U.S. patent application number 10/292532 was filed with the patent office on 2003-05-15 for method for determining retail unit specific price sensitivities.
Invention is credited to Kiefer, Nicholas M..
Application Number | 20030093313 10/292532 |
Document ID | / |
Family ID | 26967391 |
Filed Date | 2003-05-15 |
United States Patent
Application |
20030093313 |
Kind Code |
A1 |
Kiefer, Nicholas M. |
May 15, 2003 |
Method for determining retail unit specific price sensitivities
Abstract
A method of determining a price sensitivity index for one or
more retail units is based on the relation between profits, sales
or traffic and a fixed weight price index based on information from
individual retail units. Statistical regression and the theory of
the single-product firm is used to analyze the relation between
changes in performance variables and changes in the price index,
leading to a unit-specific index of sensitivity. This information
allows stores to be sorted into those which can see price
aggression, those which cannot, and those which are likely to
respond to promotions.
Inventors: |
Kiefer, Nicholas M.;
(Ithaca, NY) |
Correspondence
Address: |
CLARK & BRODY
Suite 600
1750 K Street, NW
Washington
DC
20006
US
|
Family ID: |
26967391 |
Appl. No.: |
10/292532 |
Filed: |
November 13, 2002 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
60331213 |
Nov 13, 2001 |
|
|
|
Current U.S.
Class: |
705/7.34 ;
705/7.37 |
Current CPC
Class: |
G06Q 30/06 20130101;
G06Q 30/0205 20130101; G06Q 30/0206 20130101; G06Q 10/06375
20130101 |
Class at
Publication: |
705/10 |
International
Class: |
G06F 017/60 |
Claims
What is claimed is:
1. A method of determining a price sensitivity for one or more
retail units comprising: a) creating a fixed weight price index
based on pricing information from each retail unit, wherein the
index varies only when prices vary and the weights are based on an
average menu mix per retail unit over a select period of time; b)
regressing at least one of profits or sales or quantity sold for
the retail unit on the fixed weight price index over a select
period of time, and producing a regression coefficient for the
fixed weight price index, wherein time differences in profits/sales
and time differences in the price index are used as the independent
variables in the regression analysis and the variable regressed is
the time differences in quantity sold and gross profit; and c)
assigning a price sensitivity indicator based on the magnitude of
the regression coefficient, wherein the magnitude of the indicator
reflects the level of price sensitivity of the retail unit as it
relates to the regressed variable.
2. The method of claim 1, wherein each of profits or sales and
quantity sold are regressed, and the indicator for profits/sales
shows how the store compares to an optimum pricing index, and the
indicator for quantity sold shows how sensitive the store is to
price changes.
3. The method of claim 1, wherein price sensitivity indicators for
profits/sales include high, low, and right.
4. The method of claim 1, wherein price sensitivity indicators for
quantity sold include not sensitive, sensitive, and highly
sensitive.
5. The method of claim 1, wherein the time difference is one of a
year to year time difference, a week to an adjacent week time
difference, a day to an adjacent day, or a month to an adjacent
month.
6. The method of claim 1, where the time difference is based on a
year to year time difference.
7. The method of claim 1, wherein a log of the profits or sales or
quantity sold for the retail unit are regressed on a log of the
fixed weight price index.
8. A method of determining a price sensitivity of one or more
retail units comprising: identifying a weighted price index for
each retail unit for a period of time; regressing gross profits of
the retail unit on the weighted price index to determine where the
weighted price index falls with respect to the gross profit
function in order to ascertain a magnitude of price sensitivity
against gross profit for the retail unit.
9. A method of determining a price sensitivity of one or more
retail units comprising: identifying a weighted price index for
each retail unit for a period of time; regressing quantity of items
sold for the retail unit on the weighted price index to determine
where the weighted price index falls with respect to the quantity
sold in order to ascertain a magnitude of price sensitivity against
quantity of items sold for the retail unit.
10. A method of claim 9, further comprising regressing quantity of
items sold for the retail unit on the weighted price index to
determine where the weighted price index falls with respect to the
quantity sold in order to ascertain a magnitude of price
sensitivity against quantity of items sold for the retail unit.
11. The method of claim 9, further comprising assigning a gross
profit indicator to reflect where the weighted price index falls
with respect to the gross profit.
12. The method of claim 9, further comprising assigning a
sensitivity indicator to reflect where the weighted price index
falls with respect to the quantity of items sold.
13. The method of claim 9, wherein the period of time is a year to
year time period, and the regression is based on the year to year
differences in the weighted price index.
14. The method of claim 10, further comprising assigning a gross
profit indicator to reflect where the weighted price index falls
with respect to the gross profit.
