Implementing

Restaurant
Revenue Management
A Five-step Approach Revenue management for restaurants hinges on an appropriate measure

of revenue, Here’s the basis of one such measure— revenue per

available seat-hour.
by Sheryl E. Kimes

goal of restaurant revenue

management (RRM) is to maxi­
mize revenue per available seat-hour
by manipulating price and meal
duration. An earlier Cornell Quarterly
article discussed the theory behind
RRM ,1but the application of
RRM has not been explored. In a
series of articles on the application
of RRM, of which this is the first, I
will discuss how to develop a RRM
system, demonstrate how RRM was
1 Sheryl E. Kimes, Richard B. Chase, Sunmee
Choi, Elizabeth N. Ngonzi, and Philip Y. Lee,
“Restaurant Revenue Management,” Cornell
Hotel and Restaurant Administration Quarterly,
Vol. 39, No. 3 (June 1998), pp. 32-39.
Sheryl E. Kimes, Ph.D., is an
associate professor at the Cornell
University School of Hotel Administra­
tion «[email protected]».
© 1999, Cornell University

t6 HOTEL AND RESTAURANT ADMINISTRATION QUARTERLY

 FOCUS ON FOOD S E R VI CB

applied to a pilot restaurant, and evaluated. Many managers currently an amount of time—although one
develop guidelines that restaurant measure their restaurant’s success by could argue that the true measure of
operators can use to increase their tallying the average check or by the restaurant’s product is time.4
revenue per available seat hour. maintaining certain labor- and While one can estimate a likely
In this paper I discuss the funda­ food-cost percentages. While such mean length for that meal, the ac­
mentals of restaurant revenue man­ measures are valuable for many pur­ tual duration is not set. By compari­
agement, examine methods of mea­ poses, they do not explicitly reflect son, implementing revenue manage­
suring revenue-management success, a restaurant’s revenue- (or profit-) ment is much easier for the hotel,
explain how those measures are producing performance. RevPASH, airline, cruise-line, and car-rental
different from the traditional gauges on the other hand, combines infor­ businesses, because they sell their
of restaurants’ success, and outline a mation from the average check and service for an explicitly contracted
five-step approach to establishing a seat use (or occupancy) to provide a amount of time. Restaurants rarely
revenue-management system. measure of the flow of revenue sell tables for a fixed amount of
through the system and to indicate time, and in most western cultures
Overview of Restaurant Revenue how effectively a restaurant is using are reluctant to broach this topic
Management its productive capacity. with customers. Moreover, North
Restaurant revenue management Restaurant operators have two American restaurateurs cannot even
can be defined as selling the right main strategic levers that they can rely on the practice common in
seat to the right customer at the use to manage revenue: namely, some countries of charging for the
right price and for the right dura­ price and meal duration.2 Price is a cover.
tion. The determination of “right” fairly obvious target for manipula­ One of the stumbling blocks to
entails achieving both the most rev­ tion, and many operators already successful implementation of restau­
enue possible for the restaurant and offer price-related promotions to rant revenue management is the
also delivering the greatest value or augment or shift peak-period de­ struggle that restaurant operators
utility to the customer. Without that mand (e.g., early bird specials, spe­ have in developing internal methods
balance, RM-type practices will in cial menu promotions). More so­ of managing meal duration. In the
the long term alienate those cus­ phisticated manipulations of price context of managing meal duration,
tomers who will feel that the restau­ include day-part pricing, day-of- one should not think only of reduc­
rant has taken advantage of them. week pricing, and price premiums ing diners’ average meal length.
Revenue management, or yield or discounts for different types of Quite often the factor interfering
management, is commonly practiced party size, tables, and customers. with revenue management is the
in the hotel and airline industries. Managing meal duration is a bit variability in meal lengths, and not
Companies implementing revenue more complicated. On the produc­ just their duration. Some of the
management report increases in tion side, managers must streamline ways in which managers can influ­
revenue of 2 to 5 percent over the and control their service-delivery ence meal duration include chang­
results of prior procedures. Revenue process, as well as understand ing reservation policies, redesigning
management requires a focus on the customer-arrival patterns and de­ menus, and pacing service proce­
revenue per available inventory unit. termine ways of influencing meal dures and making them more effi­
For example, hotels measure revenue duration.3 One of the things that cient. Managers can also train em­
per available room-night (com­ makes implementing revenue man­ ployees to respond to customers’
monly referred to as RevPAR), agement so difficult in restaurants is apparent wishes regarding the length
airlines measure revenue per avail­ the fact that their explicit unit of of the meal. While some customers
able seat-mile (RPSM), and cruise sale is a meal (or event) rather than may wish to linger over coffee, for
lines measure revenue per available instance, managers might be sur­
cabin. When restaurant operators 2Kimes et al, June 1998; see also: Sheryl E. prised at how often the holdup in
apply revenue management to their Kimes and Richard B. Chase,“The Strategic
Levers of Yield Management,’"Journal of Service turning a table is the restaurant’s
restaurants, I recommend that they Research, Vol. l,N o. 2 (Nov. 1998),pp. 156—166. own lackadaisical approach to ser­
measure their results in terms of 3For a discussion of the effects of altering the vice and servers’ inattention to cus­
revenue per available seat hour service-cycle time, see: Christopher C. Muller,
“A Simple Measure of Restaurant Efficiency,” tomers’ needs.
(RevPASH). on pages 31-37 of this Cornell Quarterly; for an
Concentrating on RevPASH has analysis of arrival times, see: Brian Sill and Rob­
ert Decker, “Applying Capacity-management 4 An argument made by: Daryl Ansel and Chris
major implications for the way in Science: The Case of Browns Restaurants,” Dyer, “A Framework for Restaurant Information
which a restaurant is operated and on pages 22-30 of this Cornell Quarterly. Technology,” on pages 74—84 of this issue.

