Professional Documents
Culture Documents
ABSTRACT
This article represents the first of several editorials to appear in the Journal of Retailing designed
to examine the nexus between retail practice and research, with the goal of stimulating
further research. This essay on emerging trends in pricing discusses recent advances in
retail pricing optimization. We begin with a review of how retailers typically make pricing
decisions using time-honored heuristics and attempt to infer the optimal decisions. However,
current methods are sub-optimal because they do not consider the effects of advertising,
competition, substitute products, or complementary products on sales. Most fail to take
into account how price elasticity changes over time, particularly for fashion merchandise, or
how market segments react differentially to price changes. In addition, many retailers find
it difficult to know how to price merchandise when their suppliers offer temporary ‘deals’.
They are also generally unaware of how their pricing strategy influences their overall image.
As these issues demonstrate, optimal pricing is not a static problem. Retailers must be able
to react quickly to changes in the environment or sales patterns. This paper also provides
examples of the more sophisticated pricing techniques that are currently being tested in
practice. Finally, we conclude with a discussion of the critical components that must be
incorporated into retail pricing.
1 This research was based on analysis of retail and economic trends reported by the U.S. Census Bureau
and the National Retail Federation.
2 Ibid.
Although some retailers can experiment the life-cycle of a typical fashion item
with new items to determine profit- is much shorter than that of a staple, its
maximizing prices, this approach is lifespan depends on the category and the
impractical for most retailers, because target market. For example, double-breasted
they have too many items to consider. In suits for men or certain popular interior
addition, experiments simply are not timely. decorating colors represent fashions whose
By the time the results come in, the item life-cycle may last several years. In contrast,
could be in another stage of its life-cycle. trendy fashions, such as see-through track
Experimentation could, however, work for shoes, may last for only a short season.3
a chain, such as the Cheesecake Factory, Items in the staple merchandise category
when it wants to try a new menu item. The (also called ‘basic merchandise’ or ‘fast-
restaurant could offer the item at different moving consumer goods’) are in continuous
prices in different markets to determine demand during an extended period of time.
which price is the most profitable. Most merchandise in grocery and drug
stores, as well as housewares, hosiery, basic
TWO DISPARATE PRICING PROBLEMS: blue jeans and women’s intimate apparel
FASHION AND STAPLE MERCHANDISE in specialty and department stores, are
Although there are many types of considered staple merchandise.4
merchandise, from an inventory- We depict a representation of the life-cycle
management perspective, most SKUs fit into for fashion and staple goods merchandise
one of two categories: fashion or staples. in Figure 1. There are several similarities
Fashion is a category of merchandise that between the two graphs. Both lines have
typically lasts 8-12 weeks and sales vary ‘spikes’, which represent increases in sales
dramatically from one season to the next. – usually caused by promotions.
Within the fashion category, a specific style
or SKU sells for one season or less. Although
3 One firm that specializes in pricing optimization for fashion merchandise is ProfitLogic (profitlogic.
com).
4 Firms that specialize in pricing optimization for staple merchandise include KhiMetrics (khimetrics.com)
and DemandTec (demandtec.com).
After the promotion, the level of sales techniques, the underlying principle of
settles down to somewhere near its previous maximizing profits by analyzing price
level. However, there are also important elasticities remains the same.
differences in the life-cycles of fashions and
staples in Figure 1. Firstly, the fashion curve CRITICAL COMPONENTS TO BE
is similar in form to a ‘normal’ or bell-shaped INCORPORATED INTO RETAIL PRICING
curve, in which sales start at zero, increase Retailers are interested in maximizing their
over a particular season and end at zero. The profits. To do so, they need to understand
staple line, by contrast, remains relatively flat, how to price their merchandise optimally.
