Cornell University School of Hotel Administration

The Scholarly Commons

Articles and Chapters School of Hotel Administration Collection

12-2001

Function-space Revenue Management: A Case

Study from Singapore
Sheryl E. Kimes
Cornell University, sek6@cornell.edu

Kelly A. McGuire
Radiant Systems

Follow this and additional works at: http://scholarship.sha.cornell.edu/articles

Part of the Hospitality Administration and Management Commons

Recommended Citation
Kimes, S. E., & McGuire K. A. (2001). Function-space revenue management: A case study from Singapore [Electronic version].
Cornell Hotel and Restaurant Administration Quarterly, 42(6), 33-46. Retrieved [insert date], from Cornell University, School of
Hospitality Administration site: http://scholarship.sha.cornell.edu/articles/467/

This Article or Chapter is brought to you for free and open access by the School of Hotel Administration Collection at The Scholarly Commons. It has
been accepted for inclusion in Articles and Chapters by an authorized administrator of The Scholarly Commons. For more information, please contact
hlmdigital@cornell.edu.
Function-space Revenue Management: A Case Study from Singapore
Abstract
Hotels can apply revenue-management systems to their function spaces—and boost the revenue contribution
from those spaces.

Keywords
revenue management, yield management, strategies, hotel industry, Singapore

Disciplines
Hospitality Administration and Management

Comments
Required Publisher Statement
© Cornell University. Reprinted with permission. All rights reserved.

This article or chapter is available at The Scholarly Commons: http://scholarship.sha.cornell.edu/articles/467

 REVENUE MANAGEMENT HOTEL OPERATIONS

Function-space
Revenue Management
A Case Study from Singapore

Hotels can apply revenue-management systems to their function spaces—and boost the
revenue contribution from those spaces.

BY SHERYL E. KIMES AND KELLY A. MCGUIRE

R
evenue management, also known as yield manage- event duration to affect this measure. ConPAST has the fol-
ment, is commonly applied to hotels’ guest-room in- lowing three components: contribution, space, and time. We
ventories but the practice has not yet been systemati-
ventories, suggest that measuring contribution is preferable to measur-
cally applied to hotels’ function spaces.1 Hotels’ function ing revenue because revenue alone doesn’t account for the
spaces provide substantial income that could be enhanced by varying profit margins arising from renting function spaces.
applying revenue-management, as we explain in this article. Space refers to the number of square feet or meters in the
As is the case in all revenue-management applications, the function room, and time refers to the time segment for which
goal of function-space revenue management is to maximize the revenue contribution is measured. In this paper, we ex-
the revenue contribution of each function space for each time plain how we developed a revenue-management strategy for
period that the space is available. The measure we apply for the Raffles City Convention Center at the Swissôtel Stam-
this purpose is contribution per available space for a given ford and Raffles Plaza in Singapore. (While those properties
time (ConPAST). Hotel managers can manipulate price and are now managed by Swissôtel, at the time of our study the
hotels were managed by Westin.)
1
Jerry Hartley and Peter Rand, “Conference-sector-capacity Management,”
in Yield Management for the Service Industries, second edition, ed. Ingold,
McMahon-Beattie, and Yeoman (London: Continuum, 2000), pp. 315–338. © 2001, CORNELL UNIVERSITY

DECEMBER 2001 Cornell Hotel and Restaurant Administration Quarterly 33

HOTEL OPERATIONS REVENUE MANAGEMENT

EXHIBIT 1 cording to predicted demand levels so that price-

sensitive customers who are willing to purchase
A Typology of Revenue Management at off-peak times can do so at favorable prices,
while price-insensitive customers who want to
Price purchase at peak times will be able to do so. The
Fixed Variable application of revenue management has been most
effective when it is applied to operations that have
Quadrant 1 Quadrant 2 relatively fixed capacity, demand that is variable
and uncertain, perishable inventory, appropriate
Predictable

Movies Hotel rooms cost and pricing structures, and varying customer
Stadiums and arenas Airline seats
Convention centers Rental cars price sensitivity.3 Those attributes are found in
Hotels’ function space Cruise lines the business of renting hotels’ function space.
Duration

Different industries are subject to different

combinations of duration control and variable
pricing (see Exhibit 1).4 Industries traditionally
Quadrant 3 Quadrant 4 associated with revenue management (e.g., ho-
Unpredictable

tels, airlines, car-rental firms, and cruise lines) are

Restaurants Continuing care
Golf courses Hospitals able to apply variable pricing for a product that
Internet-service providers has a specified or predictable duration (Quadrant
2). On the other hand, sellers of function space,
movie theaters, performing-arts centers, and
arenas charge a fixed price for a product of pre-
dictable duration (Quadrant 1), while restaurants
A version of this diagram was previously used in: Sheryl E. Kimes, “Revenue Man- and golf courses charge a fixed price but face a
agement on the Links: Applying Yield Management to the Golf-course Industry,” relatively unpredictable duration of customer use
Cornell Hotel and Restaurant Administration Quarterly, Vol. 41, No. 1 (February (Quadrant 3). Many health-care businesses charge
2000), p. 127; and most recently in: Lawrence R. Weatherford, Sheryl E. Kimes,
and Darren A. Scott, “Forecasting for Hotel Revenue Management: Testing Aggre- variable prices (e.g., Medicare versus private pay),
gation Against Disaggregation,” Cornell Hotel and Restaurant Administration but cannot predict the duration of patient use,
Quarterly, Vol. 42, No. 4 (August 2001), p. 54. even though some may try to control that dura-
tion (Quadrant 4). The lines dividing the quad-
rants are broken because in reality no fixed de-
marcation point exists between any of the
Revenue-management Review quadrants. Thus, a given enterprise (such as rent-
Revenue management is the application of in- ing function space) may have attributes from more
formation systems and pricing strategies to match than one quadrant.
customers with services at an appropriate time Successful revenue-management applications
and price. A common formulation of revenue are generally found in Quadrant-2 industries,
management is selling the right capacity (in this because they can manage both capacity and price.
case, function space) to the right customer at the To obtain the benefits associated with revenue
right place at the right time.2 The determination
of “right” in this instance entails achieving both 3
See: Sheryl E. Kimes, “Yield Management: A Tool for
the most contribution possible for the hotel in Capacity-constrained Service Firms,” Journal of Operations
renting a function space while also delivering the Management, Vol. 8, No. 4 (November 1989), pp. 348-
363; and Robert G. Cross, Revenue Management (New York:
greatest value or utility to the customer for the Broadway Books, 1997). See also: Robert G. Cross,
reservation and use of that space. In practice, rev- “Launching the Revenue Rocket: How Revenue Manage-
enue management has meant setting prices ac- ment Can Work for Your Business,” Cornell Hotel and Res-
taurant Administration Quarterly, Vol. 38, No. 2 (April
1997), pp. 32–43.
2 4
Barry C. Smith, John F. Leimkuhler, and Ross M. Darrow, Sheryl E. Kimes and Richard B. Chase, “The Strategic
“Yield Management at American Airlines,” Interfaces, Levers of Yield Management,” Journal of Service Research,
Vol. 22, No. 1 (January–February 1992), pp. 8–31. Vol. 1, No. 2 (November 1998), pp. 156–166.

