James H.

Myers

Benefit Structure Analysis:

A New Tool for Product Planning

HE 196O's saw the introduction of several 4. The technique also provides relatively com-
T technologies/models/processes for providing
structure for consumer markets. Most of these in-
plete information as to ambient conditions
surrounding the use of the product (time of
volved the application of multivariate statistical day, other persons present, use or task, etc.).
techniques to problems of product positioning,
market segmentation, the new product develop- A benefit structure study also provides a com-
ment process, and determining the order and plete cross-sectional view of current usage pat-
structure underlying sets of marketing variables. tems (i.e., what products/services are now used
A recent article by Green reviews and critiques for each purpose or objective) within the broad
many of these efforts.' product/service category selected. For most prod-
This article presents another approach to struc- ucts/services this information is not available
turing consumer markets: Benefit Structure Anal- elsewhere, yet it can be obtained easily from data
ysis. It differs from other methods in several collected for a Benefit Structure Analysis study.
important respects: To illustrate this technique a study of house-
hold cleaning products is presented. This study is
1. Il was developed especially for finding new hypothetical, but the data shown are actual re-
product opportunities in very broad pixjd- sults from a composite of three separate commer-
uct/service categories, such as entirely new cial studies conducted by the author in broad
types of food or beverages, new forms of product areas involving foods, beverages, and rec-
banking services {e.g., the bank credit card), reational vehicles.
or new multi-purpose tools for home repair
or redecoration. Benefit Structure Study for Household
2. It determines consumer reactions to a large Cleaning Products
number (75-100) of relatively specific benefits
Household cleaning products were defined as
desired from a type of product/service and to
any product used to clean surfaces within the
many (50-75) features or physical characteris-
home other than ixigs, drapes, furniture, or to be
tics of the product/service.
used for dusting. Examples of products used for
3. These reactions are in terms of both desire these purposes are Ajax, Lysol, 409, Ammonia, Glo
for and perceived deficiencies in each benefit Coat, Windex, etc.
and characteristic.
Qualitative Phase

• ABOUT THE AUTHOR The study began with 25-50 in-depth interviews
James H. Myers is DeBell professor of business (for some studies focus groups would work as
administration in the School of Business Adminis- well). The interviewer asks the respondent to re-
tration of UCLA call all occasions when she cleaned any interior
23
24 Journal of Marketing, October 1976