June 1999 • i?
 Measuring Success
Exhibit 1 Restaurant managers are typically
Various calculations of RevPASH evaluated by the check averages and
by the food- and labor-cost per­
Reetaurant Capacity use Average check RevPASH centages that they have been able
A 40% $18.00 $7.20 to maintain. As I indicated before,
B 60% $12.00 $7.20 none of those measures captures
C 80% $9.00 $7.20 sufficient information about the
D 90% $8.00 $7.20 revenue- (or profit-) generating
performance of the restaurant. Some
measure of the revenue-generating
potential and performance of the
Exhibit 2 restaurant must be developed.
Revenue effects of meal-duration reduction For a restaurant manager to con­
centrate only on a high average
Meal duration Percentage Cumulative check, for instance, is equivalent to
(minutes) Turns Revenue RevPASH increase increase a hotel’s focusing solely on a high
60 4.00 $ 6,000 $15.00 average room rate.5 Without infor­
59 4.07 $6,102 $15.25 1.69% 1.6 mation on the percentage of capac­
58 4.14 $6,207 $15.52 1.72% 3.45% ity use or occupancy of the restau­
rant, revenue performance cannot
56 4.29 $6,429 $16.07 1.79% 7.14% be evaluated. A high average check
55 4.36 $6,545 $16.36 1.82% 9.09% may even be detrimental in times of
54 4.44 $6,667 $16.67 1.85% 11.11% strong demand if, for example, cus­
tomers linger over their meal while
52 4.62 $6,923 $17.31 1.92% 15.38% other parties wait for a table.
51 4.71 $7,059 $17.65 1.96%
2 .00%
17.65%
20.00%
Similarly, a manager’s achieving
50 4.80 $7,200 $18.00
specified food-cost and labor-cost
percentages is laudable, but that does
48
47
5.00
5.11
$7,500 $18.75 2.08% 25.00%
not tell the entire story. In particular,
46 5.22
$7,660
$7,826
$19.15
$19.57
2.13%
17.00%
27.66%
30.43%
the margin is not a measure of prof­
itable use of capacity. A restaurant
44 5.45 $8,182 $20.45 2.27% 36.36%
manager can do a good job of
43 5.58 $8,372 $20.93 2.33% 39.53% maintaining margins and still be
42 5.71 $8,571 $21.43 2.38% 42.86% unprofitable. An overemphasis on
margins can lead to a propensity to
50.00% focus unduly on minimizing costs.
Again, reducing cost is fine, but not
when that causes reduced revenue
due to disgruntled customers.
The extent to which available
seats are occupied is another com­
monly applied measure of success,
since a busy restaurant is generally a
revenue-producing restaurant. Reli­
ance on seat occupancy as a measure
of success suffers from the same
problem as reliance on hotel-room
occupancy (in the absence of con­
sideration of ADR), because high
use does not necessarily mean high
5 See: Kimes et al. (June 1998), pp. 32—39.

