by Byron Sharp
Table 10.2: Break-even scenarios
For a brand with a 30% contribution margin at normal price For a brand with a 40% contribution margin at normal price For a brand with a 50% contribution margin at normal price
Price reduction (%) Increase in sales needed to match the current contribution (%) Increase in sales needed to match the current contribution (%) Increase in sales needed to match the current contribution (%)
5 20 10 11
10 50 33 25
20 200 100 66
Such large increases in sales are needed because a price reduction results in selling lots of volume to people who would have bought the brand anyway, at full price. Therefore, the incremental volume has to be high to 'pay for' the lost margin on what we can call the baseline volume. This is illustrated in Figure 10.1. There are some obvious spikes in sales that are one-week price specials.
Note that the blue bars represent the baseline or normal level of sales that would have been made anyway if the price promotion had not occurred. These normal sales would have been sold at full price, not a discounted price.
Because the contribution margin is reduced by an exponentially increasing rate from larger discounts, the implication for managers employing price promotions is that they should make them as shallow as possible (e.g. 10% off, not 20% off). Even if this does not result in such high volume increases, at least the volume is making more contribution to profit.
There is another factor that can impinge on the short-term profitability of a price promotion: unit sales may decline immediately afterwards. This can happen because the promotion has induced some consumers to accelerate their purchases. Empirical evidence suggests a trough can occur in sales after a promotion (Mace & Neslin 2004; van Heerde, Leeflang & Wittink 2000). If this happens, the promotion has borrowed some full-price, full-margin sales from the future; this dissipates its potential profitability. Managers should allow for this effect when evaluating promotions.
Finally there is the very likely competitor reaction, that is that they will match promotions in order to maintain their sales. In which case every extra sale you make while on discount represents a future full price sale that will now not occur. This is a very high, and largely forgotten, cost of price promotions.
Figure 10.1: Sales for a leading cereal brand
Source: Data supplied by Synovate Aztec.
We now compare pricing with advertising as a means of stimulating demand (and maintaining marketplace position).
Comparing price versus advertising
Price cutting has a large, direct effect, but a very small reach. Advertising has great reach but a small direct effect at the level of the individual consumer. Advertising arguably also has a long lasting effect, because it works on underlying propensities, whereas price promotion does not affect ongoing propensity and does not have favourable long-term effects.
It is certainly true that in-store price promotions are highly targeted: they only reach the buyers who are in the market for that category in that week. It would therefore seem they should be very efficient. However, this targeting through price promotions is very expensive – 10% or 20% off the price of every item sold! Price cutting gives a lot away to people who would buy the brand anyway.
For manufacturers who sell through distribution channels, price promotions can occur with several levels of retailer support, and retailer support has implications for reach. The lowest level of support is an in-store price cut with only the ubiquitous 'special' ticket on shelf. An in-store price cut has the lowest level of reach, because only consumers who happen to be in the store and look at brands in the category at the time will notice it. The number of consumers who would do so in any given week comprise a very small proportion of the market. For example, in grocery markets, in a typical promotion period of two weeks, approximately 60% of households buy bread, 20% buy soup and 10% purchase a category such as shampoo – but these figures are the total penetration across all retailers. A price promotion in any one retailer will only reach a fraction of these buyers: given a retailer market share of say 10%, such a price promotion in that retailer will only reach 10% x 60% = 6% of bread buyers, 10% x 20% = 2% of soup buyers and 10% x 10% = 1% of shampoo buyers. So this sort of price promotion has very low reach.
The next level of in-store support involves an 'end-of-aisle' display. Nearly everyone in the store at the time could notice this display, but there is an upper limit on its reach, being the proportion of the population shopping in that retailer in that week – which is often a small proportion.
The third level of distributor support involves the brand being advertised in catalogues or through the retailer's mass media advertising. This takes the price promotion, and the brand, to a much broader audience. Arguably, the reach of this level of support is as large as the reach of mainstream advertising. However, advertising a price promotion so widely risks lowering the reference price for the brand to a wider group of people. Therefore, it might be preferable to organise an off-shelf display or retailer advertising without the price cut. Problem is, price-cutting is usually mandatory to obtain an off-shelf display or to feature in the retailer's advertising.
To summarise, the following observations have been made in this book:
•Reach is important for brand growth.
•In-store promotions generally lack reach, but retailer advertising has much broader reach.
•Price promotions may not be profitable, particularly when the price cut is deep.
•There is the possibility that reference prices would be eroded, particularly from deep price cuts.
•However, manufacturers often feel pressurised to participate in price promotions.
Based on these observations, if a manufacturer had to choose between a promotion that emphasised deep price cuts or a promotion that focused on advertising the brand, the manufacturer should select the latter. This is because the brand would then feature in retailer advertising, which works more like normal advertising and attains reach for the brand. This communication may have to be linked to a reduced price, but the price need not be as low as it would be if the emphasis was all on price and less on communication support.