surfaces during the day prior to the interview. For Note that this random cross-section of one clean-
each of these occasions she is asked: ing occasion should represent all U.S. household
1. What was the cleaning chore? (e.g., sink, floor, cleaning occasions (within usual survey lim-
walls, bathtub, toilet bowl, shower stall, itations) in proportion to the frequency with which
appliances, cabinet facings) they occur. (Care must be taken in survey timing,
since pattems may differ over different days of
2. What product(s) were used in this operation?
the week and for the time of year—e.g.. Spring
(type and brand)
cleaning.)
3. What benefits were sought, or what were the For this one occasion respondents were asked
objectives of this cleaning? (see Exhibit 1 for the degree to which each of the 75-100 benefits
examples of benefits) was received; then she was asked whether she
4. What were the physical characteristics or attri- wanted that benefit any more or any less than she
butes of the product(s) used? (see Exhibit 1 for actually got it. For these measurements a four-
examples of product characteristics) point scale was used: 4 = a whole lot, 3 = pretty
5. What applicator (if any) was used? (e.g., mop, much, 2 = somewhat, 1 = not at all. The verbal
brush, sponge, rag) descriptors for this scale were selected based on a
6. What time of day was the work done; were pre-test using the Thurstone Equal-Interval tech-
other family members involved, etc.? nique. For example, a respondent might report
In this study it was also necessary to ask about that when cleaning her linoleum floor yesterday,
other cleaning tasks done during the past week. she wanted the benefit, "cleans without scrub-
Since many items are cleaned only on an infre- bing," "a whole lot" (scale value = 4), but that she
quent basis by housekeepers (e.g., windows, ap- received it only "somewhat" (scale value = 2),
pliances, broiler pans, etc.), they might be missed resulting in a benefit deficiency of 4 minus 2, or^-
in a discussion of cleaning chores undertaken only Similarly, respondents were asked how much
the day before. In studies of food or beverage they wanted and how much they got of each of
products consumed frequently, it is usually not the 50-75 product characteristics. Finally, various
necessary to go back beyond the previous day. types of supporting information were collected,
It may even be possible to omit the qualitative such as: (1) what cleaning product(s) were used (if
phase of the study entirely if a company already any) and brand; (2) applicator used; (3) at what
has extensive information about the product/ time of day; (4) other persons present; (5) ac-
service area from previous research. tivities both before and after the cleaning process;
(6) how often this item is normally cleaned.
The various cleaning occasions, products, bene-
fits, attributes, etc. derived from the interviews
were used to design a large scale study to deter- The resulting data comprised a multi-dimen-
mine the structural relationships among these sional matrix: Benefits Wanted x Benefits Received
elements in quantitative terms. X Product Characteristics Wanted x Product
Quantitative Phase—Methodoiogy
Characteristics Received x Cleaning Chores x
Product(s) Used x Brands x Applicators Used x
The quantitative phase of the study determined Various Supporting Data. Each dimension had
the degree to which each benefit and each product from perhaps five to as many as 100 categories.
characteristic was desired by the consumer for This n-dimensional matrix is called the Complete
each usage (cleaning occasion) and the extent to Benefit Matrix. It comprises the data bank for sev-
which they were or were not being received. eral simple and cross-tabulations, as well as for a
A nation-wide survey of 500 housekeepers was wide variety of multivariate analyses. Only a few
conducted in 12 metropolitan areas. The home of the more useful types of analysis can be shown
interview began by questioning the woman about below; space limitations preclude showing the
all the cleaning occasions she encountered during full spectrum of possible analytic approaches.
the previous day. Going through the entire day,
she told what cleaning chores she did and what Primary Analytic Paradigm
products were used for each occasion. All analyses were done on an aggregate basis,
The interviewer then selected a single cleaning with responses pooled across all survey respon-
occasion from all that were mentioned by the re- dents or across some selected sub-group of re-
spondent, using a rotational sampling pattern. spondents. Exhibit 2 shows schematically the
This one occasion became the focus of a much principal types of analyses and relationships that
more intensive probe about product(s) used, ben- are normally most useful for product planning
efits/characteristics desired and received, etc. purposes. Some of these are presented below, to
Benefit Structure Analysis 25

EXHIBIT 1
EXAMPLES OF BENEFITS
Bleaches Chrome sparkles
Removes stains Doesn't dull
Removes grease Doesn't hurt hands
Removes built-up dirt Dissolves grease
Cleans tub ring Doesn't remove gloss from paint
Less elbow grease Boosts detergents
Can see it work Strips wax
Cleans cracks (grout) better Less build-up
Doesn't leave residue Lets color come through
No rinsing necessary Stands up to damp mopping
Doesn't damage surfaces Seals porous floors
Kills mildew Doesn't yellow
Disinfects No streaking
Removes discoloration Does two jobs at once
Removes soap scum Leaves it "squeaky clean"
EXAMPLES OF PRODUCT CHARACTERISTICS
Strong smell Biodegradable
Abrasive/scratchy Concentrated
Thin liquid Self-polishing
Low suds Can spray on
Quick drying Attractive color
Can wipe on Contains deodorant
Dark color Economical
Caustic Pine smell
Contains wax Perfumed smell
Contains ammonia Lemon smell
Thick liquid Stains
Light color Little odor
Contains antiseptic No deterioration when stored

