Content uploaded by Ann Melissa Campbell
Author content
All content in this area was uploaded by Ann Melissa Campbell on Oct 14, 2015
Content may be subject to copyright.
Practice Article
Revenue management opportunities for
Internet retailers
Received (in revised form): 6th November 2012
Niels Agatz
a
, Ann M. Campbell
b
, Moritz Fleischmann
c
, Jo van Nunen
a
and
Martin Savelsbergh
d
a
Erasmus University, Rotterdam, The Netherlands;
b
University of Iowa, Iowa City, Iowa, USA;
c
University of
Mannheim, Mannheim, Germany; and
d
University of Newcastle, Newcastle, New South Wales, Australia
Niels Agatz is an assistant professor of Logistics and Operations Management at the Rotterdam School of
Management, Erasmus University, The Netherlands. His research interests are in transportation optimization,
demand management and pricing and the interface between marketing and operations, specifically in B2C
Internet retailing.
Ann M. Campbell is an Associate Professor at the University of Iowa in the Department of Management
Sciences in the Tippie College of Business. Her research is in the field of transportation and focuses primarily
on vehicle routing, delivery network design, and disaster logistics. She has written over twenty journal articles,
many of which are published in leading transportation journals.
Moritz Fleischmann is Professor and Chair of Logistics and Supply Chain Management at the University of
Mannheim, Germany. His research interests are in the field of supply chain management, with focal points on
the integration of revenue management and supply chain management, coordination of the marketing-
operations interface, e-fulfillment, closed-loop supply chains, and inventory management.
Jo van Nunen was professor and chair of decision and information sciences at the Rotterdam School of
Management, Erasmus University. His research on e-commerce, closed-loop supply chain management and
multi-modal transportation is published in 6 books and more than 150 journal articles. He passed away
unexpectedly in May 2010.
Martin Savelsbergh is a professor in the Faculty of Mathematical and Physical Sciences at the University of
Newcastle, New South Wales, Australia. He is an optimization and logistics specialist with over 20 years of
experience in mathematical modeling, operations research and logistics. He has published over 100 research
papers in many of the top optimization and logistics journals.
Correspondence: Niels Agatz, Rotterdam School of Management, Erasmus University, Burgemeester
Oudlaan 50, 3062 PA, Rotterdam, The Netherlands.
E-mail: nagatz@rsm.nl
ABSTRACT In this article, we explain how Internet retailers can learn from proven revenue management
concepts and use them to reduce costs and enhance service. We focus on attended deliveries as these
provide the greatest opportunities and challenges. The key driver is service differentiation. Internet retailers
have strong levers at their disposal for actively steering demand, notably the offered delivery time windows
and their associated prices. Unlike traditional revenue management, these demand management decisions
affect both revenues and costs. This calls for a closer coordination of marketing and operations than current
common practice.
Journal of Revenue and Pricing Management (2013) 12, 128–138. doi:10.1057/rpm.2012.51; published online 4 January 2013
Keywords: E-fulfilment; demand management; revenue management; home delivery
& 2013 Macmillan Publishers Ltd. 1476-6930 Journal of Revenue and Pricing Management Vol. 12, 2, 128–138
www.palgrave-journals.com/rpm/
INTRODUCTION
Since the burst of the Internet bubble, e-tailers
have taken major strides towards improving the
profitability of online distribution channels.
Operations managers have increased the effi-
ciency of fulfilment processes. Marketing man-
agers have unveiled the purchase behavior
of different customer segments. However, the
coordination between operations and market-
ing remains weak – whereas they are closely
interdependent. At the same time, information
technology provides the means for an ever
richer interaction with the customer. Exploit-
ing these means to enhance the marketing-
operations coordination represents a vast yet
untapped potential for sustaining the profit-
ability of online distribution. In this article, we
discuss how revenue management concepts can
help achieve this goal.
Internet retail sales have been growing at a
staggering pace over the last decade, and noth-
ing indicates that this trend will change any
time soon. This expansion often comes at the
expense of the physical stores. From 2002 to
2010, online retail sales in the United States
increased at an average annual growth rate of
17.9 per cent, compared with 2.6 per cent for
total retail sales (US Census Bureau, 2012).
Similar growth is seen in Europe, where the
Centre for Retail Research predicts online
retail sales to grow by 16.1 per cent to h232.8
billion in 2012.