Managing Price Promotion Addiction
A price promotion will deliver an increase in sales in most cases. How large this sales increase is and whether the promotion totally decimates all profits depends on factors such as the size of the brand and what its price is relative to the competition before and after the price cut. Current evidence suggests there is no long-term increase in sales as a result of the spike in sales during the price promotion: buyers return to their pre-promotion buying patterns and sales return to pre-promotion levels. This means that a price promotion's effect on sales and profit can be easily evaluated – the analyst need only consider the sales spike during the promotion period, any possible trough in sales immediately afterward, and likely competitor retaliatory promotions.
In many cases, managers have other motives for running price promotions than just chasing sales spikes. A commonly cited reason for manufacturers to run promotions is to please or placate the retailer. In the long run this may be the only justifiable reason for price promotions, but if maintaining retailer relations is the objective of a price promotion then this should be the metric that is evaluated – yet such evaluation is seldom done in any formal sense.
To conclude, much price discounting is significantly costly: it erodes brand margins for the benefit of the retailer and consumer. The advice to brand owners is to think long and hard about what price-cutting activity is achieving in the long term, and try to reverse the long-term trend of marketing money being transferred from brand-building activities to discounting. Ask how many price promotions are really needed to avoid delisting from a retailer, and avoid doing more than this.
Chapter 11:
Why Loyalty Programs Don’t Work
Byron Sharp
Loyalty programs are structured marketing efforts that reward, and theref
ore encourage, loyal buying behaviour. The most typical programs provide consumers with points each time they make purchases. Gradually, after many purchases, these points add up and reach a level where they may be redeemed for rewards. This points system is an incentive for consumers to purchase a particular brand more frequently (i.e. increase their loyalty) in order to gain more rewards faster. Such loyalty programs might also discourage defection from the brand as, once customers have accumulated points, and if they enjoy and value the loyalty program, they should be less likely to defect to another brand. At least that’s how the logic goes.
Huge sums of money have been invested in loyalty programs over the past decade, be everyone from airlines to coffee shops. Some marketers implemented these programs because it was fashionable to do so, or launched programs because they now had the technology to do so. However, most marketers were motivated by the hope that loyalty programs would improve loyalty. They anticipated great business results, with large increases in sales and profit. 75
However, as discussed in Chapter 1, marketing strategies that are based on false assumptions will never deliver high returns, no matter how well executed the strategies. The key assumptions that underpinned the large investments in loyalty programs are faulty. Many people assumed that loyalty levels could easily be dramatically improved; that customer defection could be reduced (to zero even); and that existing customers could be encouraged to devote all their purchasing to one brand (100% loyalty). It was assumed this would lead to substantial growth in revenue and profits. It was also mistakenly assumed that targeting a brand's most loyal customers would generate the greatest return. Enthusiasm for loyalty programs has now waned, and some firms are winding down loyalty programs. There is no evidence that outstanding business results are driven by loyalty programs, and in spite of thousands of programs strangely no-one is coming forward showing off their success.
Loyalty programs are still big business. One of the problems with large-scale consumer loyalty programs is that they are difficult for firms to exit from them. Consumers who have accumulated points do not take kindly to losing them. If consumers are told they are losing their points, they tend to place an unrealistic, inflated value on the points – even if they were unlikely to redeem them76. There are also contractual and legal impediments to shutting down a loyalty program, and there are the considerable sunk costs (including management ego) invested in the program. Firms turn out to be quite loyal to their loyalty programs, and tend to continue them far longer than rational economics would recommend.
Do loyalty programs improve loyalty?
Do loyalty programs affect loyalty? You may be surprised to hear that yes, they do. However, the effect is very weak. The Ehrenberg-Bass Institute published the first large-scale empirical study (see Sharp & Sharp, 1997a) that analysed one of the world's largest (in terms of retail coverage) loyalty programs in Australia. The study observed a weak effect – an effect that may have been as much to do with the program launch, and its marketing spend, than the program itself.
The technical difficulty in evaluating the effect of loyalty programs on loyalty is to have a baseline to make comparisons from. Simply comparing the behaviour of loyalty program members to non-members is inappropriate because a brand's more loyal customers will join the program (i.e. a selection effect that occurs because these shoppers have more to gain). Looking for changes in consumers' behaviour from when the loyalty program was implemented is difficult, because this would require having data on long periods of continuous purchasing before and after the implementation of the program. A long period of data is required to correctly classify the loyalty level of individual consumers because their buying fluctuates. The law of buyer moderation (see Chapter 4) means that many members of loyalty programs will be misclassified as less loyal than they really are. Over time it will look as if these buyers are becoming more loyal and this 'regression to the mean' effect will be mistakenly attributed to the loyalty program. One analysis of a very small convenience store loyalty program suffered from this (Liu 2007).
Fortunately, the scientific laws introduced in this book mean that benchmarks are available for loyalty metrics; these remove the issue of selection effects and are used to see if brands that ran loyalty programs have unusual loyalty for their market share.