illustrate procedures and suggested applications. The Product-by-Use matrix shows one form of
Each analysis presented is indicated by the small market structure: what products are used for each
numbers in Exhibit 2. Several different tech- cleaning chore, and what cleaning chores are seen
niques can be utilized to establish relationships, as appropriate for each product type. These rela-
depending upon objectives; they include cross- tionships show the market planner how products
tabulations, graphic plots, simple correlations, currently on the market are being used; they
factor analysis, discriminant analysis, canonical show what products are used as substitutes or as
analysis, and simultaneous row-eolumn cluster- complements to one another; they suggest how
ing. I single-use products might be altered to make
them into multi-use products; they reveal clean-
Products Used By Cleaning Task ing tasks where now only a single product is con-
One of the most basic analyses is the Product- sidered appropriate, so that competitive entries
by-Use matrix, with product types as rows and might be considered. Perhaps most important,
cleaning chores as columns. (See relationship #1 they show the relative frequency of the various
in Exhibit 2; space Hmitations preclude showing cleaning tasks; if "amount used" is requested,
the matrix itself.) This is quite similar to the then volumetric analysis is also possible.
Stefflre Item-Use matrix;^ it differs in that re-
spondents were not asked which products they
would "consider appropriate" for a particular Benefits Wanted vs. Benefits Received
cleaning use, but rather which product(s) they ac- At the heart of a Benefit Structure Analysis
tually used for the particular cleaning occasion en- study is an examination of how much each of the
countered during the previous day. Since the ma- 75-100 benefits is wanted and received. This is
trix for this study is summed over all 500 respon- done by summing the ratings across all respon-
dents to reveal a representative cross-section of dents and across all cleaning occasions. Exhibit 3
products and uses on a joint basis, it puiports to shows a Benefit Deficiency matrix for one specific
represent the total U.S. market for household benefit: "Removes Grease." Marginal totals show
cleaning products. the extent to which removing grease was both
26 Joumal of Marketing, October 1976

EXHIBIT 2
MOST USEFUL LINKAGE ANALYSES

Benefits Benefit Benefits

Wanted Deficiencies Received

/N I
Products

Cleaning Chores

\/
t \J

Product Product Product

Characteristics Characteristic Characteristics
Wanted Deficiencies \ Received

wanted and received by housewives^ on the four- ficiency" score of 1, making a total of 162 of the
point scale (see relationship #2 in Exhibit 2). 500 respondents with some perceived deficiency in
Exhibit 3 shows that 196 of the 500 respondents the desired benefit, "removal of grease"—almost
said they wanted to remove grease "a whole lot"; exactly one-third of the sample. While some of the
143 said "not at all"; frequencies for benefits re- 79 deficiency scores of I should be discounted be-
ceived are similar (147 reported being able to re- cause of the basic unreliability of any such survey
move grease "a whole lot" while 187 said "not at data, it is still apparent that there exists a rather
all"). However, simple comparisons of marginal sizeable group of respondents who were not satis-
frequencies obscure the extent to which specific fied with the grease removal properties of the
benefits are wanted but not received. cleaning compound they used.
For example, 34 respondents said they wanted "Deficiency Indexes" can be calculated from
to remove grease "a whole lot" but the cleaning these data in a number of ways:
product they used did so "not at all", yielding a 1. Average Deficiency: .335
rather impressive "benefit deficiency" score of 3 (4 This is a simple weighted average of both
minus I). Another 26 wanted to remove grease "a positive (wanted but not received) and nega-
whole lot" but did so only "somewhat" (4 minus 2 tive (received but not wanted) deficiencies.
= 2); 23 wanted this "pretty much" but got it 2. Average Positive Deficiency: .566
"not at all" (3 minus 1 - 2), for a total of 49 Since most benefit statements are worded in
respondents with a "benefit deficiency" score of 2. the affirmative (something potentially desir-
An additional 79 respondents had a "benefit de- able), "positive deficiencies" would normally
Benefit Structure Analysis 27