Internet retail channels have a great potential
for increasing customer value by providing
access 24/7 to a huge assortment right from
the customer’s home. However, Internet retail-
ing also requires a physical distribution struc-
ture, either in-house or outsourced to a third
party, that fulfills these promises by promptly
delivering the product to the customer. Inabil-
ity to design cost-efficient fulfilment processes is
widely recognized as a root cause of many of the
failures of early online retail initiatives.
Cost-efficient fulfilment is particularly chal-
lenging in the case of attended home delivery,
which is common for many types of products
that cannot easily be delivered in the customer’s
mailbox, such as grocery (due to perishability),
electronic equipment (value), or white goods
and furniture (size) (Boyer et al, 2004). In
attended home delivery, the retailer and the
customer agree on a delivery time window. The
window choice gives rise to a cost-service
trade-off. A narrow delivery window provides
certainty to the customer but in general implies
higher costs for the retailer by limiting the
retailer’s flexibility (Punakivi et al, 2001). Inter-
net retailers have learned from past f ailures to
make this trade-off carefully. For example, most
of today’s Internet grocers use one to four hour
delivery windows (see Table 1) rather than the
aggressive 30-min window offered by Webvan
in the late 90s. At the same time, some Internet
retailers will find out that their current delivery
offering is too conservative. Many customers
no longer accept half-day or even full-day
delivery windows, for example, for the delivery
of home appliances.
Internet retailers have been improving their
supply chain efficiency also in other ways.
These include the ‘bricks-and-clicks’ bundling
of online and offline channels to leverage buy-
ing power, the incremental expansion of deliv-
ery networks instead of a rush to giant
automated fulfilment centers, and the smart
deployment of inventories across the delivery
network to offer a large assortment while limit-
ing the risk of overstocking (Agatz et al, 2008).
However, these supply-side decisions cover
only half of the supply-demand equation. De-
mand management has an equally strong impact
on profitability, affecting both costs and reven-
ues, and may hold a much greater potential for
many of today’s Internet retailers than further
supply-side optimization.
Effective demand management requires a good
understanding of customer preferences and the
flexibility to tailor the product proposition to
them. Internet retailers are in a unique position
in both the respects. First, online transaction and
click-stream data pr ovides a w ealth of information
on customer behavior . Second, online commu-
nication allows close interaction with the custo-
mer , including rich deliv ery negotiations.
Revenue management opportunities for Internet retailers
129& 2013 Macmillan Publishers Ltd. 1476-6930 Journal of Revenue and Pricing Management Vol. 12, 2, 128–138
In this article, we explain how Internet
retailers can learn from proven revenue man-
agement concepts and use them to reduce costs
and enhance service. We focus on environments
with attended deliveries as these provide the
g reatest opportunities and challenges. We pro-
ceed as follows. In Section 2, we review the
e-fulfilment process of a large Internet grocer in
The Netherlands. The example serves to illus-
trate the main supply chain processes of an
online retail channel. In Section 3, we compare
these processes to those in airline revenue
management. We show that both setti ngs share
a number of characteristics, but argue that
there are also import ant differences. There-
fore, tradi tional revenue management concepts
have to be adj usted to be e ffective in an
Internet retailing context. We classify the
resulting approaches on the basis of two criter-
ia: the demand management lever and the
degree of time flexibility. In Sections 4–7, we
discuss the four resulting solutions i n detail.
We highlight the specific benefits of each
solution and explain how to realize them.
Section 8 summarizes our findings and dis-
cusses implications f or industries beyond In-
ter net retailing.
ILLUSTRATIVE CASE: DEMAND
MANAGEMENT AT Albert.nl
To help make things concrete, we briefly sketch
the home-delivery process of a specific Internet
retailer , namely Holland-based e-grocer Albert.nl.
The grocery sector is commonly recognized as
one of the most challenging env ir onments for
successful e-fulfilment because of fierce com-
petition, low pr ofit margins and logistically
demanding products, such as fresh food. Not
surprisingly, the sector has seen some of the
most spectacular e-business failures. On the
other hand, practices that work in e-gr ocery hav e
a high chance of success with other Internet
retailers.
Table 1: Delivery policies of E-grocers in Europe and US
Delivery area Delivery time
slot length
Timing Delivery fee
a
Dynamic
incentives
Albert.nl B65% of Dutch
households
2-hour 8 am–2 pm/
4 pm–9 pm
h4.95–h8.95 (t)—
Sainsburys.co.uk 83% of UK postcodes 1-hour 10 am–10 pm £5 —
Ocado.com B80% of UK
households
1-hour 6 am–11 pm £3-£6 (t, s) |
Tesco.com 96% of UK households 2-hour 9 am–11 pm £3.99-£5.99 (t)—
Peapod.com Chicago, Washington,
DC, Boston,
Baltimore, e.a.