34 Cornell Hotel and Restaurant Administration Quarterly DECEMBER 2001

 REVENUE MANAGEMENT HOTEL OPERATIONS

management, industries outside of Quadrant 2 have a great deal of difficulty with forecasting
should attempt to emulate Quadrant-2 businesses their demand for group rooms, and most have
by deploying the appropriate strategic levers (for not even attempted to forecast their function-
example, by controlling duration). This implies space demand.6 If a hotel can obtain informa-
that Quadrant-1 businesses, such as renting func- tion on the lead time and the amount of busi-
tion space, should concentrate on developing ness of major market segments, it can make better
demand-based pricing approaches. This is not to decisions regarding the level of demand from key
say that duration management cannot be im- customers, when that demand will occur, and
proved, but only that most of the benefits asso- when to release space to other market segments.
ciated with revenue management will be obtained While hotels have a difficult time in forecast-
from differential pricing. ing function-space demand, they do a good job
In the next sections we review some of the of reducing cancellation and no-show rates
options available for managing meeting-space du- by requiring nonrefundable deposits and pre-
ration, revenue, and demand. payments for function space. In some high-
demand cities, hotels even require prepayment
Function-space Revenue for the requested number of guest rooms con-
Management nected to the function-space reservation. This
Managing function-space revenue is more diffi- practice protects those hotels from holding room
cult than dealing with transient guest rooms blocks that no one picks up at the agreed-on rate.
because of the interaction of function space with Duration uncertainty. Since events can span
room sales and food and beverage demand. While multiple days or day parts, hotels must ensure
function space will reap most of its revenue- that the guest rooms associated with the event
management benefits from pricing, this is not to do not unnecessarily displace higher-paying tran-
downplay the role of duration control. sient business and must also ensure that the event
Duration control. Duration can be managed lasts for the agreed-on length of time. The pre-
in four ways: refining the definition of duration, payment that we just mentioned also helps en-
reducing arrival uncertainty, reducing duration sure that the contracted function space is used
uncertainty, and reducing the amount of time for the required length of time. However, hotels
between functions. 5 Each is briefly described and have a much more difficult time ensuring that
the issues associated with each discussed. the guest rooms will be occupied for the speci-
Definition of duration. Most hotels define fied length of time. Many groups, particularly
function-space duration by day part, but the defi- conventions, suffer from high room attrition on
nition and number of day parts varies by hotel. the last day or two of the event. If group mem-
Since events can span multiple day parts or use bers check out early, the hotel is usually stuck
only a fraction of a day part, we suggest that ho- with an empty room, unless the hotel also im-
tel function-space managers define duration as poses early departure fees or requires prepayment
an hour. Most computerized systems track this for all rooms. Then again, early departure fees and
information, but it may be difficult to retrieve. room prepayments may incur customers’ wrath.
If it’s impossible to track the function-room Turnaround time. In high-demand periods,
schedule by the hour, tracking by day parts will the time needed to turn the room around can
work, as we demonstrate below. affect the hotel’s ability to meet demand. Hotels
Arrival uncertainty. Arrival uncertainty in- that try to minimize the labor costs associated
volves two factors: (1) the timing and quantity with set-up and tear-down often end up turning
of requests, and (2) the no-show and cancella- down business that could have been accommo-
tion rates of booked events. Determining the tim- dated if the hotel had employed a sufficient num-
ing and number of requests from different mar- ber of employees to turn over the room. For ex-
ket segments is a challenging task. Hotels still
6
Sheryl E. Kimes, “Group Forecasting Accuracy for Ho-
tels,” Journal of the Operational Research Society, Vol. 50,
5
Ibid. No. 11 (November 1999), pp. 1104–1110.