though it may veer up or down, depending What does optimal price really mean? It is
on the general trend for the SKU and the the price at which profits are maximized
season. Secondly, the demand line for staple by item or group of items. Of course, this
merchandise never approaches zero. Unlike process is not as easy as it sounds. For
fashions, staples continue selling, at least example, by maximizing the profits of
over a reasonable planning horizon. one item by setting its ‘optimal’ price, the
The systems designed to optimize price retailer may sacrifice profits on another
for both fashion and staple goods are item whose demand correlates with the first
complicated, but for different reasons. For item. To illustrate, if a retailer reduces the
fashion goods, the objective is to maximize price of an 8-ounce can of Hunt’s tomato
the profits for the item or category and, at sauce, the demand for the 12-ounce can
the same time, price the merchandise so might decrease.
that inventory approaches zero at the end Next, we discuss this and six other factors
of the fashion cycle, because at that time, that must be taken into consideration to
it is out of style and has little or no value determine optimal prices. To summarize,
in the marketplace. That is, if a store has a these factors are as follows:
lot of sweaters left over in January, it must 1. Price sensitivity, or how demand for an
aggressively mark them down to ensure SKU changes with its price; for fashion
that they are gone in time for the arrival of merchandise, how does price sensitivity
spring merchandise. change over time? (e.g. a markdown for
Because staple merchandise continues a sweater in May may not create the
to sell throughout the year, staple retailers same sales lift as the same markdown
do not need to consider the complication would in January.)
of pricing to be out of stock on a certain 2. Substitution effects, namely, how
date. However, the staple merchandise demand for an SKU changes with the
optimization problem is complicated due to price of a competing SKU.
the number of potential decisions that must 3. Dynamic effect of price promotions
be made. Pricing decisions can be made at over time, or how changing prices
the SKU level for staple merchandise and today affects tomorrow’s demand.
retailers must take into consideration the 4. Segment-based pricing, which
effect that the price of one SKU has on investigates how prices vary across
another or the effect one category of SKUs different markets/customer segments.
has on another category. Therefore, the 5. Cross-category effects, or accounting
sheer size of the optimization problem for for demand complementarities across
staple merchandise can be daunting. categories.
Although the differences between pricing 6. Retailer costs (wholesale prices and
optimization systems for staple and fashion trade deals) and discounts.
merchandise appear to be substantial 7. To what extent competition at the retail
and, though, software vendors approach level influences retail prices.
the problems with a variety of analytical
Price sensitivity effects SKU because of its price, the retailer should
At the most basic level, to determine an evaluate the relative margins of the two
optimal initial or markdown price, the SKUs before lowering the price of the target
retailer must assess its own-price elasticities SKU. Implicit, therefore, in this effect is the
(derived from demand curves, which interesting observation that ‘cross-pass-
are usually non-linear) to measure how through’ effects exist; that is, changes in
sensitive demand is to price for a given the wholesale cost of one SKU can drive
item over a period of time. Although price changes in the prices of other SKUs. For
elasticities generally have a negative sign, example, if the wholesale price for SKU ‘A’
to suggest that an increase in price usually is temporarily lowered, it may be optimal
results in a decrease in demand, in some for a retailer to ‘convert’ those customers
situations, however, a decrease in price who normally purchase SKU ‘B’ to buy SKU
can lead to a perception of lower quality, ‘A’ because of its now higher margin. Such
which thus decreases demand. Such price– a conversion is determined, in part, by the
quality inferences are well documented in cross-price effect of SKU ‘A’ on SKU ‘B’. In
behavioral pricing research (e.g. Dodds, other words, if SKU ‘A’ cannot steal sales
Monroe, and Grewal, 1991). In addition, from SKU ‘B’, it may not be worthwhile
the role played by quality signals (e.g. to attempt to convert buyers to ‘A’. This
price marching guarantees, warranties, complicated effect suggests that retailers
store reputation and brand image) must should adopt a category management
be incorporated (Estelami, Grewal, and approach to develop a pricing strategy
Roggeveen, 2004; Kukar-Kinney and (Basuroy, Mantrala and Walters, 2001;
Walters, 2003; Miyazaki, Grewal and Chen et al, 1999; Chintagunta, 2002; Hall
Goodstein, 2004; Srivastava and Lurie, et al, 2004; Zenor, 1994).