be much more important than negative. An Product Characteristic Deficiency matrix may be
Average Positive Deficiency Index can be prepared for each of the 50-75 characteristics
calculated by ignoring negative deficiencies using the same format as for benefits.
(setting them equal to zero) and recalculat-
Plots of Benefits Wanted vs. Deficiencies
ing the weighted average deficiency, which
more accurately reflects the extent to which While the above information provides useful
housewives wanted a particular benefit but diagnostics for each benefit and product charac-
did not get it. teristic, it needs to be organized and summarized
3. Proportion With Deficiency: .329 for the market planner. Plotting the Average
The above weighted averages are useful in Wanted Benefit scores (as columns) vs. Average
later plots but they do not communicate Deficiency scores (as rows) for each benefit/charac-
very meaningful information to manage- teristic is a simple and vei"y useful first step.
ment. The proportion of the sample with Using benefits to illustrate several possible plots
some specified deficiency (1 or more, 2 or would be meaningful: Average Wanted Benefit
more, etc.) is much clearer and is calculated score vs. Average Positive Deficiency score for
easily. each benefit; Average Wanted Benefit score vs.
Proportion of Sample with Some Deficiency;
4. Average Negative Deficiency: .231 Average Wanted Benefit score vs. Proportion
Setting all positive deficiencies equal to zero Wanting the Benefit "pretty much" or "a whole
and calculating a weighted average negative lot" that felt some deficiency.
deficiency yields the Average Negative De-
ficiency. The plots enable rapid identification of "oppor-
tunity points" as a basis for focusing further diag-
5. Proportion With Negative Deficiency: .150 nostic and planning efforts. Opportunity points
Proportion of all respondents having a nega- are benefits/characteristics that are wanted but
tive deficiency of 1 or more (or 2 or more, if not received, or product characteristics that are
desired), indicating they got the benefit more received although not wanted. In the case of
than they wanted it. characteristics, the planner can instruct the de-
Similar deficiencies can be calculated for each velopment labs to add or remove certain product
of the product characteristics; however, the situa- features (e.g., color, odor, antiseptic), assuming
tion here is quite different than for benefits. Since this is possible.
pixiduct characteristics are neutral (i.e., purely A single plot can portray the entire cleaning
descriptive), whereas most benefits are positive, products market. But even more useful informa-
negative deficiency indexes for the former should tion may be gained from plots for each of the
be more important than the negatives in the case major product types as well as plots for each of
of benefits, since they reflect the extent to which a the major cleaning uses. Such plots often show
particular characteristic was received although two or three products types (or u,ses) which are
not wanted (e.g., strong aroma, grittiness). A particularly vulnerable because they are not ade-

EXHIBIT 3
BENEFIT DEFICIENCY MATRIX

Benefit #49: "Removes grease

Received: "Not at all" "Somewhat" "Pretty much" "A Whole lot" Marginal Sums
Wanted Wanted Got
"A whole lot" 34 26 27 109 196 147
"Pretty much" 23 30 25 20 98 75
"Somewhat" 22 15 13 6 56 84
"Not at all" 108 13 10 12 143 187
TOTAL: 187 84 75 147
Averages: Wanted: 2.70 Received: 2.36
Want minus Received - - 3 -2 -1 0 1 2 3
No. of Respondents: 12 16 46 257 79 49 34
(Please note that these totals include different combinations of "Want minus Got").
28 Journal of Marketiiig, October 1976

EXHIBIT 4
COMPARATIVE SIZE OF MAJOR BENEFIT WANTED
SEGMENTS & BENEFIT DEFICIENCIES
% of Sample % of Sample
Wanting Beneht Having
BENEFIT GROUPING Pretty Much or More Deficiency

A. Removes dust/ 76% 32%

dirt/film
V////A
B. Removes porcelain 63
stains 28%
V/TT\
C. Leaves no film/
residue/scratches 61 26%

D. Removes grease/
wax/stains 59 41%
/////I
E. Convenient to use/
store 58 34%
V/ //A
F. No unpleasant odor
during/after use 54 62%

G. Gentle on hands/skin 35 56%

V / / /k
quately meeting some needs that are highly de- These benefit segments are produced by group-
sired or even moderately desired. Carefully de- ing together specific benefits that are perceived as
signed new product entries could be expected to similar across all respondents and all cleaning oc-
compete effectively in these product areas because casions. What is needed is a technique that simul-
they are specifically targeted at benefits which people taneously groups benefits and weights them in
say they want a great deal but are not getting now. rank-order of frequency desired. If one or more
This type of analysis was used for two clients in benefits that are wanted "a great deal," are, in
the food and beverage industry and clearly re- fact, "isolates" (i.e., are not perceived as similar to
vealed the existence of such gaps for several any other benefits) then the technique used must
major product types; for example, a "pick-me-up" identify "isolates" as well as groupings. Presuma-
beverage that failed to "pick up"; a type of sauce bly the desired technique would be sequential: ex-
that was considered far too fattening and that tracting first the benefit or benefit cluster desired
stained clothing and table coverings. most, then the one desired next, and so on until
all of the most important benefits are accounted
Major Benefit and Deficiency Structure
for.
The next analytic step is to construct a Benefit Unfortunately, no such technique appears to
and Deficiency Structure graph for the household have been developed. The best that can be done is
cleaning products market on an aggregate basis. to judgmentally combine the information from a
This analysis shows that the primary benefits factor analysis of all 75-100 benefits (from "bene-
which people want from cleaning products can be fits wanted" ratings) plus the average "benefit
grouped into seven major categories, labeled A wanted" scores for key "marker" benefits which
through G, as shown in Exhibit 4. For example, most clearly define a cluster (i.e., have the highest
Benefit Group A includes benefits that are similar factor loadings). These groupings can then be dis-
in terms of removal of surface dust, dirt, film; played in bar graph form, in rank-order of desired
Segment D includes the removal of heavy grease, intensity. Deficiencies for each of these benefit
wax, or stains; Segment F involves odor and structure dimensions may be developed and jux-
aroma. taposed on the same graph.
Benefit Structure Analysis 29