Table 11.1 illustrates the results of the Ehrenberg-Bass Institute's study, which examined the cross-category loyalty program, FlyBuys (FlyBuys is a separate program in two countries: Australia and New Zealand). As the name suggests the loyalty program rewarded shoppers with points that could be redeemed for flights.
Loyalty programs should make brands look niche, that is, their market share should be made up of unusually high loyalty for their market share and penetration. This is because loyalty programs reach a restricted part of the customer base and encourage these consumers to buy more frequently.77 So a loyalty program might or might not grow a brand’s share much but it should make that share be an unusual combination of penetration and loyalty. Or put another way, it probably won’t drive penetration but it may drive loyalty (which is not the usual way brands grow) Table 11.1 shows each brand in the FlyBuys loyalty program and the penetration and loyalty metrics necessary to make up its market share. By comparing the predicted and observed market share composition, we can see there does appear to be a loyalty effect. However, the effect is so weak that it is inconsistent; some brands don't show the pattern presumably because other marketing mix factors swamp the weak effect.
Table 11.1: The FlyBuys loyalty program
Size of customer base Loyalty related metrics
Penetration (%) Average purchase frequency Share of requirements (%) Sole buyers (%)
Observed Predicted Observed Predicted Observed Predicted Observed Predicted
Department stores (Australia)
Kmart 48 52 3.7 3.4 34 31 10 7
Target 43 42 2.9 3 27 27 5 6
Myer 35 34 2.8 2.8 23 25 5 5
Supermarkets (Australia)
Coles 61 64 9.8 9.4 31 29 5 3
Bilo 58 60 9.1 8.9 29 27 3 3
Retail petrol (Australia)
Shell 46 51 6.3 5.8 42 35 11 10
Bank credit cards (NZ)
NAB 20 25 9.6 7.2 87 70 79 63
BNZ 27 28 8.1 7.9 88 84 79 80
Phone calls (NZ)
Telecom 86 85 24.8 24.7 94 93 88 87
Retail petrol (NZ)
Shell 54 57 6.9 6.5 52 47 22 16
Average 48 50 8.4 8 51 47 31 28
Source: Ehrenberg-Bass Institute (see Sharp & Sharp 1997)
Professors Lars Meyer-Waarden and Christophe Benavent (2006) replicated the Ehrenberg-Bass research using the AGB Behavior Scan panel in France. They examined the effect of four supermarket loyalty programs and they too observed weak and inconsistent effects. Leenheer and colleagues (2007) used a different approach (a one-off statistical model) to control selection effects (i.e. that the most loyal consumers join a brand’s loyalty program). Their study of Dutch supermarket loyalty programs also found very small loyalty effects. Another Dutch researcher (Verhoef 2003) found small, positive effects on retention and share in the financial services market; however, Verhoef acknowledged that the longitudinal design of his study didn't fully account for the selection effects. It is a consistent story. Loyalty programs produce very slight loyalty effects, and do practically nothing to drive growth. The consequent effect on profits is presumably negative.
Why don't loyalty programs work better?
The answer lies in the fact that from a marketing strategy perspective there is something unusual about loyalty programs: they skew, more than other marketing interventions, towards heavier, more loyal buyers of the brand. This selection effect occurs for two reasons. The first is physical and mental availability: it is easier for more loyal buyers to notice the loyalty program and to join. This is particularly true of retail loyalty programs where usually the only way of joining is to sign up in-store; more regular shoppers have a statistically (mu
ch) higher chance of joining due to chance exposure. The second reason is the economic incentive to join, which is much stronger for already loyal buyers who can see that they will be rewarded for doing what they already do – getting 'something for nothing' is an attractive deal.
Essentially, consumers who rarely buy a brand (or shop in a particular store) don't see the loyalty program – and if they do they can't see the point in joining. This substantially restricts the reach of loyalty programs, and hence limits their capacity to drive substantial growth. There is also the question of what effect the loyalty program has on those consumers who it does reach (i.e. consumers who join the program and remember to participate). Logically, this effect will not only depend on the strength of the loyalty program, but also on which particular consumers the loyalty program recruits. Table 11.2 shows that there are four types of consumers a loyalty program can recruit and potentially influence (i.e. influence their loyalty towards the brand with the loyalty program).
Table 11.2: Types of customers who could join the loyalty program
Level of loyalty
Low High
Category purchase level Light Unlikely to join the program (but this is the largest group of consumers, so a number still do join); not particularly desirable Likely to join the program, but undesirable
Heavy Very unlikely to join the program, but very desirable Likely to join the program, but undesirable
Most consumers fall into the top left-hand quadrant because most buyers are lighter-than-average consumers (of products in a particular category) and most of a brand's buyers also buy other brands.78 A loyalty program will inevitably recruit some of these light consumers because there are so many of them. Most will become lapsed users of the program as they generally fail to accumulate sufficient points to qualify for rewards and so gradually forget about, or reject, the program (e.g. they lose the loyalty card). Hopefully a few of these consumers will become heavier buyers. 79