2004).
Estimating price elasticities for fashion Dynamic effects of price promotions
merchandise is more complex, because Retailers often assume that sales (both
fashions are not stable over the course when there is no promotion [baseline] and
of the season. A price reduction prior to when there is a promotion offered) for a
Christmas, for example, will cause a higher given SKU are independent of past pricing
sales spike than if the same reduction were activity. Yet recent evidence suggests that
introduced in September. Furthermore, sales may be affected by prior discounting
the joint effects of advertising and price activity.
promotions on price sensitivity and Research in consumer behavior has
demand must be incorporated explicitly demonstrated that consumers evaluate
(Kalra and Goodstein, 1998; Sethuraman retail prices for items relative to certain
and Tellis, 2002). internal benchmarks or reference
prices (Winer, 1986). Some retailer- or
Substitution effects manufacturer-supplied information that
At the general level, the substitution effects plays a prominent role in affecting these
– cross-price elasticities or cross-price internal reference prices includes MSRP
effects – of a brand refer to the effect of and retailer-supplied reference prices (e.g.
the change in the price of an SKU on the regular price, original price, compare at
demand for a competing SKU (Besanko, price). (Lichtenstein and Bearden, 1989;
Dubé and Gupta, 2004). Bell, Chiang, and see also reviews by Compeau and Grewal,
Padmanabhan (1999) note that almost 1998; Krishna et al, 2002; Urbany, Bearden,
75% of consumer response to promotions and Weilbaker, 1988). Consumers’ internal
is due to brand switching. Hence, if an SKU reference prices also can be influenced by
can steal market share from a competing past prices, brand promotion frequency
and the type of store (Kalwani et al, 1990). Furthermore, excessive price promotions
Therefore, price promotions are likely to over time may result in increased customer
affect consumer reference prices or price price elasticity. If these long-term negative
expectations. effects of promotions on baseline sales and
In addition, it is difficult for retailers price response are high, retailers should
to understand the differences in the sales decrease their use of price promotions.
lift generated from a promotional vehicle In a large-scale field experiment involving
(an advertisement) compared with that durable goods sold through a direct mail
attributable to the offer itself (sales price catalog, Anderson and Simester (2004, p.4)
or discount). Ignoring this dynamic can find that ‘[d] eeper price discounts in the
substantially affect the optimal price of current period increased future purchases
an SKU. Furthermore, it is important to by first-time customers (a positive long-
understand that different promotions (e.g. run effect), but reduced future purchases
discounts, coupons, rebates or bundles) do by established customers (a negative long-
not only have differential effects (Compeau run effect)’. Most theories of the effect
and Grewal, 1998; Hardesty and Bearden, of price promotions, such as purchase
2003; Raghubir, 2004), but also result in acceleration, selection, customer learning
consumer dynamics such as stockpiling and increased deal sensitivity, predict lower
and purchase deceleration (Neslin, 2002). future purchases, so their finding regarding
Consider the following example (Kopalle, first-time buyers is puzzling and should be
Mela and Marsh, 1999): a few years ago, investigated further.
a major discount store chain used sales
promotions relatively infrequently. Its off- Segment-based pricing effects
discount (baseline) sales were moderate Consumers in different markets behave
and consumer response to sales promotions differently with regard to their own and
was good. Observing the promotional cross-price elasticities, as well as how
response, the retailer decided to increase they react to price changes. For example,
sales promotions, which led to a decrease in customers in an upper-income area may be
baseline sales. Believing that the additional less sensitive to price and the relationship
sales promotions were successful, the among the prices of various products than
retailer added even more. Eventually, the those in a less affluent region. By taking
retailer started offering special promotions these differential factors across markets
almost every week, and its management into consideration, retailers can implement
wondered why profitability was so low, different price and promotion plans across
since the large incremental demand over various markets.