DECEMBER 2001 Cornell Hotel and Restaurant Administration Quarterly 35

HOTEL OPERATIONS REVENUE MANAGEMENT

ample, a hotel that requires a two-hour set up day (e.g., customers who book events for week-
and two-hour break-down for all events would ends or mornings receive a discount).
not be able to book a meeting that ends at 5:00
in the same room with a dinner that begins at RM Program for Function Spaces
6:30. If more employees were assigned to speed To develop a revenue-management program for
the transition between events, however, the labor function spaces, managers should:
cost would be more than covered by the increased (1) establish the baseline of performance,
revenue associated with the incremental event. (2) understand the drivers of that performance,
Pricing. Function-space pricing is complicated (3) develop a revenue-management strategy,
by the fact that function space cannot be priced (4) implement that strategy, and
without considering its effect on room sales and (5) monitor the strategy’s outcome.7
food and beverage. A price that may seem too In the following sections, we will discuss how we
low for the function space when considered alone developed a revenue-management program for
may be more than compensated for when room the Raffles City Convention Center at the
and food and beverage revenue are considered. Swissôtel Stamford and Raffles Plaza hotels in
Similarly, a high function-space price may ulti- Singapore. We will conclude the case study with
mately be unprofitable for the hotel if it occa- a discussion of the steps a hotel should take to
sions a low guest-room rate and displaces poten- implement function-space revenue management.
tially higher-paying business from the transient
rooms. The Case of the Raffles City
Function-space managers face two pricing Convention Center
decisions: what price to charge and how to de- The Raffles City Convention Center (RCCC) at
termine who pays which price. When analyzing the Swissôtel Stamford and Raffles Plaza in
which business to accept, the manager must as- Singapore will be used to demonstrate how rev-
sess the days and day parts requested, the num- enue management can be applied to function
ber of function rooms required, the expected space.8 The RCCC is reasonably typical of a con-
number of guest rooms sold, and the expected vention hotel’s function space. The convention
sales of food and beverage, as well as other items. center occupies a 70,000-square-foot space on
Since the profit margins for different revenue the hotel’s fourth floor, comprising 18 function
streams vary, the manager must determine the rooms ranging in size from 20-person meeting
expected contribution associated with each event rooms to the Raffles Ballroom, which holds over
and compare that with any potential displace- 1,000 people for a sit-down dinner. The three
ment of other business. If the expected contri- ballrooms (Raffles Ballroom, Atrium Ballroom,
bution is higher than the potential displacement, and Stamford Ballroom) drive convention, wed-
the group should be accepted. Otherwise, its ding, and social revenue, while three sets of meet-
business should be declined or referred to another ing rooms (the Executive Conference Center,
day or day part. Plaza Meeting Rooms, and Stamford Meeting
Along with determining the appropriate price Rooms) handle demand for small meetings and
to charge, the hotel must also justify the prices conference breakout sessions. Each of the ball-
charged to different clients. Rate fences can be rooms can be divided into three sections, and
used to determine who pays which price. So- the Plaza and Stamford Meeting Rooms also can
called “physical” rate fences might include the
presence of certain amenities (e.g., high-tech 7
Sheryl E. Kimes, “Implementation of Restaurant Revenue
rooms command a premium) or location (e.g., Management: A Five-Step Approach,” Cornell Hotel and
rooms on the first floor command a premium). Restaurant Administration Quarterly, Vol. 40, No. 3 (June
Non-physical rate fences might include customer 1999), pp. 15–22.
characteristics (e.g., repeat customers receive a 8
Much of this section comes from: Kelly A. McGuire,
discount), transaction characteristics (e.g., cus- “Function-space Revenue Management: A Case Study of
the Raffles Convention Center at the Westin–Singapore,”
tomers who book events through the internet Cornell University School of Hotel Administration, unpub-
receive a discount), or day of week or time of lished monograph, 2001.

36 Cornell Hotel and Restaurant Administration Quarterly DECEMBER 2001

 REVENUE MANAGEMENT HOTEL OPERATIONS

be divided. Two high-tech rooms provide pro- Expo Center has guest rooms of its own, how-
jection and sound equipment for executive meet- ever, so RCCC competes by focusing on its abil-
ings. Foyer space is used for registration and cof- ity to provide the services of a full-service hotel,
fee breaks. in addition to its function space. Because they
The director of marketing for the hotel over- have so many guest rooms, the Swissôtel Stam-
sees the four sales groups, which are divided ford and Raffles Plaza’s secondary competitive set
among weddings, conventions and exhibitions, includes other hotels in Singapore. However, the
meetings, and dinners and dances. Each sales RCCC is nearly double the size of any of those
group has its own director and team of sales man- hotels’ function space, which forms a point of
agers. Each subgroup director is responsible for differentiation compared to the other hotels.
setting the sales targets for her team, and all four
directors work together to set prices for the next Step 1: Establish a Baseline
year. A catering-sales analyst, who reports directly The first step in implementing a revenue-
to the director of revenue management, works management system for a property’s function
with all four sales groups. Her primary responsi- space is to establish the baseline performance for
bilities are to oversee the programming and ad- that space. Baseline-performance measures
ministration of the Delphi catering- and sales- should include both performance characteristics
system software and to generate reports and of the meeting space and demand behavior of
analyses. the various market segments. The performance
The Raffles City Convention Center drives measurements should include both occupancy
guest-room sales at the Swissôtel Stamford and and our proposed revenue-contribution factor
Raffles Plaza, and the function space is frequently (i.e., contribution per available space-time).
sold at a reduced rate to generate room-nights. These should be applied to the entire space, each
This arrangement is not atypical for a hotel of room type, each day of week, and each day part.
this size. The hotel has 2,000 rooms, divided into Demand behavior should include information on
three main sections: the Stamford, and the North the demand patterns and booking patterns for
and South Towers of the Plaza. Compared to the each market segment by month, by day of week,
Plaza’s towers, the Stamford is a lower-rated prop- and by day part.
erty in terms of décor and amenities offered, but When analyzing space performance, hotels
at 70 stories it is the tallest hotel in the world. should look to the traditional measures used for
The Plaza is more upscale than the Stamford. The tracking guest-room sales: occupancy and
South Tower has recently been renovated and RevPAR. With few exceptions, hotels track oc-
commands a considerable rate premium. Plans cupancy on a daily basis, and most hotels also
have been made to renovate the North Tower, and track their RevPAR. Very few hotels know the
eventually the Stamford itself will be upgraded. occupancy of their function space and even fewer
The function-space market in Singapore is di- know their ConPAST.
vided into two segments: convention centers and Occupancy. Function-space occupancy is
hotels. Singapore has two convention centers: tricky to calculate because a single room can be
Suntec City, located three blocks from the divided (or not) and the appropriate time unit
RCCC, and the Expo Center, which is near the must be determined. As we indicated above, we
airport, or about 25 kilometers from downtown. believe that the best time unit to use in calculat-
The major-hotel market includes Marriott, ing function-space occupancy is one hour, but
Hyatt, Hilton, Pan Pacific, and the Oriental, all most hotels do not have sufficiently accurate data
located within a few kilometers of the Swissôtel for a calculation at this level. Because of this, the
Stamford and Raffles Plaza. most practical time unit is the day part. The num-
The RCCC considers its primary competitive ber of day parts per day varies by hotel, but it is
set to be the two convention centers—chiefly, usually related to the number of times each day
Suntec City. Until Suntec City was built, the that a function space can be sold. Most hotels
RCCC had the largest function space in down- use two or three day parts per day for each func-
town Singapore. Neither Suntec City nor the tion space.