15. The method of claim 10, further comprising assigning a
sensitivity indicator to reflect where the weighted price index
falls with respect to the quantity of items sold
16. The method of claim 10, wherein the period of time is a year to
year time period, and the regression is based on the year to year
differences in the weighted price index
17. The method of claim 11, wherein the period of time is a year to
year time period, and the regression is based on the year to year
differences in the weighted price index
18. The method of claim 12, wherein the period of time is a year to
year time period, and the regression is based on the year to year
differences in the weighted price index
Description
[0001] This application claims priority under 35 USC 119(e) based
on provisional patent application No. 60/331,213 filed on Nov. 13,
2001.
FIELD OF THE INVENTION
[0002] The present invention is directed to a method for
determining retail unit specific price sensitivities, and in
particular to a method that directly links a weighted price index
to profits and traffic and further eliminates seasonality effects
by comparing year over year changes.
BACKGROUND ART
[0003] In the prior art, it is common to implement pricing or
promotion strategies for a chain of retail outlets. However, a
problem often arises because one or more local unit managers
complain that the overall pricing or promotion strategy does not
apply to their stores; the "yes, but my store is different"
syndrome. Often times, the local manager's observations are
accurate due to the access to local information and experience that
is typically unavailable to corporate headquarters.
[0004] Consequently, there is a need to develop better techniques
for identifying the price sensitivities of a store or business
unit. The present invention solves this need by providing a method
to permit the identification of the price sensitivities of one or
more stores. With this information, a business owner can determine
whether a particular store can raise prices or is too price
sensitive and should concentrate on promotions rather than raising
prices.
SUMMARY OF THE INVENTION
[0005] It is a first object of the present invention to provide a
method of identifying store price sensitivities for marketing
purposes.
[0006] Another object of the invention is a method of identifying
store price sensitivities that eliminates seasonal effects.
[0007] Still another object of the invention is a method that
enables a store owner to better maximize profits through price
promotions rather than higher prices or vice versa.
[0008] Other objects and advantages of the present invention will
become apparent as a description thereof proceeds.
[0009] The store sensitivity analysis produces summary numbers for
individual units in a chain, allowing classification of units
according to how price sensitive both profits and traffic are
(sales are used instead of profits when sales are available and
profits are not). Two regressions can be used together or
individually to categorize stores into groups reflecting various
pricing status and traffic sensitivity similarities. One is the
gross profit regression and the other is the traffic regression.
These two regressions by themselves return valuable information on
the pricing status and sensitivities of the stores in the system.
Moreover, as the combination of these two regression results is
used to categorize stores into groups that are homogenous, similar
revenue management and profit maximizing strategies may be employed
on each store in the category.
[0010] Stores that are determined to be price sensitive by the
invention in both profits and traffic should exercise care in
raising prices, but opportunities to exploit price promotions may
still exist. Stores that are not price sensitive can be more
aggressive in pricing across the board.
BRIEF DESCRIPTION OF THE DRAWINGS
[0011] Reference is now made to the drawings of the invention
wherein:
[0012] FIG. 1a is a graph comparing gross profit function compared
to price;
[0013] FIG. 1b is a graph comparing quantity of items sold versus
price;
[0014] FIG. 2 is a pie chart showing gross profit sensitivity to
price; and
[0015] FIG. 3 is a pie chart showing traffic sensitivity to
price.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0016] The invention offers significant advantages in the field of
pricing and promotion strategies by being able to identify the
price sensitivity of a retail unit amongst a number of retail
units. Identifying this price sensitivity provides invaluable
information in permitting an owner to better identify which
marketing tools are better suited for that store.
[0017] The inventive method involves a number of steps, the steps
principally analyzing the affects of gross profit and traffic or
customer count on pricing.
[0018] A first step involves developing a database of information
over time for each store in terms of various variables relating to
price sensitivity, e.g., prices, profits, sales, items being sold,
quantity of items, time periods, for determining a fixed weight
price index. The index is used to assist in analyzing gross profits
and traffic for stores and sets of stores. In analyzing gross
profit, a regression analysis is made wherein the analysis delivers
a category measure of pricing status: "Low" indicates that price
increases are likely to lead to improvements in gross profit, the
store can be more aggressive in pricing. "High" indicates that the
store's prices are high and care needs to be taken in considering
future increases, i.e., the store is price sensitive. It is
important to understand that a pricing performance categorization
of "High" for the store does not limit the pricing strategy of that
store from improving profits. On the contrary, it indicates that
increased profits can still be realized, potentially by decreasing
prices or increasing promotional activity using appropriate items.
"Right" indicates that the price level, as measured by the
sensitivity statistic, is about right; judicious price increases
can be made, but perhaps there are opportunities in manipulating
menu mix by pricing policy. "Insufficient evidence" indicates that
the evidence for the store is mixed.