18 j HOTEL AND RESTAURANT ADMINISTRATION QUARTERLY

revenue. A restaurant can run at
90-percent of capacity and still not Exhibit 3
make money if menu items are sold
at too low a price, for example.
Because it embraces capacity use, 5:00 pm 6 :0 0 pm 7:00 pm 8 :0 0 pm 9:00 pm

check averages, and cost margins, Sunday $2.39 $6.72 $6.43 $6.36 $3.63
revenue per available seat-hour Monday $0.19 $2.91 $2.92 $3.52 $2.95
(RevPASH) is a much better indica­ Tuesday $0.61 $2.96 $5.46 $4.61 $5.47
tor of the revenue generating per­ Wednesday $0.75 $2.70 $3.92 $4.29 $2.26
formance of a restaurant than the Thursday $0.22 $1.47 $4.86 $3.37 $2.84
commonly used measures that I just Friday $1.49 $6.04 $8.76 $8.17 $9.21
discussed. RevPASH indicates the Saturday $2.72 $6.22 $11.89 $12.60 $11.59
rate at which revenue is generated
and captures the trade-off between
average check and facility use. If RevPASH is closely related to may require substantial investment—
occupancy percentages increase the number of turns and the length for example, by adding kitchen
even as the average check decreases, of the meal, or service cycle. As the equipment or more employees. A
for instance, a restaurant can still number of turns increases and meal return-on-investment analysis that
achieve the same RevPASH. Con­ length decreases, the RevPASH considers the effects of service-cycle
versely, if a restaurant can increase increases. Just a one-minute reduc­ changes on RevPASH can help op­
the average check, it can maintain a tion in meal time during a high- erators decide whether a prospective
similar RevPASH with a slightly demand period can lead to an in­ investment is worthwhile. In making
lower facility use. crease in RevPASH of 1.5 to 2.0 their plans managers should remem­
Exhibit 1 gives a hypothetical percent. Returning to our hypo­ ber that customer preferences and
illustration of this principle. The thetical 100-seat restaurant with its expectations limit the minimum
four restaurants in the exhibit all four-hour dinner, say that its aver­ feasible meal duration and will set
have the same RevPASH ($7.20), age service cycle is 60 minutes. In a theoretical minimum acceptable
but each achieves it in a different that case, the restaurant can poten­ RevPASH.
manner. Restaurant A has a facility tially handle 400 customers per
use of 40 percent and an average night. If the average check is $15, RevPASH-based Strategies
check of $18.00, while Restaurant its maximum nightly revenue is Once operators understand their
D has a use ratio of 90 percent but $6,000, and its potential RevPASH RevPASH patterns, they can de­
an average check of $8.00. Restau­ is $15. If the meal time can be re­ velop strategies for dealing with
rants B and C also achieved a duced to 59 minutes, the restaurant high and low RevPASH periods. A
RevPASH of $7.20, but with vary­ can handle an additional 6.8 cus­ full discussion of the RRM strate­
ing facility-use and average-check tomers. If the average check re­ gies available will be presented in a
statistics. mains at $15, its potential nightly subsequent paper of this series. Dur­
The easiest way to calculate revenue increases to $6,102 and ing low RevPASH periods, manag­
RevPASH is to divide revenue (or its potential RevPASH increases ers can either try to attract more
profit) for the desired time period to $15.26 (a 1.7-percent increase). customers and increase use or rely
(e.g., day part, day, month) by the With the increased volume, the on suggestive selling to increase the
number of seat-hours available dur­ check average could even drop average check. During those periods
ing that interval. For example, as­ by $0.20, and revenue would still with high RevPASH, operators
sume a 100-seat restaurant makes increase. should consider raising menu prices
$1,500 on Fridays between 6:00 and Reduced meal times can be or try to reduce meal duration so