DECEMBER 2001 Cornell Hotel and Restaurant Administration Quarterly 37

HOTEL OPERATIONS REVENUE MANAGEMENT

EXHIBIT 2 booked was multiplied by that percentage and

added to the values for the other individual rooms
PM Occupancy for November and December 2000 to determine the total number of events booked
in that type of space. Thus, if two of the indi-
Facility Sun Mon Tue Wed Thu Fri Sat Total vidual Plaza meeting rooms were occupied, the
occupancy for the overall Plaza meeting rooms
Atrium Ballroom 52% 38% 71% 48% 81% 89% 100% 69%
was 50 percent, and if three of the rooms were
ECC: Bailey 33% 38% 50% 56% 67% 33% 33% 44% occupied, the occupancy was 75 percent. Using
ECC: Bras Basah 44% 63% 44% 39% 56% 56% 44% 49% this rubric, if the dividers were not in use for a
ECC: Orchard 56% 50% 54% 48% 70% 59% 33% 53% particular space, its occupancy was by definition
100 percent for that particular day part. The only
ECC: Wilberforce 0% 63% 75% 56% 67% 56% 33% 49%
exception to this rule is how we treated the rooms
Plaza rooms 28% 53% 72% 69% 83% 81% 61% 64%
in the Executive Conference Center. Since the
Raffles Ballroom 67% 42% 42% 44% 52% 89% 85% 61% ECC rooms cannot be combined into one space,
Stamford Ballroom 111% 63% 56% 72% 106% 89% 133% 91% we calculated the occupancy for each of those
Stamford meeting room 22% 23% 42% 41% 44% 43% 39% 36% rooms separately.
As indicated in Exhibits 2 and 3, PM occu-
pancy was much higher than AM occupancy in
November and December. Exhibit 2 shows that
Function-space occupancy can be calculated overall (total) PM occupancy ranged from 36 per-
by dividing the number of day parts used by the cent in the Stamford meeting rooms to 91 per-
number of day parts available. For example, if a cent in the Stamford Ballroom. Occupancy also
hotel has ten function rooms and three day parts varied by day of week. For example, the occu-
per day, its monthly capacity available (assum- pancy of the popular Stamford Ballroom ranged
ing a 30-day month) would be 900 day parts (10 from 56 percent on Tuesdays to 133 percent on
× 3 × 30). If the hotel sold a total of 450 day Saturdays (indicating that it was turned more
parts of function space in a month, its occupancy than once during the PM day part). As the table
would be 50 percent. in Exhibit 3 shows, overall AM occupancy ranged
With that calculation in mind, we obtained from 3 percent in the Raffles Ballroom to 48
detailed information on functions booked at percent in the Bailey Meeting Room (one of the
these Singapore hotels from their Delphi system ECC rooms). Occupancy tended to be a bit
for November 2000 through May 2001. We ex- higher during the week than on weekends.
cluded coffee breaks in our calculation of occu- Ballroom occupancy. Since the ballrooms
pancy and contribution figures, to reduce the could be divided, we tracked the occupancy of
possibility of double counting room use. both the ballroom parts and each entire ballroom.
We used just two day parts, AM (before 5:00 As hoteliers know only too well, when one divi-
PM) and PM (after 5:00 PM). On that basis, we sion of a ballroom is booked, the entire space
tracked daily occupancy for each major meeting cannot then be rented—just the other section(s).
area (Atrium Ballroom, ECC rooms, Plaza meet- This factor created intriguing occupancy patterns.
ing rooms, Raffles Ballroom, Stamford Ballroom, The Stamford Ballroom, for instance, had a
and Stamford meeting rooms). higher occupancy when the entire space was
As mentioned previously, all of the meeting rented as a unit. On the other hand, the Atrium
areas in our convention center can be divided Ballroom and Raffles Ballroom had higher week-
into smaller spaces (except for those in the ECC). day occupancies when they were sold as divisible
To calculate the rooms’ occupancy, we expressed spaces on weekdays, but as the entire space on
each meeting space as a percentage of the total weekends.
space for that particular venue. For example, the ConPAST. As we suggest above, hotels should
hotel has four Plaza meeting rooms, so each of also calculate each space’s contribution per avail-
those rooms was considered as 25 percent of the able space-time to determine that space’s contri-
total. The number of times a specific room was bution to operating profit. Again, we argue that