Exhibit 4 shows at a glance the "benefit and particular benefit under study, when viewed
deficiency structure" of the household cleaning across all products and respondents.
products market. Note that it shows each benefit
Reiationships of Benefits and Products
grouping separately; combinations of two or more
desired benefit groupings that are wanted can be In a manner parallel to linking benefits with
developed also. The exact same analysis should be product characteristics, benefits can be linked to
done for product characteristics, to develop a products to show what types of products are con-
market structure in comparable terms. Large de- sidered best at delivering each of the benefits. One
sired benefit linkages with large deficiencies can useful approach is to rank-order all product cate-
be quickly spotted, as well as smaller desired gories in terms of their average benefit received
benefit linkages with large deficiencies. The latter scores for each benefit, with summation across all
often signal market segments that offer real prom- respondents who reported on a particular prod-
ise for some new product, or for the repositioning uct. The resultant rankings have been found to be
of some existing product. For example, analysis of quite useful in indicating the general types of
these linkages showed that not many respondents products that are most (and least) likely to pro-
wanted to remove grease/wax/stains with a prod- duce a specific benefit (see relationship #4 in
uct that had no unpleasant odor, but among those Exhibit 2).
who did, a sizeable proportion perceived that the Simultaneous Row/Column Linkages
products they were, using were not delivering this
combination of benefits. Up to this point, all benefit linkages have been
done one-at-a-time, and the resultant data are
more diagnostic than structural. The same gen-
Relationships of Benefits and Characteristics eral types of information can be presented in a
form that provides more over-all structure to the
Once it is determined what benefits and combi- market at the same time that it allows specific
nations of benefits are wanted, how are these diagnostic information although presented in a
benefits achieved? A Benefit Structure Analysis different format. This is done using a process that
study shows what types of product characteristics groups and displays two variables simulta-
are seen by respondents as being most closely asso- neously.
ciated with any single benefit or any group of related The simultaneous row/column clustering algo-
benefits. This provides a linkage between benefits rithm used in Benefit Structure Analysis was de-
and those product characteristics which produce veloped by Dr. Edward W. Forgy^ and named
the desired benefits (see relationship #3 in Ex- "LARC" (large row/column simultaneous cluster-
hibit 2). ing). Exhibit 5 shows the LARC analysis of clean-
While this can be done in several ways, the ing tasks (columns) vs. benefits wanted (rows).
simplest is to correlate benefit received scores Cleaning tasks that are seen as similar in terms of
with product characteristics received scores, benefits desired are grouped together, and bene-
across all respondents and cleaning occasions. fits that are seen as similar in terms of cleaning
Since both are measured on the same four-point tasks are grouped. Heavy lines have been drawn
interval scale, the size of the correlation should be arbitrarily around blocks of the largest numbers;
quite meaningful. these blocks show the most clearly-defined link-
The sizes of the correlations are important. Sel- ages between benefit groupings and cleaning task
dom are any correlations in excess of .40 found. groupings. That is, lines show what groups of
Most benefits show highest correlations in the .30 similar benefits are desired by housewives for a
range. This makes the job of the new product particular grouping oi similar cleaning tasks, and
planner more difficult, in that relationships are vice versa.
not high enough to clearly indicate which product It is important to note that LARC or any other
characteristics should be used to produce a given similar process will not define clusters of rows or
benefit. Still, comparisons of those relating high- columns as clearly and accurately as will the var-
est with those relating lowest usually indicate ious algorithms that have been designed solely for
rather clear differences in the general types of clustering, such as optimal partition clustering,
product characteristics that are and are not per- hierarchical clustering, and even factor analysis.
ceived to be related to a particular benefit. If none However, this weakness is compensated for by the
of the relationships is high (perhaps .30 or more), ability of a LARC-type analysis to both group and
it would suggest that the planner is free to use display simultaneously the linkages between the
any product characteristics he chooses, since no two variables under study (see relationship #5 in
specific characteristic is related strongly to the Exhibit 2).
30 Joumal of Marketing, October 1976