2 hour/
3.5 hour
6 am–1 pm/
4 pm–9.30 pm
$6.95–$9.95 (t, s) |
Albertsons.com Seattle, Portland, San
Fransisco,
Los Angeles, San
Diego,
Las Vegas, e.a.
1.5 hour 10 am–2.30 pm/
3.30 pm–9.30 pm
$9.95 —
Safeway.com Seattle, Portland,
San Fransisco,
Los Angeles,
San Diego, e.a.
2 hour/
4 hour
10 am–3 pm/
4 pm–9 pm
$9.95 (s)—
a
t=time-dependent; s=size-dependent.
Agatz et al
130 & 2013 Macmillan Publishers Ltd. 1476-6930 Journal of Revenue and Pricing Management Vol. 12, 2, 128–138
Albert.nl is the Internet channel of Albert
Heijn, the Netherlands’ largest supermarket
chain and a subsidiary of Royal Ahold, the retail
multinational, who also owns Peapod, one
of the major US e-grocers. Albert.nl offers
approximately 10 000 SKUs, including fresh
g roceries such as meat, milk, and fruit, thereby
corresponding with a mid-sized Dutch super-
market. The product prices are identical to
those in the conventional Albert Heijn
stores, plus a time-dependent delivery fee.
Albert.nl has organized its warehouse and
delivery operations in two shifts per day. Cur-
rently, the company offers its attended home
delivery service in all major urban areas in the
Netherlands.
Upon login, the customer reserves a 2-hour
delivery window. The cut-off point for order
placement is about 16 hours before the actual
delivery. After that cut-off point, when all the
orders for the given shift are known, the
company plans the corresponding delivery
routes. Subsequently, the orders are picked in
the warehouse and grouped by vehicle before
actually being delivered. A typical delivery
route visits between 10 and 20 customers.
Zooming in on demand management reveals
two main levers, which the e-tailer uses to steer
demand, namely, the set of delivery windows
offered to the customer and the corresponding
delivery fees. To improve the capacity utiliza-
tion of the delivery service, the company uses
a differentiated delivery fee to balance the
demand over the week as well as over the day.
The delivery fee ranges from h4.95 to h8.95,
based on the popularity of the time slot. As
expected, this price differentiation smoothes
the demand, reducing the ratio between the
busiest to the least busy window in terms of
number of customers visited from 3:1 to 1.5:1.
The second demand management lever con-
cerns the offered delivery windows. Albert.nl
offers different sets of windows, dependent on
the zip-code of the delivery location. This lever
mainly serves to balance regional differences in
demand volumes. Low demand areas receive
fewer windows than areas with more customers
in order to concentrate deliveries and achieve
economies of scale. Moreover, Albert.nl takes
into account the proximity of different zip-
codes in choosing the delivery windows in
order to support efficient delivery routes with-
out harming customer service.
Some of the above details are, of course,
specific to Albert.nl. For example, Tesco.com,
the world’s largest e-grocer, uses store-based
rather than warehouse-based order picking.
Others, like UK-based retailer John Lewis,
outsource their entire fulfilment operation.
However, the main steps of the fulfilment
process – order in-take, routing, picking, ex-
ecution – and the corresponding planning tasks
are generic and apply to many other Internet
retailers with attended home delivery and even
to other delivery services. This includes parcel
carriers such as FedEx and Velocity Express
who are targeting the B2C delivery market
with evening deliveries and more narrow time
windows than in the traditional B2B market
(Demery, 2007).
The role of pricing and delivery windows as
the main demand management levers is generic,
as is their impact on both customer ser vice and
operational efficiency. In this article, we will
explain how to exploit these levers in a sys-
tematic way.
LEARNING FROM REVENUE
MANAGEMENT
Inspiration for effective demand management
for e-fulfilment comes from the example of
revenue management. In fact, revenue manage-
ment is demand management. In order to
maximize the revenues generated with a given
amount of capacity, revenue management aims
to exploit market heterogeneities. Therefore,
the market is partitioned into different market
segments with different price sensitivities and
different preferences. In the airline case, the
prototypical application of revenue manage-
ment, the classical market segmentation is be-
tween business and leisure travelers. Business
travelers tend to have a higher willingness to pay
Revenue management opportunities for Internet retailers
131& 2013 Macmillan Publishers Ltd. 1476-6930 Journal of Revenue and Pricing Management Vol. 12, 2, 128–138
and value flexibility regarding late booking and
cancellations. Leisure travelers have a lower
willingness to pay but are more flexible in
general. Each group also prefers different travel
times, such as weekday versus weekend stays.