7:00 p m . Its RevPASH would be achieved by changing the service that the restaurant can increase its
$15 ($1,500 -5- 100 seats X 1 hour). process, altering staffing levels, or turn rate.
Similarly, if that same 100-seat res­ altering the menu. The first few RevPASH can be used at differ­
taurant made $5,000 over a four- minutes of reduction are not that ent levels of analysis and for different
hour meal period, its RevPASH difficult or expensive to achieve, for purposes. At the individual restau­
would be $12.50 ($5,000 - 100 example, by picking up the pace of rant level, managers may choose to
seats X 4 hours, or $5,000 -s- 400 greeting, seating, and check settle­ develop hourly or quarter-hourly
available seat-hours). ment. Deep reductions, however, RevPASH figures to help develop a
 selling or even reduce menu prices
Exhibit 4 | to boost traffic.
>• t ■ - -u - > m i RevPASH oertormance j Comparative RevPASH
Average Check RevPASH Regional or national managers
Restaurant check RevPASH performance performance could use RevPASH to compare
1 $10.50 $6.45 0.99 1.24 performance of different restaurant
2 9.75 4.50 0.92 0.87 units. One would want to adjust the
3 11.25 5.25 1.06 1.01 unit RevPASH according to an
4 12.10 4.25 1.14 0.82 area’s cost of living, but a unit-by­
5 9.45 6.25 0.89 1.20 unit comparison of RevPASH
6 10.60 4.50 1.00 0.87 would give a good indication of the
relative performance of different
restaurants in an area, region, or
nation.
Consider a city with six restau­
Exhibit 5
rants franchised by a particular
Con-.ut
chain, as shown in Exhibit 4. Res­
f ■•. ••
taurant 4 enjoyed the highest aver­
Unit Average check RevPASH Check ratio RevPASH ratio age check ($12.10) of the six units,
but it also has the lowest RevPASH
Your restaurant $12.50 $5.60 0.89 1.12
($4.25). On the other hand, restau­
Competitive set (mean) $14.00 $5.00 — —
rant 5 has the lowest average check
Note: The two ratios compare the subject restaurant’s performance to its competitive set. ($9.45), but the second-highest
RevPASH ($6.25). Relative perfor­
mance measurements can be calcu­
lated by dividing one restaurant’s
revenue-management strategy best performance by the average perfor­
suited to their restaurant. RevPASH mance of all the units. Thus, the
can also be used to evaluate the check performance of restaurant
effectiveness of an operation’s serv­ 4 is above average (1.14), but its
ers and managers. RevPASH performance is only 0.83
Say that the manager of our hy­ ($4.25 + $5.20). By using relative-
pothetical 100-seat restaurant performance measurements like
wanted to understand her hourly these, regional managers can better
RevPASH patterns for September. evaluate the revenue generation of
Obtaining data from the restaurant’s the restaurants they oversee.
POS system, the manager found One could also calculate and
that her highest RevPASH periods compare competitors’ RevPASH to
were on Fridays and Saturdays from get a sense of how well a restaurant,
6:00 to 10:00 p m and on Sundays a region, or a chain is performing.
from 6:00 to 9:00 p m . The manager For example, the RevPASH of a
can use this information to help particular restaurant could be com­
develop revenue-management tac­ pared with the average RevPASH of
tics specific to high and low the competitive set to evaluate per­
RevPASH periods. For example, formance (see Exhibit 5). This type
during high-RevPASH periods, she of calculation is done for hotels, for
may focus on reducing the meal example, by Smith Travel Research
time by having table servers skip reports.6 While competitive infor­
suggestions of desserts or after­ mation is not readily available, one
dinner drinks. On the other hand, at
low-RevPASH times she may de­ 6 See: «http://str-online.com/products/
cide to increase the use of suggestive star.html».