the baseline indicated that the promotions A retailer’s ability to segment and charge
were working so well. differential pricing may also hinge on the
Why didn’t the retailer’s management price awareness levels of the consumer
recognize that the increase in sales segments. For example, prior research has
promotions led to a decrease in baseline demonstrated that consumers have low
sales and that its pricing decisions were levels of price recall and awareness for many
sub-optimal? Many retailers have little products (Binkley and Bejnarowicz, 2003;
understanding of how such estimates Dickson and Sawyer, 1990; Mazumdar and
arise. Clearly, retailers should consider the Monroe, 1990). Thus, in some categories,
possibility that increases in the use of price retailers may be wasting profits by over-
promotions can have long-term negative discounting their merchandise in an effort
effects on their baseline sales; in other to appeal to a deal-prone segment, for
words, baseline sales could decrease with which a small discount might be sufficient
frequent promotions (Kopalle et al, 1999). (Inman, McAlister and Hoyer, 1990).
researchers can assess the elasticity of of an item can range from $0.25 to $0.50
an SKU for which the price has never per item (Levy et al, 1997). A price change
changed. cost, therefore, should be built into any
optimization model. If the cost of changing
Psychological price thresholds and reference the price is greater than the additional
effects revenue projected from the price change, it
Assume that a pricing optimization program makes more sense to leave the price alone.
recommended a price of $2.90 for a staple
product, but the retailer, consistent with Good data in, good data out
Anderson and Simester (2003), believes that A multi-billion-dollar retailer can be run
shoppers in a grocery store do not notice on bad, or unrefined, data no more than
the last digit of a price, so the retailer is free a rocket-ship can be run on crude oil. The
to round the price up to the nearest nine, typical problem with retailers, however, is
or $2.99. This tactic would increase dollar not that they lack sufficient data, rather,
sales by approximately 3%, with almost they have too much, but not in a useable
no increase in costs. Pricing optimization format!
programs systematically examine the When it comes to data requirements,
recommended price – $2.90 in this case retailers are unlike other businesses that
– and exercise analytical rules to round it make similar decisions. Airlines, for
up to a higher price that yields more profit, example, have used yield management
because customers are insensitive to the techniques to make pricing decisions
difference. Consistent with this example, for years. Until recently, however, these
in an analysis of scanner data in 29 sophisticated statistical techniques were
categories over an eight-year period, Levy unavailable to retailers, which must
and colleagues (2004) find that small price manage a great deal of data. For example, a
increases occur more frequently than small typical retail chain might have data about
price decreases. 20,000 SKUs for each of 2,000 stores for
A related issue is to incorporate the effects 104 weeks, during which time it offers
of reference prices – the anchoring levels, or three promotions. At 100 bytes/record, this
standards that consumers use to compare data set would require approximately 1.2
observed purchase prices of a product – on terabytes of storage capacity.
consumers’ internal reference prices and Not even a great buying team could make
demand (e.g. Blair, Harris and Monroe, sense out of such a mountain of data. To
2002; Chandrashekeran and Grewal, 2003; prepare the data for analysis, better systems
Kalyanaram and Winer, 1995; Kopalle have algorithms that combine individual
and Lindsey-Mullikin, 2003; Krishna et SKUs into affinity groups that behave
al, 2002). If the observed price is greater similarly in the marketplace. For example,
than the reference price, it is perceived as all six packs of Pepsi would be priced the
a loss. In contrast, if the observed price is same, because customers expect it and the
less than the reference price, it is perceived manufacturer requires it.
as a gain. Kopalle, Rao, and Assunção’s Pricing optimization software also will
(1996) results suggest that dynamic (or hi- identify outlier sales points, fill in missing
low) pricing is optimal when the positive data or smooth out a demand spike that
impact of a gain on sales outweighs the was caused by an aberrant factor, such as
negative impact of a corresponding loss. weather, a non-recurring promotion or
non-recurring competitive action. Most
Price change costs important, it is difficult for retailers to
It is expensive for retailers to change prices. capture lost sales that result from a product
The price of finding and changing the cost being out of stock. If a customer goes to buy
wish to peg their prices to those of their Shopping: The Case of Quantity
competition or set prices to maintain a Surcharges’. Journal of Retailing, 79 (1):
certain image. How do these conflicting 27-35
goals affect their customers and their Blair, E., Harris, J. & Monroe, K. B. (2002).
