating customers can guide promotional efforts that are more broadly targeted.

Without personal information (a loyalty card presented at the register), a retailer can use payment information (debit or credit card numbers) on the assumption that customers habitually use the same card at the same merchant. The obvious limitation is that the assumption is not strictly true (a given person may use different cards at different times and make cash purchases), so you don't have a precise indication - but it is still a good rough estimate (and better than nothing at all).

A more significant limitation to using payment cards to identify customers is that the customer remains largely anonymous: you do not have an address to send a promotion, do not know when they walk into the store that they are a regular customer, etc. There are still opportunities to market: rewards can be offered at the register, after the card is swiped (these function as "blind" rewards) and you can offer them an incentive for their next visit.

Case Study: Kmart's School Spirit Loyalty Program

The author describes a program implemented by Kmart that was marketed in local communities, in which donations were made to local schools based on the purchasing behavior of consumers in a specific area. In aggregate, this program resulted in millions of dollars donated to K-12 schools across the US in a single year, a "couple hundred thousand" raised in the program's first three weeks.

It's also mentioned briefly that the donations were in the form of gift cards, but without much detail. It's likely Kmart provided cards to schools or school districts to be given to teachers to purchase supplies at Kmart itself- such that the actual amount of the financial cost was less than the nominal amount because its donation was the wholesale, not retail, value of the goods. (EN: but this is my supposition - the precise details are not disclosed.)

The participant was required to enroll in the Kmart loyalty program and tender their card when making purchases - which itself gave customers an incentive to enroll in the program and got them in the habit of using a discount card, so the benefits to Kmart outlasted the one-year event.

While any parent or socially-minded individual might choose to participate, their intended target was the top segment of busy and budget-conscious mothers, who were incidentally a significant target segment for Kmart in general.

An additional challenge was that the team was tasked by the corporation with building a substantial loyalty program on a fairly tight budget within sixe months time. As a prerequisite to getting customers signed up, they had to rig the POS system and back-end information systems to process the cards and store the data ... which itself was a significant project.

A specific problem with an early implementation is that the card had to be scanned before the merchandise was rung up. If they presented it late, the transaction had to be voided and the merchandise re-scanned . So naturally, many customers decided at that point not to use the card, and it likely caused some to toss the card entirely. Even though the problem was eventually addressed so the card could be scanned at any time, this likely harmed their results.

The author speaks in vague terms of the "heavy analytic rigor" that went on at the corporate office to track the program, and the high level of scrutiny and attention given to the program by senior leadership.

The author is also vague in his assessment of results - they enrolled "millions" in the first month, "several more millions" afterward, and drove "millions" in incremental sales. Which is to say the program was considered a success, but he can't disclose the specific amounts.

Case Study: Australian Loyalty

At the onset, the author had misgivings about how his experience would apply to the Australian grocery industry - the buying behavior of customers is a derivative of their culture, so what is true in one location may not hold true in another, or at best it requires adjustment.

Moving into his new home, he immediately noticed the lack of storage space in the kitchen: the refrigerator was half the size of the typical American ones, and there was about a third as much cabinet space. In general, the place was designed for "more space rather than for things."

The patterns in Australia were significantly different to America: customers tend to buy as they need, shopping for a small number of items every few days rather than making a weekly or biweekly trip and stocking food at home.

The majority of customers purchased perishables (vegetables, dairy, meat, etc.) on a daily basis, with an emphasis on freshness. Australian shoppers also maintain the "pleasant excursion" rather than "time-wasting chore" perspective on shopping. The grocery-specific sortie was less common that buying groceries as a stop on a trip that included various other items.

Another major difference was the lack of parking proximate to retailers: typically, a person has to park a block or so away, which meant they had to carry their goods some distance to their vehicle, which again would reduce the volume a person would purchase. Though in his case, the store he was assigned had a small parking lot on the roof of the store.

Another difference: it is not uncommon for grocery stores to be located in shopping malls - some of them had to be accessed through a third-floor entrance likewise making bulk-buying inconvenient.

As to consumer behavior: this meant customers shopped three times as often as their US counterparts, buying more each visit. This meant that the benchmarks for recency, frequency, and monetary levels would need to be reset for Australia. (EN: It likely also meant the goals would need resetting - if customers shop once per day, you can't get much more frequency, and if they do not stock up, there will be fewer fluctuations in the amount spent caused by stocking and drawing from the pantry.)

However, the author remarked that the methods he used in the US worked very well in Australia, though it was necessary to adjust to local customs and habits.

Loyalty Programs in Australia

The author considers the Australian "FlyBuys" loyalty program, which initially used a mag-stripe card that enabled customers to accumulate points that could be exchanged for airline tickets or other merchandise. It is the largest such program, with 10 million cardholders in 5.5 million households.

The card was used at Coles, a conglomerate that operated various grocery, general merchandise, and liquor stores under various names and includes sundry other lines such as a home office supply store. Coles operates both Kmart and Target stores in Austrailia.

The program covered all these merchandisers, such that points could be earned and redeemed at any store, a great benefit to the consumer, but somewhat different in that the loyalty was to the chain rather than a particular store.

This required the analysts to combine all market data from all customers to form a more comprehensive view of each customer's behavior at each store, before any consideration could be made to the initiatives that would benefit any particular store.

It's also noted that the loyalty program was tied to a general-purpose credit card (MasterCard) that could be used at other merchants as well - so in addition to basket data for purchases from Coles, the retailer also had some perspective on the buying behavior in general. (EN: Though I'm assuming it was limited, as in the US a credit card issuer gets only the total amount and merchant name, not basket data, so Coles would only have a vague sense of how much is being spent with whom, but not on the specific items at any non-Coles store.)

The author wraps by stating "this is a great example of having a standalone loyalty card as well as a bundled credit loyalty card" and that it was "very appealing to the consumer." But here, his commentary ends: he doesn't describe what was done with the program or what results were achieved.

Other Foreign Loyalty Programs

The author presents some notes on Australia:

The author then mentions the Nectar program in the UK, which is "considered the best in class for coalition loyalty programs." Multiple retailers participate in the program and share data on consume behavior. The mechanics are that customers earn two points per pound that is redeemable for merchandise.

Necessity of Retail Analytics

In established markets, senior members in retail organizations have become believers in the notion that understanding the customer is critical to the survival of a retail business, and they recognized the importance of consumer data in doing so. It is no longer a new thing that is pushed aside, but an established practice that drives business decisions at any sizable retailer.

Customer loyalty is critical to the operation of retail - getting more revenue means getting more purchases from more buyers, and doing so requires having a keen insight into their buying behaviors and motivations. IT drives the top-line and its impact carries through to the bottom-line.

At the same time, there is no single set of universal practices that can perfectly suit the needs of all retailers - and likely there never shall be. Each retailer must choose analytics and undertake initiatives that are customized to its industry, store format, and customer base. It is only through the rigorous application of analytics that a firm can do so.

There is also the problem of loyalty programs requiring a lot of cost and effort to manage, and the fact that they do not offer an immediate bump in revenue - but instead take time for customers to adopt and for the program itself to be refined. Many retailers begin with high hopes and abandon quickly.