38 Cornell Hotel and Restaurant Administration Quarterly DECEMBER 2001

 REVENUE MANAGEMENT HOTEL OPERATIONS

contribution is preferable to revenue because of EXHIBIT 3

the varying profit margins of the different streams
of function-space revenue. For example, food and AM Occupancy for November and December 2000
beverage expenditures usually have a contribu-
tion margin of 30 to 35 percent, while room Facility Sun Mon Tue Wed Thu Fri Sat Total
rental may have a contribution margin of 85 to
Atrium Ballroom 0% 29% 29% 30% 33% 11% 26% 22%
95 percent, and AV rentals may have a contribu-
tion margin ranging from 50 to 95 percent. Re- ECC: Bailey 33% 38% 75% 56% 67% 33% 33% 48%
lying only on revenue generation (as opposed to ECC: Bras Basah 33% 50% 25% 44% 22% 22% 11% 30%
contribution) can lead hotel managers to develop ECC: Orchard 56% 25% 4% 48% 41% 44% 30% 36%
unprofitable selling patterns.
ECC: Wilberforce 0% 63% 63% 56% 56% 56% 33% 46%
Contribution should be calculated by space
Plaza rooms 17% 31% 47% 58% 47% 47% 22% 39%
(either square feet or square meters) because this
measurement provides a standard for the com- Raffles Ballroom 0% 4% 4% 0% 0% 11% 4% 3%
parison of different types of rooms and room Stamford Ballroom 28% 31% 25% 17% 28% 17% 44% 27%
configurations. For example, if a hotel has a di- Stamford meeting room 19% 15% 33% 33% 41% 39% 26% 30%
visible ballroom and finds that the ConPAST is
higher when the room is divided than when it is
sold as one space, it should direct its marketing
efforts at smaller groups or ensure that larger booked (constrained demand) and those that
groups are charged a premium. were turned down (unconstrained demand).
Most hotels have information on their guest- Booking behavior shows how far in advance par-
rooms’ competitive performance, but very few ticular market segments make their reservations.
have information on the competitive perfor- Information on booking behavior can help
mance of their function space. In some North hotels better forecast sales and improve decisions
American markets, HotelRevMax™ provides this on when to release space to different market
information. For example, if a hotel has a segments.
monthly ConPAST of $5 and its competitive set Constrained and unconstrained demand. As
has a monthly ConPAST of $4, its yield index with guest-room reservations, hotels need infor-
would be 1.20 ($5/$4). mation on both the constrained and uncon-
The contribution per available space-time was strained demand for function space.9 To review
calculated by dividing the contribution per day this topic, the level of constrained demand can
part (AM and PM) by the amount of space (num- be determined from actual sales, while uncon-
ber of square feet) for each meeting room type strained demand can be estimated by declined
(see Exhibits 4 and 5 on the next page). reservations, or the requests that could have been
In the PM day part, the Raffles Ballroom had accommodated if the hotel had unlimited space.
by far the highest overall ConPAST ($4.21), fol- We observe that most hotels do a poor job of
lowed by the Stamford Ballroom at $0.73. The tracking unconstrained demand, but such infor-
ConPAST of the Raffles Ballroom was highest mation is essential for understanding the true
on Fridays ($5.59), Saturdays ($7.71), and nature of demand. Take the case of a convention
Sundays ($4.81). Mondays were low, at $1.90. that occupies a hotel’s entire function space for a
The AM day part had an extremely low week. Had other groups wished to book space
ConPAST across the board (Exhibit 5). The Ex- during that time, they would have been turned
ecutive Conference Center had the highest over- down due to space constraints. If function-space
all ConPAST ($0.24), followed by the Plaza
meeting rooms ($0.17).
Demand behavior. Information on the de- 9
See: Eric Orkin, “Wishful Thinking and Rocket Science:
mand and booking behavior of each market seg- The Essential Matter of Calculating Unconstrained Demand
for Revenue Management,” Cornell Hotel and Restaurant
ment should be tracked and analyzed. Demand Administration Quarterly, Vol. 39, No. 4 (August 1998),
behavior includes both functions that were pp. 15–19.

DECEMBER 2001 Cornell Hotel and Restaurant Administration Quarterly 39

HOTEL OPERATIONS REVENUE MANAGEMENT

EXHIBIT 4 and the number of regrets. Unless an automated

tracking system can be developed, it is unlikely
PM ConPAST that accurate information will be captured.
Booking behavior. Information on how far
Facility Sun Mon Tue Wed Thu Fri Sat Total
in advance various market segments make their
Atrium Ballroom $0.36 $0.06 $0.24 $0.15 $0.34 $0.46 $0.66 $0.33 bookings can help hotel managers improve their
ECC $0.36 $0.21 $0.25 $0.29 $0.30 $0.65 $0.42 $0.36 forecasting and can also help them decide when
to release space to different market segments. For
Plaza rooms $0.21 $0.32 $0.27 $0.22 $0.34 $0.38 $0.39 $0.30
example, most hotels have a conflict between
Raffles Ballroom $4.81 $1.90 $2.47 $4.17 $2.35 $5.59 $7.71 $4.21 catering sales and food and beverage, as the two
Stamford Ballroom $1.11 $0.49 $0.28 $0.48 $0.70 $0.79 $1.18 $0.73 departments attempt to book the same space. If
Stamford meeting room $0.21 $0.09 $0.16 $0.15 $0.17 $0.14 $0.18 $0.16 sufficient booking information were available,
managers would be able to make better decisions
on which type of business to accept.
The booking curve shows the booking pace
EXHIBIT 5 of each segment. Booking-curve data can be used
to set holds and releases for function spaces and
AM ConPAST to give an indication of the comparative revenue
performance of a particular month (i.e., are we
Facility Sun Mon Tue Wed Thu Fri Sat Total behind or ahead of typical bookings?). The book-
Atrium Ballroom $0.00 $0.04 $0.04 $0.05 $0.06 $0.02 $0.06 $0.04 ing curves we developed were based on average
ECC $0.29 $0.17 $0.11 $0.26 $0.15 $0.47 $0.19 $0.24
revenue-on-hand (RevOH) for each month be-
fore arrival (MBA).
Plaza rooms $0.18 $0.17 $0.19 $0.19 $0.13 $0.11 $0.19 $0.17
A steep booking curve indicates a fast book-
Raffles Ballroom $0.02 $0.21 $0.16 $0.18 $0.00 $0.01 $0.09 $0.09 ing pace while a gradual slope illustrates a slow
Stamford Ballroom $0.17 $0.14 $0.08 $0.06 $0.04 $0.03 $0.17 $0.10 and steady pace. Knowing the booking pace can
Stamford meeting room $0.20 $0.06 $0.16 $0.13 $0.17 $0.10 $0.12 $0.13 help sales managers decide on release dates for
spaces. Thus, if a sales manager knows that nearly
all conventions are booked by 12 months before
sales alone were considered in this instance, ho- arrival, she can release all of the larger ballrooms
tel managers would be overlooking the presence that are not booked for conventions a year ahead
of additional demand. They might well make of time, because it is unlikely that there will be
biased future allocation decisions because they additional requests for a convention.
lack information about the business that was
turned down. A Look at Demand
Unconstrained demand can be divided into The Swissôtel Stamford and Raffles Plaza serve
denials and regrets. Denials are customers who five major market segments: convention and ex-
are turned away because of capacity constraints hibition, meetings, weddings, social, and tours.10
or restrictions. For example, if a function space Except for tours, those market segments were rela-
is not available, or if sufficient guest rooms are tively balanced in 2000. Social events had the
not available, the hotel may be forced to decline largest percentage of total revenues that year (28
the request for event space. Regrets are custom- percent), while conventions were close behind at
ers who are turned down because they are not 24 percent. Weddings and meetings each pro-
inclined to pay the price of the space or a room vided 23 percent of the hotel’s revenues, while
(i.e., shoppers). Those inquiries may not be a true tours contributed only 3 percent. We’ll use the
representation of customers who intended to convention-and-exhibition and meetings seg-
book. For this reason, it is more important to ments to illustrate the use of demand and book-
track denials than to be concerned about regrets. ing data.
Tracking unconstrained demand implies that
sales managers must tally the number of denials 10
Much of this section comes from: McGuire, op. cit.