[0019] The second regression analysis focuses on traffic or
customer count. This analysis delivers a category measure of
traffic sensitivity in relation to the price index (how does
quantity vary with the price index.) "Not Sensitive" indicates that
the price increases have no adverse affects upon traffic. A flat or
even estimated positive slope of the line basically indicates no
relationship between traffic and increases in price. "Sensitive" or
"Highly Sensitive" reflects a traffic sensitivity to price
increases that begins to evidence a downturn in traffic when
increased prices are implemented, i.e., a negative slope showing
that when the price index increases, traffic decreases.
[0020] While it is preferred to perform both regression analyses
for a complete picture of store sensitivity, either analysis could
be done alone.
[0021] The following better demonstrates the impact that the
inventive analysis framework could have on store profitability.
Stores in the "Sensitive" traffic category and in the "High" price
performance category are likely to see revenue gains from price
decreases and should be extremely cautious about increases. Those
with "Insensitive" traffic and "Low" price performance are in line
for price increases. To understand how these measures affect store
performance, consider the likely shape of the gross profit function
graphed as a function of price as shown in FIGS. 1a and 1b.
[0022] Referring to FIG. 1a, at prices below unit cost, of course
gross profit is negative (this pricing strategy is easy to rule
out), and at price equal to unit cost, labeled below (1) in FIG.
1a, gross profit is zero. As price increases from (1), gross profit
can be expected to increase, as long as consumers want the product
at all. Now consider what happens as the price becomes high.
Contribution margin increases, but the quantity sold can reasonably
be expected to decline. At some price, the quantity will be zero,
and hence so will gross profit. This is below point (3) in FIG. 1a.
As the two effects of prices operate, a gross profit function of
the general shape given in FIG. 1a can be produced. Maximum gross
profit (2) occurs at price p*. Of course, the seller would like to
choose the price p*, at which gross profit is maximized.
[0023] An important aspect of the inventive store sensitivity
system is the ability to examine data on prices and gross profits
and determine whether stores are operating at or near p*, at prices
below p*, or at prices above p*. Identifying the store's
relationship to p* gives the store owner insight as to what should
be done to improve profits.
[0024] In contrast to FIG. 1a, which focuses on the relationship
between gross profit and price, another aspect of the invention
relates to the relationship between traffic or quantity sold and
price. This relationship can be used to illustrate the development
of the gross profit function from assumptions on consumer demand.
Referring to FIG. 1b, suppose that q represents the quantity sold
of an item. Further suppose that the quantity sold q of the item in
question is a function of price. Suppose further for illustration
that it is a linear function q=a-b*p, as graphed in FIG. 1b. Assume
that the per-unit cost is c (refer to FIG. 1a). Consider the price
p* in FIG. 1b. At that price, the quantity sold is q*, from the
demand function. Revenue realized is p* times q* namely the area of
the indicated rectangles 1 and 2. Cost (food cost) is given by q*
times c, also indicated as the area of rectangle 1 on the graph.
The difference between these areas is exactly gross profit or
rectangle 2. Thus FIG. 1a, the gross profit function, can be
developed from FIG. 1b by considering different prices, reading the
corresponding quantities from the demand function, calculating
revenue and cost and taking the difference for gross profit, and
graphing gross profit against prices. However, this method is quite
tedious and requires a number of steps to arrive at gross profit
based on individual products. In the store sensitivity analysis of
the invention, gross profit is studied directly; not via the
demands for individual products. This is a tremendous
simplification and advantage when dealing with multi-product
situations.
[0025] The stylized case of a firm selling one product in varying
quantities provides a useful framework for focusing ideas, but
implementation in the case of restaurants with full menus or retail
stores with full product lines are different situations entirely.
The store sensitivity approach emphasizes restaurant-level
characteristics, not item-level characteristics.
[0026] In order to develop a single summary measure of pricing
status, it is preferred to develop a single index summarizing the
prices in a particular store. This index of prices can be
calculated for individual stores over many periods, and the
relation between the price index and a measure of gross profit can
be examined on the basis of co-variation between the two
variables.
[0027] When dealing with indices, one question to consider is the
use of weighted averages of the prices of the different menu items
as a summary measure of the prices in a given store in a given
period. If a weighted index is selected, the question then becomes
what weights should be used. One possibility is to weight by menu
mix. In this case, the index is simply the check average defined as
total revenue divided by total items sold ($10.00 in revenue/5
items=2.0). This calculation involves the use of the price of the
items weighted by the number sold. The problem with this approach
is that because menu mix changes from period to period as consumer
purchasing behavior varies, changes in the check average will occur
even when prices have not moved. Put another way, while prices may
stay the same, the number of items may change, thus changing the
price index.