20 H IK \ ¡ HOTEL AND RESTAURANT ADMINISTRATION QUARTERLY

may be able to develop information grams9 can be used to better RRM cannot be ensured
to perform such an analysis. understand the possible reasons without measurement of im­
for why meals last as long as provement. After establishing
The Five Steps to RRM they do and to help identify the the baseline and implementing
When developing an RRM system a most important problems in RRM, operators must develop
restaurant operator must first under­ controlling meal duration. a system to measure RRM
stand current conditions and perfor­ (3) Make recommendations. After performance. One should, for
mance. Following this, the operator identifying the causes of the instance, monitor RevPASH
must evaluate the possible drivers of most important problems af­ and the average and standard
that performance. This understand­ fecting the service cycle, man­ deviation of dining time and
ing will help managers determine agers should develop detailed compare those figures to
how to improve RevPASH statistics. recommendations on how to baseline performance.
Finally, the manager must monitor correct those problems. Some
the impact of implemented changes solutions may deal with reduc­ Future Papers
on revenue performance. I describe ing the overall meal duration, This article is the first of a series
each of these steps below. while others may deal with on restaurant revenue management.
(1) Establish the baseline. Most man­ reducing variability in particu­ In future articles I will discuss spe­
agers know their average check lar service steps (e.g., order­ cifically how to develop an RRM
and their labor- and food-cost taking, bussing), and still others system and establish a baseline—
percentages, but few can accu­ may involve table management including the types of data to
rately estimate the capacity use or customer-arrival manage­ gather, the possible sources of data,
or RevPASH of their restau­ ment. The manager should and the analysis and interpretation
rants. To develop a RRM pro­ analyze potential return on of the information collected. These
gram, operators must collect investment for each recommen­ and the other restaurant revenue
detailed information on arrival dation to ensure prudent deci­ management points will be illus­
patterns, meal times, RevPASH sion making. A later article trated with the actual experience
patterns, and customer prefer­ will present a simple model of a 100-seat restaurant in Ithaca,
ences. This information can be for evaluating the ROI of New York.
collected from a variety of RevPASH improvements. The series will review methods
sources, including the POS (4) Implement the changes. For RRM for analyzing the possible causes of
system, guest checks, and me­ to be successful, restaurant op­ problems, including service blue­
thodical observation. Once erators must ensure that manag­ prints, fishbone diagrams, and pro­
collected, the data must be ers, servers, bussers, and other cess analysis, and it will examine
analyzed to determine the employees clearly comprehend restaurant revenue-management
mean and deviation of dining the purpose and practice of strategy and tactics. I will also dis­
time and daily and hourly RRM. This requires a position- cuss how to combine the informa­
RevPASH patterns. specific training program that tion obtained from the baseline and
(2) Understand the drivers. Once helps employees understand causal analysis to develop specific
the baseline data have been their role in RRM and how strategies for increasing RevPASH.
collected, managers should RRM can benefit both the Finally, the papers will examine
analyze the factors that affect restaurant and employees. Addi­ issues surrounding RRM imple­
meal duration and RevPASH tionally, operators should align mentation and discuss methods of
performance. Simple tools such any employee-incentive pro­ establishing appropriate incentive
as process analysis,7 service grams to coincide with the and training programs for servers,
blueprints,8 and fishbone dia­ objectives of RRM. managers, and bussers, along with
(5) Monitor outcomes. As with any suggestions on how to monitor the
business practice, the success of success of the revenue-management
7 For example, see: Sheryl E. Kimes and
Stephen A. Mutkoski, “The Express Guest
system. By the end of the series,
Check: Savings Steps with Process Design,” 9D. Daryl WyckofF,“New Tools for Achieving I hope to have drawn an outline
Cornell Hotel and Restaurant Administration Quar­ Service Quality,” Cornell Hotel and Restaurant of how to implement revenue
terly, Vol. 30, No. 2 (August 1989), pp. 21-25.
8 G. Lynn Shostack, “Designing Services That
Administration Quarterly, Vol. 25, No. 3 (November
1984), pp. 78-91; and U. Apte and C. Reynolds,
management in your restaurant and
Deliver,” Harvard Business Review, January- “Quality Management at Kentucky Fried thereby improve its revenue­
February 1984, pp. 133-139. Chicken,” Interfaces, Vol. 25, No. 3 (1995), pp. 6-21. generating performance.