40 Cornell Hotel and Restaurant Administration Quarterly DECEMBER 2001

 REVENUE MANAGEMENT HOTEL OPERATIONS

Conventions. Conventions are major revenue EXHIBIT 6

generators for the hotel, since they typically bring
in a significant number of room-nights along with Convention and exhibition booking curve
the function-space revenue and additional F&B
sales. Since the events are so large and are booked
far in advance, they form the cornerstone of the $600,000
function-space use. $500,000
The convention-segment managers report that
$400,000
their peak month is October. The average num-
ber of covers per function for conventions in 2000 $300,000
was 290. This average includes all breakout ses- $200,000
sions and VIP luncheons, as well as the large con- $100,000
vention banquets and exhibit-hall food service. —0— I I I I I I I I I I I I I I
Convention managers report an average length 0 2 4 6 8 10 12 14 16 18 20 24 28 33 35
of stay of four days per convention. Most con-
vention and exhibition events begin in the morn- Months before arrival
ing—most frequently at 9:00 AM.
The 2000 bookings for conventions and ex-
hibitions started earlier than for other market
segments (Exhibit 6). For instance, the hotel al- EXHIBIT 7
ready had convention RevOH at 36 months be-
fore arrival. Half of the RevOH had been booked
by 19 months before arrival and 90 percent had
Meetings booking curve
been booked by seven months before arrival.
Since 75 percent of all functions were booked $500,000
by one year before the event, the property could $450,000
begin to release unbooked space to other seg- $400,000
ments at the one-year point. At seven months $350,000
before arrival it could release all remaining space,
$300,000
as few convention requests are likely to come in
after that point. $250,000
Meetings. Although the percentage of total $200,000
revenue of the meetings segment is close to that $150,000
of conventions, meetings are usually smaller in $100,000
size, with an average of only 69 covers per event.
$50,000
Many more meetings must be booked to gener-
ate revenues equivalent to those of conventions —0— I I I I I I I I I I I I I I
00 2 4 6 8 10 12 14 16 22 26 30 32 34 36
and exhibitions. For example, in 2000, 3,190
meeting events were booked, compared to 710 Months before arrival
convention and exhibition events. Meetings typi-
cally last one day, occur during the week, and
mainly comprise morning and afternoon sessions,
each with a coffee break. Sales managers report
peak seasons in October, November, and May.
December, January, and July are off peak.
The meetings’ booking curve stands in con-
trast to the booking curve for conventions (Ex-
hibit 7). Relatively few meeting bookings oc-
curred until six to eight months before arrival.
So, just as convention bookings began to slow

DECEMBER 2001 Cornell Hotel and Restaurant Administration Quarterly 41

HOTEL OPERATIONS REVENUE MANAGEMENT

down, meetings bookings started to pick up. External factors include:

Based on that analysis, hotel managers can be (1) City-wide events—Some city-wide events
reasonably sure that they will get meeting requests may be so large that it precludes having
until close to the actual event. The meetings busi- space available for other groups; and
ness is a good complement to conventions, be- (2) Market conditions—If the economic con-
cause the two segments have opposite booking dition of the overall market has deterio-
patterns. Therefore, the hotel can turn away most rated, business from certain market seg-
meeting requests until seven months before ar- ments may decrease.
rival—which is about the time that the remain- An example of internal factor number six
ing convention space is released. (above) occurred at the RCCC convention cen-
Understanding demand and booking patterns ter in July 2000, when an extremely large con-
for each segment is important when developing vention used all of the hotel’s function space. That
an appropriate RM strategy. For example, it helps booking resulted in the denial of requests from
the hotelier determine which spaces to hold if other customers. July is normally a slow month
she knows that weddings typically book week- for conventions, but the large convention made
end evenings, conventions book weekdays, and July look like a good month.
that both groups are similar in size. If a conven- The Atrium Ballroom may be an example of
tion wanted to set up on a Sunday, which is a internal factor number one (above). That space
peak day for weddings, for example, the hotel recorded a much lower occupancy and ConPAST
could charge a premium fee for that privilege. than did either the Stamford or Raffles Ballrooms.
This weak performance may have been caused
Step 2: Understanding Causes of by the physical characteristics of the Atrium Ball-
Booking Behavior room. It has a low ceiling, which makes it less
Once the baseline has been established, manag- desirable for weddings and social events, and the
ers must understand the causes behind past perfor- space dividers are not soundproof, so that mul-
mance and must develop a revenue-management tiple small events cannot be booked into the
strategy to help improve performance. A thorough space—that is, the Atrium can be rented either
understanding of the internal and external envi- as a large space or a single small space, but not as
ronment of the hotel and its sales policies is es- multiple small spaces simultaneously.
sential for identifying performance drivers and
for developing revenue-management strategies. Step 3: Developing a Revenue-
Potential causes can be divided into internal and management Strategy
external factors. Effective revenue management is predicated on
Internal factors include: the use of demand-based pricing in which full
(1) Physical constraints—Rooms may not be price is charged during high-demand periods,
large enough for certain meetings or may while discounted rates are available during low-
have other limiting characteristics; demand periods. To apply demand-based pricing,
(2) Labor availability—The hotel may not it is essential to identify high- and low-demand
have enough workers available to turn periods for function space.
rooms around between events; A simple approach to identifying and com-
(3) Sales incentives—The sales-incentive sys- municating high- and low-demand periods is
tem may preclude making good revenue- to color code them on a table. This approach has
management decisions; been applied effectively in hotel rooms and
(4) Incomplete data—The data available may restaurants.11 For example, high-demand periods
be inaccurate or difficult to obtain; are designated as hot, or red; periods of moder-
(5) Guest-room availability—Function-space
requests often hinge on whether sufficient
11
guest rooms are available; and Sheryl E. Kimes, Deborah I. Barrash, and John E.
Alexander, “Developing a Restaurant Revenue-management
(6) Prior bookings—Other business may have Strategy,” Cornell Hotel and Restaurant Administration Quar-
already booked the space. terly, No. 40, Vol. 5 (October 1999), pp. 18–30.