[0028] The present invention avoids this pitfall through the use an
index, which is an indicator of movements of prices within the
store's control. The check average mixes up changes in prices and
changes in quantities sold from period to period and is therefore
not desirable. A fixed-weighted index is preferred since it does
not suffer from the problems of a check average and is more
appropriate for determining price sensitivities. Fixed weight
indices are well known in the statistic art, and a detailed
explanation is not deemed necessary for understanding of the
invention.
[0029] It is preferred to weight the different prices by a measure
of the relative importance of each price in revenue production. The
inventive store sensitivity analysis approach uses a fixed-weight
system in which the weights are the average menu mix per store over
the period considered. This method produces an index which moves
only when prices move, but which still does weight prices according
to their revenue contribution. This technique does not use the
check average approach, which can move even if prices do not.
[0030] Referring back to the regression analysis of gross profit or
quantity sold, a number of store/period specific variables for use
in the analysis include: (1) lnpp, the logarithm of profits (these
can be actual profits, or a measure adjusted for changes in costs,
or if profits are unavailable, sales; (2) lntraffic, the logarithm
of a measure of traffic (either customer counts or number of items
sold); and (3) lnpind, the logarithm of the price index constructed
as described above. Periods can vary such as by day, week or
month.
[0031] In the ideal, full data case, data are available for more
than one year. Having data for more than a year allows new
variables to be formed, i.e., dlnpp, dlntraffic, and dlnpind, the
year over year changes in each of these variables. For example,
dlnpp can represent the difference between profits in week 27 in
the current year and week 27 of the previous year. The regression
coefficient of dlnpind in the regression of dlnpp on dlnpind is the
price sensitivity index (corresponding approximately to the slope
of the function shown in FIG. 1a). The coefficient in the
regression of dlntraffic on dlnpind is the promotion or traffic
sensitivity index. This year over year comparison is a significant
advantage when determining true price sensitivities. By looking at
the difference in gross profit and traffic in the same season but
between two different years, the potential confounding effects of
seasonality are eliminated in the estimate. The regressions are
preferably performed separately for each store if the data permit;
however, importantly, this specification in differences allows
combining information across similar stores to obtain an overall
"market" sensitivity for any commercially interesting group of
stores.
[0032] The regression can also be done without year over year data,
e.g., lnpp on lnpind, by store for a selected period of time.
[0033] Once the regression coefficients are generated, a summary
report for the chain as a whole can be developed which is of
significant importance in determining the price and traffic
sensitivity for all stores. An example in terms of a restaurant is
shown below. While not shown, a similar report could be which would
show a listing of the particular results by store, e.g., what
stores are highly sensitive, not sensitive, etc. in traffic and
which stores are high, low, or right in terms of gross profit.
[0034] The FIGS. 2 and 3 summarize the chain's gross profit and
traffic sensitivity for US restaurants only.
[0035] FIG. 2 represents gross profits and illustrates that 47% of
all US stores in this example have a "Low" gross profit
sensitivity. This indicates that these stores are performing below
the optimal gross profit point and there are significant profit
opportunities remaining within these stores. Eighteen percent of
the stores are operating at the right gross profit point, and 25%
are operating beyond the optimal gross profit point. There was
insufficient evidence to determine the sensitivity ratings for 10%
of the stores.
[0036] Referring to FIG. 3, stores characterized by "Not Sensitive"
to price do not drive traffic through price promotions, while
stores that are "Highly Sensitive" to price can improve traffic
with price promotions on items. Stores characterized by "Low" gross
profit sensitivity and "Low" traffic sensitivity have an
opportunity for increased margins by increasing prices on the
proper items. The second group of stores, evidencing "High" gross
profit sensitivity and "High" traffic sensitivity must exercise
caution when implementing price changes and may do better with
price promotions.
[0037] As noted above, a final part of the report is the list of
stores and their categorizations. Stores with "insufficient
evidence" simply do not have enough data variation to identify
sensitivities (i.e. the regression t-statistics are <1.5 in
absolute value; this number can be varied according to the level of
confidence required). Sensitive stores have significantly negative
coefficients, and insensitive stores have zero or positive
coefficients.
[0038] While the invention is described in terms of gross profits,
this measure is not always available. In these instances, sales can
be substituted for profits.
[0039] While the example uses variables based on the difference in
year to year, other time periods could be used such as week to
adjacent week, month to adjacent month, day to adjacent day, year
to adjacent year, etc.
[0040] As such, an invention has been disclosed in terms of
preferred embodiments thereof which fulfills each and every one of
the objects of the present invention as set forth above and
provides new and improved method for determining price
sensitivities for retail units.
[0041] Of course, various changes, modifications and alterations
from the teachings of the present invention may be contemplated by
those skilled in the art without departing from the intended spirit
and scope thereof. It is intended that the present invention only
be limited by the terms of the appended claims.
* * * * *