The key insight is that capitalizing upon these
heterogeneities allows companies to do better
than simply selling the capacity first-come-
first-served at a constant price. Segmentation as
such is a classical marketing instrument. What is
new about revenue management is the in-
creased flexibility, which allows companies to
adjust prices and volumes offered to different
segments in real-time.
In Table 2, we compare airline revenue
management with Internet retailing. For the
sake of simplicity, we restrict ourselves to the
simplest case of a single-leg passenger flight,
recognizing that today’s airline revenue man-
agement systems address many additional
complexities, notably the optimization of
flight networks rather than single flights (Talluri
and Van Ryzin, 2004). The table highlights key
ingredients of revenue management structured
around supply-side and demand-side elements.
In the Table 2, we can see that the main
conditions for revenue management also apply
to Internet retailing. The Internet retailer serves
a heterogeneous market with a delivery capacity
that is relatively inflexible in the short run, and
he can change prices and customer access
relatively easily. This implies that also Internet
retailers can do significantly better than offering
delivery on a first-come-first-served basis for a
constant price.
However, we also see significant differences
between Internet retailing and traditional rev-
enue management environments. First, Internet
retailing concerns the combination of physical
products plus a delivery service. Effective
demand management needs to take the product
dimension into account, notably through its
effects on revenues and capacity. Second, de-
mand management has a significant cost impact
in Internet retailing, as the location of the
customers impacts the costs of delivery, unlike
an airline setting, which fixes operations prior
to order in-take. Consequently, demand man-
agement in Internet retailing translates to profit
management rather than revenue management.
Figure 1 highlights this important difference
between the airline and the e-fulfilment context
and links it to the differences in the underlying
planning processes.
Next we examine how to transfer revenue
management concepts to Internet retailing
Table 2: Key characteristics of airline and e-fulfillment
Airline e-Fulfilment
Supply
Product Travel service Physical product þ delivery service
Capacity Number of seats: fixed, perishable Product inventory: flexible
Picking þ delivery capacity: inflexible, perishable
Costs Sunk at order in-take Variable, interdependent transportation costs
Booking Up to months in advance, specific
depar ture time
Days in advance, delivery time window
Demand
Revenues Fare Product margin þ delivery fee
Transaction size Single seat Varying order size þ driving time
Customer
heterogeneity
Willingness to pay, flexibility,
travel time
Willingness to pay, flexibility, delivery time, order
size, delivery location
Response to
stockout
Lost, up-sell/down-sell,
alternative flight
Lost, alternative delivery time, off-line store
Agatz et al
132 & 2013 Macmillan Publishers Ltd. 1476-6930 Journal of Revenue and Pricing Management Vol. 12, 2, 128–138
while taking these distinctions into account. Akin
to traditional revenue management, we will dis-
tinguish betw een quantity-based and price-based
solutions. The first concerns decisions on which
delivery options, namely which time slots, to
makeavailabletowhichcustomers.Thesecond
focuses on the deliv ery fee as the main lev er to
manage customer demand. A retailer can apply
both of these options, slotting and pricing, at
different moments in the sales process, either off-
line prior to the actual order in-take or real-time
as demand unfolds. These distinctions leave us
with four different types of demand management
in e-fulfilment, as summarized in our framework
in Table 3. We next address each of these options
and explain its potential and its requirements.
DIFFERENTIATED SLOTTING
The first lever for managing demand concerns
the delivery time slots offered to the customers.
Many retailers express these in terms of a
weekly schedule. The time slot offering sets
the conditions before the actual order in-take.
As such, the collection of time slots forms the
base of the demand management system in
Internet retailing.
Obviously, the number, length and selection
of delivery slots impacts customer service. Pre-
ferences can vary for different customers and in
different situations. The more choices, both in
terms of time and length, the more attractive
the service offering is for the customer, which
translates into higher expected sales. At the
same time, however, the offered delivery slots
directly affect the efficiency of the delivery
operations through transportation costs. Limit-
ing the number of delivery options in a given
geographical area helps concentrate customer
orders, thereby reducing the distance traveled
per order. Furthermore, the timing of the
offered slots to nearby customer areas (see
Figure 2) impacts the routing efficiency.