profits? ‘Effects of Shopping Information on
Finally, an emerging trend in retailer Consumers’ Responses to Comparative
strategies is that of frequent shopper Price Claims’. Journal of Retailing, 78 (3):
programs. Kopalle and Neslin’s (2003) 175-181.
analysis suggests that such strategies Bolton, R. N. & Shankar, V. (2003). ‘An
have the potential to be effective Empirically Derived Taxonomy of Retailer
multi-period sales promotion tools Pricing and Promotion Strategies’. Journal
when primary demand can expand. of Retailing, 79 (4): 213-224.
It will be interesting to examine how Chandrashekaran, R. & Grewal, D.
such programs impact retail sales and (2003). ‘Assimilation of Advertised
consumer behavior over time. Reference Prices: The Moderating Role of
Involvement’. Journal of Retailing, 79 (1):
These are but a few of the many research 53-62.
initiatives that we hope this article will Chen, Y., Hess, J. D., Wilcox, R. T. & Zhang,
stimulate. Z. J. (1999). ‘Accounting Profits Versus
Marketing Profits: A Relevant Metric for
REFERENCES Category Management’. Marketing Science,
Anderson, E. T. & Simester, D. I. (2003). 18 (3): 208-229.
‘Mind Your Pricing Cues’. Harvard Business Chen, Y. & Iyer, G. (2002). ‘Consumer
Review (September): 97-103. Addressability and Customized Pricing’.
Anderson, E. T. & Simester, D. I. (2004). Marketing Science, 21 (2): 197-208.
‘Long-Run Effects of Promotion Depth Chintagunta, P. K. (2002). ‘Investigating
on New Versus Established Customers: Category Pricing Behavior at a Retail
Three Field Studies’. Marketing Science, 23 Chain’. Journal of Marketing Research, 39
(1): 4-20. (May): 141-154.
Basuroy, S., Murali, K. M. & Walters, R. Compeau, L. D. & Grewal, D. (1998).
G. (2001). ‘The Impact of Category ‘Comparative Price Advertising: An
Management on Retailer Prices and Integrative Review’. Journal of Public Policy
Performance: Theory and Evidence’. & Marketing, 17 (Fall): 257-274.
Journal of Marketing, 65 (October): 16-32. Dhar, S. K., Hoch, S. J. & Nanda Kumar, N.
Bell, D. R., Chiang, J. & Padmanabhan, (2001). ‘Effective Category Management
V. (1999). ‘The decomposition of Depends on the Role of the Category’.
Promotional Response: An Empirical Journal of Retailing, 77 (2): 165-184.
Generalization’. Marketing Science, 18 (4): Dickson, P. & Sawyer, A. (1990). ‘The Price
504-526. Knowledge and Search of Supermarket
Bell, D. R., Teck-Hua Ho & Tang, C. S. Shoppers’. Journal of Marketing, 54 (July):
(1998).‘Determining Where to Shop: 42-53.
Fixed and Variable Costs of Shopping’. Dodds, W. B., Monroe, K.B. & Grewal, D.
Journal of Marketing Research, 35 (August): (1991). ‘The Effects of Price, Brand, and
352-369. Store Information on Buyers’ Product
Besanko, D., Dubé J-P. & Gupta, S. (2004). Evaluations’. Journal of Marketing Research,
‘Retail Pass-Through on Competing 28 (August): 307-19.
Brands’. Marketing Science, forthcoming. Drèze, X. & Hoch, S. J. (1998). ‘Exploiting
Binkley, J. & Bejnarowicz, J. (2003). the Installed Base Using Cross-
‘Consumer Price Awareness in Food Merchandizing and Category Destination