42 Cornell Hotel and Restaurant Administration Quarterly DECEMBER 2001

 REVENUE MANAGEMENT HOTEL OPERATIONS

ately high demand are designated as warm, or EXHIBIT 8

yellow; and low-demand periods are designated
as cold, or blue. All other demand levels are clas- AM demand levels
sified as neutral and not colored at all.
Each demand level (or color) should have a
different set of prices associated with it, based on Facility Sun Mon Tue Wed Thu Fri Sat Total
qualified discounts with rate fences. The highest
prices should always be available regardless of the Atrium Ballroom 0 29% 29% 30% 33% 11% 26% 22%

demand level, but a variety of qualified discounts ECC: Bailey 33% 38% 75% 56% 67% 33% 33% 48%
should be offered during other periods. A quali- ECC: Bras Basah 33% 50% 25% 44% 22% 22% 11% 30%
fied discount implies that customers must meet ECC: Orchard 56% 25% 4% 48% 41% 44% 30% 36%
some restriction (or rate fence) to receive the
ECC: Wilberforce 0 63% 63% 56% 56% 56% 33% 46%
lower price. Companies use rate fences in an at-
tempt to prevent price-insensitive customers from Plaza meeting rooms 17% 31% 47% 58% 47% 47% 22% 39%

obtaining discounts that are not necessary to en- Raffles Ballroom 0 4% 4% 0 0 11% 4% 3%
courage their business. The rate fences are gen- Stamford Ballroom 28% 31% 25% 17% 28% 17% 44% 27%
erally designed to screen out customers who are Stamford meeting room 19% 15% 33% 33% 41% 39% 26% 30%
not eligible for discounts. Rate-fence rules include
physical attributes of the room (e.g., high tech,
good view), transaction characteristics (e.g., cus-
EXHIBIT 9
tomers who prepay, those who book over a year
ahead of time), customer characteristics (e.g., fre-
quent purchasers, type of organization), or con- PM demand levels
trolled availability (e.g., coupons or direct-mail Facility Sun Mon Tue Wed Thu Fri Sat Total
offers).
The definition of demand levels is subjective Atrium Ballroom 52% 38% 71% 48% 81% 89% 100% 69%

and can be based either on occupancy or ECC: Bailey 33% 38% 50% 56% 67% 33% 33% 44%
ConPAST. For example, at our Singapore prop- ECC: Bras Basah 44% 63% 44% 39% 56% 56% 44% 49%
erties, we designated occupancy rates over 80 per- ECC: Orchard 56% 50% 54% 48% 70% 59% 33% 53%
cent as hot; those between 65 and 80 percent as ECC: Wilberforce 0% 63% 75% 56% 67% 56% 33% 49%
warm; those less than 20 percent as cold; and the
range of 20 to 64 percent as neutral. Alterna- Plaza meeting rooms 28% 53% 72% 69% 83% 81% 61% 64%

tively, demand-level classifications can be based Raffles Ballroom 67% 42% 42% 44% 52% 89% 85% 61%
on ConPAST, but since the contribution from Stamford Ballroom 111% 63% 56% 72% 106% 89% 133% 91%
each function space varies, occupancy may be a Stamford meeting room 22% 23% 42% 41% 44% 43% 39% 36%
better and more comparable measure.

AM Pricing Strategy Key for the two tables above: “hot” “warm” “low”
A key part of the RCCC management’s revenue-
management strategy was to grapple with the
problem presented by the weak AM day part.
Riddled with cold periods, this slow day part had
no hot periods at all and just two warm periods
(see Exhibit 8). The brightest spot was the Bailey
Room, which displayed fairly high demand on
Tuesdays and Thursdays. Given that this is the
only strong point for this day part, the hotel
should quote rack rate for this room. However,
if the room’s high occupancy comes from frequent
customers or standing meetings, those custom-

DECEMBER 2001 Cornell Hotel and Restaurant Administration Quarterly 43

HOTEL OPERATIONS REVENUE MANAGEMENT

ers would likely qualify for a discount. Customers Step 4: Implementation

interested in booking the Bailey Room who are For revenue management to be successful, hotel
extremely price-sensitive should either be steered managers must ensure that the director of sales,
towards another room or another day of the week. sales managers, the director of catering, and the
During cold periods, managers should focus director of food and beverage clearly understand
on filling the rooms by applying any appropriate the purpose and practice of revenue management.
discounts for groups requesting space. In addi- This requires a position-specific training program
tion, sales managers should use the information that can help employees understand their role in
on low-demand periods to develop a marketing revenue management and how revenue manage-
plan for generating business for those periods. ment can benefit both the hotel and employees.
Additionally, managers should align employee-
PM Pricing Strategy incentive and -evaluation programs to coincide
Evenings were much busier at our Singapore con- with the objectives of revenue management.
ference center. This day part enjoyed several hot Sales-incentive systems that reward achieving
periods, most notably in the ballrooms (see Ex- volume or revenue goals can lead to suboptimal
hibit 9, on the previous page). The Atrium and profitability. When volume quotas are used, sales
Stamford Ballrooms, for example, were hot on managers may book business regardless of its con-
tribution. On the other hand, if revenue quotas
are the target, sales managers may “pad” expected
We believe that revenue management revenue with unprofitable add-ons. In addition,
volume- and revenue-quota systems often lead
can help improve the profitability of to unproductive competitions between the sales
hotel function space. managers of different market segments. The use
of contribution quotas can help alleviate the
above conflicts.
Some hotels have experimented with the con-
Thursdays, Fridays, and Saturdays, and the Stam- cept of rewarding sales managers for booking
ford Ballroom was hot on Sundays as well. The business during low-demand periods. For ex-
Plaza meeting rooms and the Raffles Ballroom ample, business brought in during cold periods
were hot on Thursdays and Fridays. The hotel may count for three times as much as the busi-
should be quoting only rack rate for those rooms ness brought in during hot times, while business
during the hot periods. If customers are price brought in during warm periods may count for
sensitive and unwilling to pay full price, they twice as much as hot-period business.
should either be directed to another night of the
week, to the AM day part, or to a “cooler” func- Step 5: System Evaluation
tion room. As with any business practice, the hotel must ap-
This analysis can highlight not only when to ply a reliable measure for the performance of the
offer discounts, but when to boost rack rates. For revenue-management strategy. The success of a
example, we suggest that the extremely high oc- revenue-management system can be measured in
cupancy of the Stamford Ballroom (sometimes three ways: ConPAST, the yield index, and the
even over 100 percent, indicating two turns dur- revenue-opportunity model. ConPAST can be
ing the day part) may indicate that the rates be- monitored for growth, but may not be a reliable
ing charged for the hot periods are too low. Simi- measure because of the complications arising
larly, the Atrium Ballroom recorded a relatively from external market conditions. The yield in-
low ConPAST, despite its high occupancy rates dex and the revenue-opportunity model can re-
on Thursdays through Saturdays. Again, this out- duce the effects of market conditions on the
come may indicate that the hotel is charging rates model and give a clearer indication of perfor-
that are too low. Even though the Atrium Ball- mance.
room is due for renovation, it remains extremely North American hotels have long used the
busy. yield index to measure the success of their rev-