An effective time-slot offering has to balance
these effects. To this end, customers should first
be grouped geographically (for example by zip-
code). Second, the number of visits per week to
each group needs to be determined. This
involves gauging the sensitivity of demand to
the number of delivery oppor tunities offered as
well as assessing delivery costs as a function of
the number of deliveries made per visit. Third,
the slots have to be allocated over the week.
The allocation should reflect the different pre-
ferences of the different customer g roups, such
as assuring availability of evening slots in areas
with many busy professionals. Given these
constraints, the windows have then to be
coordinated across customer areas so as to
facilitate efficient delivery routes. Note that
not all windows necessarily have to have the
same length. Offering longer windows, for
example, in a rural area, increases the routing
flexibility. Windows of different lengths asso-
ciated with different prices also provide a means
for segmentation (see Section 5).
In computer simulations, we found that
optimizing the number of time slots offered
per week per zip-code reduced delivery costs by
about 10 per cent compared with unrestricted
delivery slot availability (Agatz et al, 2011). In
the case of thin margins, as for example in
e-groceries, these savings can have a huge
impact on profitability. The specific savings
potential depends on the customer density. Past
Internet failures often made the costly mistake
of over-servicing low volume areas. Thus, we
recommend critically assessing the minimum
demand required to justify the visit to an area,
considering both the average margin per order
and the future growth potential. Although
Order
Placement
Route
Planning
Capacity
Planning
Capacity
Planning
Order
Placement
Execution
Route
Planning
Airlines
Demand Management
Costs: Sunk
Demand Management
Costs: Variable
Execution
E-fulfillment
Figure 1: Planning processes – airlines and e-fulfilment.
Revenue management opportunities for Internet retailers
133& 2013 Macmillan Publishers Ltd. 1476-6930 Journal of Revenue and Pricing Management Vol. 12, 2, 128–138
certainly not negligible, the exact allocation of
the windows has a smaller cost impact than the
number of windows. The savings potential of a
smart allocation depends on the nature of the
delivery trips. The more customer areas visited
per route, the more important the coordination
of the delivery windows across these areas.
Standard routing softw ar e typically does not
support the above slotting decisions explicitly.
Therefore , companies are as of y et dependent on
custom-made solutions. Ho w e ver , our research
indicates that rough approximations already result
in substantial savings. Even simple spreadsheet
tools can be of significant value for this purpose.
DIFFERENTIATED PRICING
Pricing is probably the most obvious demand
management lever. In our context, this concerns
the delivery fee charged for an order. As for
any business, the appropriate pricing level has
to strike a balance between increasing de-
mand volume and decreasing revenue per order.
However, pricing also provides means for dif-
ferentiating between different delivery options
offered to the customer. In that sense, pricing
and slotting are complementary tools. Specifi-
cally, pricing can provide incentives for attract-
ing customers to a particular slot.
As attended delivery requires the customer to
be present, some delivery times, for example, in
the evening or in the weekend, are more
popular than others. From a marketing view,
this alone suggests charging different prices for
different delivery times. There is also an opera-
tions argument for this conclusion. Uniform
pricing typically results in imbalanced demand.
As the delivery capacity is relatively inflexible,
in general, this means either costly over-
capacity or losing peak-load demand. Differen-
tiated pricing, such as peak-load premiums and
off-peak discounts, help counter the above
effects by smoothing demand (see Figure 3).
Differentiated pricing is fairly intuitive and
is commonly practiced by many businesses,
including hotels (weekends versus weekdays)
and package delivery services (Monday–Friday
versus Saturday delivery). The pricing schemes
for several e-grocers are found in Table 1. The
difficulty with differentiated pricing lies in
determining the appropriate magnitude of dis-
counts and premiums. Excessive discounts not
Table 3: Classification of demand management
Capacity allocation Pricing
Static Differentiated slotting Differentiated pricing
Off-line, Regional demand clustering Demand smoothing
Forecast-based Balance service offering and delivery efficiency Increase capacity utilization
Dynamic Dynamic slotting Dynamic pricing
Real-time, Differentiate service Counterbalance underutiliz ed capacity
Order-based Maximize contribution of congested capaci ty Stimulate efficient routes
Avoid inefficient routes Segment on customer flexibility
Unrestricted demand Differentiated slotting
Driver 2
Driver 1
Shift 1 Shift 2
Slot 1+2
Slot 2
Slot 3+4
= Slot 1
= Slot 2
= Slot 3
= Slot 4
Figure 2: Differentiated slotting.