44 Cornell Hotel and Restaurant Administration Quarterly DECEMBER 2001

 REVENUE MANAGEMENT HOTEL OPERATIONS

enue management of guest rooms, and could ment can help improve the profitability of hotel
apply the same concept to function space. As function space. In this paper, we attempted to
described previously, the yield index is simply the
demonstrate how that might occur by proposing
ConPAST for an individual hotel divided by the a five-step approach for developing a function-
ConPAST of its competitive set. As long as a space revenue-management strategy and apply-
hotel’s ConPAST remains above one, it is out- ing that approach to the Raffles City Conven-
performing its competition, regardless of market tion Center in Singapore. As function-space
conditions. Competitive information is becom- revenue management matures, more sophisti-
ing available in many North American cities cated forecasting and optimization approaches
through the HotelFlash banquet reports available similar to those currently used for guest rooms
from HotelRevMax (www.hotelrevmax.com). will be developed, and hotel operators will be-
The reports provide information on banquet- come accustomed to measuring the occupancy
related food, beverage, and room rental per avail-
and ConPAST of function space, just as they now
able room, per occupied room, per group room, refer to the occupancy and RevPAR of guest
and per square foot for the subject hotel and forrooms.
its competitive set. However, hotels located in Hotels that wish to develop a function-space
cities not covered by reports of that kind may revenue-management program should follow
find accurate competitive data difficult to obtain.
these five steps:
For a hotel to use the revenue-opportunity (1) Establish a baseline. Hotels must first de-
model, it must track the timing of both con- termine the baseline performance of their
strained and unconstrained demand. The function space. Baseline performance in-
revenue-opportunity model measures the follow- cludes information on demand patterns,
ing three variables: (1) the amount of contribu- room-occupancy percentages, average
tion that would have been achieved if requests rent per available square foot per day, and
for space were accepted on a first-come-first- contribution per available square foot per
served basis, (2) the amount of contribution that day part.
could have been achieved if perfect information (2) Understand the causes. After establishing
were available, and (3) the amount of contribu- the baseline performance, hotels must de-
tion that was actually obtained.12 The difference termine the possible causes of the perfor-
between the first-come-first-served revenue and mance. Understanding the causes of per-
the perfect revenue is referred to as the revenue formance is important because it can help
opportunity. The performance of a revenue- managers determine why certain things
management system is measured as a percentage have occurred.
of that ideal revenue opportunity that the hotel (3) Develop a strategy. We demonstrated a
actually obtained. For example, assume a hotel simple way to identify peak and non-peak
would have made $20,000 if it had accepted func- days and day parts. Once that is in place,
tion-space requests on a first-come-first-served a hotel’s managers can develop appropri-
basis, but would have made $25,000 if it had ate pricing strategies for each day part. In
perfect information. Its revenue opportunity in addition, the hotel can consider impos-
this instance would be $5,000. If the revenue- ing length-of-stay controls for certain
management system led to a contribution of times, and determining the pricing and
$22,000, the hotel would have earned 40 per- reasoning for the pricing by time of year,
cent of its revenue opportunity ($2,000/$5,000). day of week, and time of day.
(4) Implement the strategy. Implementation is
A Five-step Approach the most difficult part of revenue man-
Function-space revenue-management is still in agement. With function-space revenue
its infancy, but we believe that revenue manage- management, managers must ensure that
sufficient training of appropriate person-
nel is completed, that incentive and re-
12
Smith et al., pp. 8–31. ward systems are aligned to match the

DECEMBER 2001 Cornell Hotel and Restaurant Administration Quarterly 45

HOTEL OPERATIONS REVENUE MANAGEMENT

goals of revenue management, and that

information systems are positioned to
provide the data necessary for decision
making and performance measurement.
(5) Monitor success. Once revenue manage-
ment has been implemented, hotels must
monitor performance and compare it to
the baseline performance. Several perfor-
mance measurements exist, including
the increase in revenue per available
square foot, increase in yield index, and
the revenue-opportunity gauge.
The intent of this paper was to discuss
function-space revenue management by using the
experience of a major operation in Singapore as
an example to illustrate how a hotel can imple-
ment revenue management. We believe this ar-
ticle is just the beginning of what promises to be
a successful and profitable approach to manag-
ing function space. Future developments will
most likely include the sophisticated forecasting,
optimization, and duration-control methods
commonly found in revenue management for
guest rooms. While more sophisticated methods
hold great promise, a hotel will not be able to
take full advantage of such methods unless it
views its function space as a resource that can be
profitably managed. ■

Sheryl E. Kimes, Ph.D. (above, left), is a professor at the Cornell

University School of Hotel Administration (sek6@cornell.edu).
Kelly A. McGuire (above, right), who holds a Master of Manage-
ment in Hospitality (M.M.H.) degree from Cornell University, is a
consultant in the Hospitality Division at Radiant Systems (kmcguire@
radiantsystems.com). The authors would like to thank Jeannette Ho,
revenue manager of Singapore’s Raffles Plaza and Stamford hotels,
for her support and assistance with this project.

© 2001, Cornell University;

refereed paper: submission received on October 11, 2001;
revisions requested on November 20, 2001; accepted on
December 2, 2001.

46 Cornell Hotel and Restaurant Administration Quarterly DECEMBER 2001