Agatz et al
134 & 2013 Macmillan Publishers Ltd. 1476-6930 Journal of Revenue and Pricing Management Vol. 12, 2, 128–138
only give away margins but also spoil the
market’s reference price. Systematic market
research and small incremental price changes
help avoid these dangers.
Fur thermore, delivery prices may also impact
the basket composition and corresponding rev-
enues of Internet retailing. In order to stimulate
sales, several retailers offer delivery fee discounts
for large orders. Other effects may be less
obvious. Albert.nl, for example, experienced
that a lower delivery fee at off-peak moments
attracted customers with a smaller basket size.
Pricing allows a differentiation not only
between different delivery times but also
between different lengths of the delivery win-
dows. Peapod for example, offers the customer
a $1 discount for choosing a delivery window
of 3.5 hours instead of 2 hours. In this way,
the Internet retailer can exploit differences
in the customers’ flexibility by offer ing
windows of different lengths simultaneously,
g ranting a discount for wider windows. In this
case, the discount should not only reflect the
willingness-to-pay of the different customer
segments but also the e-tailer’s efficiency gain
because of greater planning flexibility. This
closely links length-based pricing to the slotting
analysis discussed previously.
DYNAMIC SLOTTING
The previously discussed demand management
approaches are purely forecast-based in the
sense that they set conditions prior to receiving
the actual orders. However, even richer oppor-
tunities arise for Internet retailers through their
interaction with the customer s during the
actual sales process. Slotting and pricing deci-
sions then provide a means for managing
demand dynamically and in real time.
The most basic example of a dynamic slot-
ting decision regards closing a time slot once the
corresponding capacity is depleted. Although
some kind of capacity check is in fact a necessity
for any Internet retailer, an accurate assessment
of remaining available capacity is less obvious
than it may look at first sight. Effective capacity
involves the picking capacity in the warehouse,
physical fleet size and available driving time.
The latter depends on the clustering of orders
into routes and thereby directly links slotting to
transportation planning. Systematically, asses-
sing this interaction helps Internet retailers
increase their capacity utilization.
The potential of dynamic slotting goes much
further. The fundamental lesson from revenue
management is that there is a smarter way than
selling the capacity first-come-first-serve until its
depletion. Heterogeneous markets call for more
differentiation between orders. In the prototypi-
cal airline case, this leads to a trade-off between
selling a seat at a discount fare now versus
reserving it for a potential full-fare customer
later. Thus, in e-fulfillment it may be beneficial
to reserve scarce capacity, that is, busy time
slots, for the most profitable customers. Trans-
portation costs add another dimension to the
trade-off, namely, whether to serve a customer
in the given time slot or whether to try and
convert him to another slot that allows for
demand
Mo Tue We Thu Fr Sa Su
Capacity requirement
price
demand
Mo Tue We Thu Fr Sa Su
Capacity requirement
price
differentiated pricing
constant pricing
Figure 3: Differentiated pricing.
Revenue management opportunities for Internet retailers
135& 2013 Macmillan Publishers Ltd. 1476-6930 Journal of Revenue and Pricing Management Vol. 12, 2, 128–138
a more efficient delivery (Figure 4). In essence,
revenue management shifts the focus from
capability (Can we deliver this order at that
moment?) to profitability (Is it profitable to
deliver this order at that moment?).
Customers can be segmented in multiple
ways for this purpose. First, one may segment
by order size, or more generally by customer
value. Losing a large order from a regular
customer is worse than losing a small order
from an incidental customer. The second seg-
mentation is by delivery location. Given the
already accepted and the still expected future
orders, a delivery to A in a given slot may result
in a less costly route than a delivery to B.
Similarly, a delivery to A may be cheaper in slot
X than in slot Y. Third, the degree of customer
flexibility is of importance. A busy slot is best
used for the customer that is least willing to
accept an alternative slot.
There are many potential benefits of tailoring
the delivery service to individual customers
dynamically, rather than fixing everything up
front. Our simulation studies indicate potential
unit transportation cost savings of upto 20 per
cent (Campbell and Savelsbergh, 2005). In
addition, dynamic slotting enables Internet
retailers to offer shorter time windows without
hurting their delivery efficiency. This not only
improves customer service but also reduces the
risk of failed deliveries, which are a significant
challenge in Internet retailing. The specific
improvement potential of dynamic slotting in-
creases with the degree of customer flexibility,
heterogeneity between customers, the variabil-
ity of demand and the tightness of capacity.
Many Internet retailers face a significant market
g rowth. Dynamic slotting enables them to
increase the return on their current capacity,
thereby postponing expensive capacity expan-
sions.
In order to reap these benefits, companies
need good visibility of customer behavior and
delivery cost dependencies. Detailed transaction
data provides a rich source for analyzing custo-
mer response to different delivery options. We
see room for further improvement through a
better understanding of the customer choice
models in the marketing literature. On the cost
side, the analysis of dynamic slotting is closely
related with the planning of delivery routes.
The difficulty lies in the fact that the cost
impact of an individual request needs to be
estimated before all orders are known. In addi-
tion, all real-time computations have to be very
fast in order not to derail the customer’s order-
ing process. Several vendors of commercial
vehicle routing packages have started extending
their software to support real-time scheduling
(for example, Paragon, Descartes, Sidewinder,
ORTEC). For example, the Sidewinder real-
time scheduling tool enabled Sainsburys to
reduce the number of failed deliveries and at
the same time offer shorter time windows to
the customer, that is, Sainsburys’ 1-h promise
(www.sainsburys.com). As an easier yet bene-
ficial starting point, we advise extending a well-
designed static slotting schedule with a few
simple dynamic rules, for example, withholding
a slot from an order that lies fur ther apart than a
certain threshold from all other (accepted or
expected) orders in that same slot. The thresh-
old is dependent on the margin of the order and
on the risk of losing it.
DYNAMIC PRICING
Dynamic pricing provides an even richer tool
for real-time demand management. Pricing
Sell Slot X
to Customer A
Withhold Slot X
from Customer A
Customer A
accepts slot Y
Customer A
leaves
Sell Slot X
to Customer B
Slot X
remains idle
Slot X
remains idle
Sell Slot X
to Customer B
Slotting decision
upon arrival of
Customer A
Response
customer A
Future
demand
$
$$$
$
$
$
Figure 4: Dynamic slotting.
Agatz et al
136 & 2013 Macmillan Publishers Ltd. 1476-6930 Journal of Revenue and Pricing Management Vol. 12, 2, 128–138
allows a much finer gradation of incentives than
the yes-no type of decisions in slotting. In
addition, pricing can provide incentives for
pulling customers to a particular delivery option
whereas dynamic slotting pushes them away
from certain options.
Delivery-related price incentives can aim at
many different goals. Traditional airline revenue
management uses dynamic pricing as a means
for segmenting based on customers’ willingness
to book in advance. Typically, prices increase as
the departure draws nearer. In similar vein,
delivery services often segment their customers
based on their lead-time preferences, for exam-
ple, standard (5–7), 2 days or next day. As in the
airline case, customers with different lead-time
preferences and different willingness-to-pay
then compete for the same delivery slots.
Another option is to use price incentives,
namely discounts, to steer an order to a time
when it can be delivered efficiently. The under-
lying economics are similar to those of dynamic
slotting, involving the same trade-offs of
delivery efficiency and customer flexibility
(Campbell and Savelsbergh, 2006). For exam-
ple, discounts can be used for matching a
delivery with a visit to a nearby customer, and
for moving demand to temporarily underuti-
lized delivery periods, thereby enhancing
capacity utilization. The experience of Peapod
indicates that even a small discount (for exam-
ple, $1) can change the customer’s slot selection
(T. Parkinson, 2005, executive officer and
cofounder Peapod, personal communication).
One of the particular challenges of dynamic
pricing is its appropriate communication to the
customer. More than in the case of dynamic
slotting, customers may perceive unexpected
price changes as unfair (Xia et al, 2004). In
order not to overstrain the customer, we see
limited, well-targeted price discounts as more
appropriate in Internet retailing to date than full
dynamic pricing. To assure visibility of tempor-
ary price discounts, Internet retailer s may ap-
proach target customers proactively, for
example, by means of SMS or e-mail notifica-
tions. Another challenge concerns opportunistic
customer behavior. If discounts follow a regular
pattern customers will learn to anticipate them
and thereby limit the directive effect of the
pricing tool. This is another argument for a
careful use of dynamic price incentives.
CONCLUSIONS
To summarize, we have explained how active
demand management can benefit the profit-
ability of e-fulfillment services. The key driver
is differentiation. Revenue management has
shown that companies can do much better than
a one-size-fits-all first-come-first-serve strategy
when selling scarce capacity to a heteroge-
neous market. The same argument applies to e-
fulfillment. Customers differ in their willingness
to pay, their time preferences and their flexibility.
A smart strategy should recognize these differ-
ences through a differentiated service offering.
In contrast with classical revenue manage-
ment, the cost side has an important role in
e-fulfilment. Delivery costs differ between cus-
tomers and, for the same customer, between
different delivery windows. These cost effects
add a second dimension to demand manage-
ment in e-fulfilment.
Internet retailers have strong levers at their
disposal for actively steering demand. From a
fulfilment perspective, the offered delivery time
windows and their associated prices are of
particular relevance. Internet retailers can use
both of these levers off-line to manage systema-
tic demand patterns, such as weekly demand
peaks and regional demand clustering. Even
more importantly, however, they can adjust
time slot offering and delivery fees real-time,
based on actual orders, thereby tailoring their
service proposition to individual customers.
The future will allow even richer interactions,
such as negotiations through software agents.
Close cooperation between marketing and
operations is a prime prerequisite for success-
fully exploiting the great potential of demand
management in e-fulfilment. Although the im-
portance of the marketing-operations inter-
face is growing in many industries, its role in
Revenue management opportunities for Internet retailers
137& 2013 Macmillan Publishers Ltd. 1476-6930 Journal of Revenue and Pricing Management Vol. 12, 2, 128–138
Internet retailing is pivotal as demand manage-
ment decisions have an immediate and funda-
mental impact on delivery costs. Software
developments that support an integrated de-
mand management are encouraging. Several
vendors of routing software have taken first
steps toward extending their applications with
demand management capabilities, notably re-
garding time slot management. Standard solu-
tions, including dynamic pricing support, are
yet to come.
The potential of demand management
extends well beyond Internet retailing. In prin-
ciple, the above arguments apply to any business
that involves appointments with the customer.
This includes, for example, delivery of furniture
or kitchen appliances from a store or visits by a
repairman or a service engineer. The main
differences with Internet retailing concern the
pricing lev er. Typically, customers do not pay
separately for the deliv ery component in these
cases.However,thetimeslotofferingremainsa
po w erful steering instrument. In particular , dy-
namic slotting pr o vides a means for differentiation,
thereby increasing service without compromis-
ing efficiency. The key revenue management
lesson is to prioritize service to the most profit-
able customers. There is still a lot to be learned
from this lesson in the context of delivery
services.
REFERENCES
Agatz, N.A.H., Fleischmann, M. and van Nunen, J.A.E.E. (2008)
E-fulfillment and multi-channel distribution – a review.
European Journal of Operational Research 187(2): 339–356.
Agatz, N.A.H., Campbell, A.M., Fleischmann, M. and Savelsbergh,
M.W.P. (2011) Time slot management in attended home
delivery. Transportation Science 45(3): 435–449.
Boyer, K.K., Frohlich, M.T. and Hult, G.T.M. (2004) Extending
the Supply Chain: How Cutting-Edge Companies Bridge
the Critical Last Mile into Customers’ Homes. New York:
AMACON.
Campbell, A.M. and Savelsbergh, M.W.P. (2005) Decision sup-
port for consumer direct grocery initiatives. Transportation
Science 39(3): 313–327.
Campbell, A.M. and Savelsbergh, M. (2006) Incentive schemes
for attended home delivery services. Transportation Science
40(3): 327–341.
Demery, P. (2007) Where’s my package? Growing E-retailers
steer through a changing delivery scene. September, www
.internetretailer.com.
Punakivi, M., Yrjola, H. and Holmstrom, J. (2001) Solving the
last mile issue: Reception box or delivery box? International
Journal of Physical Distribution & Logistics Management 31(6):
427–439.
Talluri, K.T. and Van Ryzin, G.J. (2004) The Theory and Practice of
Revenue Management. New York: Springer.
US Census Bureau. (2012) E-stats Report.
Xia, L., Monroe, K.B. and Cox, J.L. (2004) The price is unfair! A
conceptual framework of price fairness perceptions. Journal of
Marketing 68(4): 1–15.
Agatz et al
138 & 2013 Macmillan Publishers Ltd. 1476-6930 Journal of Revenue and Pricing Management Vol. 12, 2, 128–138