Several types of row/column analysis are useful: Row/column clustering provides a highly useful
1. Products by cleaning task form of structure for the market planner. By look-
ing at the linkages of cleaning tasks and benefits
2. Products by benefits wanted, received, and
(indicated by the larger numbers that have been
the various benefit deficiencies
enclosed within the heavy lines on Exhibit 5), he
3. Cleaning tasks by benefits wanted, received, can see at a glance which general types of cleaning
and the various benefit deficiencies tasks require which general types of benefits (or
4. Products by characteristics desired, received, product characteristics). For example, the first
and their deficiencies four columns of Exhibit 5 show that cleaning
5. Cleaning tasks by characteristics desired, re- tasks involving furniture (nos. 19, 4. 14, 20) re-
ceived and their deficiencies quire products that remove dust, dirt, and film
6. Benefits/characteristics wanted, received, and (benefit nos. 12, 38, 42, 14), and that leaves no
their deficiencies vs. such other question- residue or scratches (benefit nos. 17, 36, 43). The
naire information as demographics, psycho- market planner can check the array of existing
graphics, brand, usage volume, etc., as avail- products to see if they deliver each of the specific
able benefits wanted in each grouping. And the group-
Exhibit 5
Benefits Wanted by Cieaning Occasion

Cieaning Tasks
Data ordered by row/column simiiaritles
6 7 6 9 10 11 12 13 14 15 lo 17 16 19 £0 21
la 3 13 19 6 12 21 1 7 9 17 10 20 16 8 4 il S iS 14
Benefit Freq. Ij 4 14 20 7 13 22 2 ti 10 18 11 21 17 9 it id o 16 15

2 12 74 76 11 19 77 t<* 99
28 38 b8 49 49 18 12 4 61 35 42
32 42 66 41 45 9 10 to 20 19
t 14 4t 47 40 tl .ii 44 a3 6b
12 22 4 4 3
50 60
22 32 11 10 3 a
18 14 14 3 7
8 12 16
47 57 lO -12
37 47 12 11 8 9
ID 59 U 14 t 6
11 **9 14 17
12 30 40 6 10
19 29 19 ib 12 10
13 61 17 12
14 51 12 14 9 b
15 39 49 7 U
21 31 16 la 8 Ii
16 23 26 17
17 13
29 l t i
l7 16
18 36 46 It 11
41 51 25 17 8 b
19 54 33 33
20 44 3 i
21 10 20 10 17 40 36 3 b
2Z 17 48 46 b
7 36 17 17 19 45 47 15
23
24 26 43
36
30 33 36 u IU
25 33 44 32 32 7
26 34 58 2b 27
ta 30 40 26 3a 5 b
27 27 19 32
28 20 66 3b 31 2<4
29 17 19 75 31 29 24 24 13 15
30 9 50 20 24 21 30 to U 17
31 40 52 14 13 25 24 32 31 It 19
32 42 39 494 14 20 24 26 26 27 26 20
33 29 53 454 16 14 14 15 20 2Q 36 19
34 43 15 424 15 14 22 24 20 24
35 5 55 406 13 22 22 2d 17 16
36 45 33 373 15 12 24 29 6 b
37 23 16 357 11 14 la 20 6 b
38 6 41 295 21 22 il IU
39 31
25 250 19 22 22 22
15
24 233 24 22 18 19
to 14
45 235 11 14 11
35
35 236 14 14 6 13
*! 13 372 13 13
42 25 34 30 25 23 2a 30 20
14 13
43 3 395 27 29 24 24
24
48 458 27 24 29 28 25 17
44 26 439 37 25
45 38 56 10 11 13 lb i!0 21 Jl c!6 ;j4 •i.i Ji 4i &b
16 632 33 27 JO 2o
46 28 15 21 27 i4 4b 35 43 30 27 29 2b 2a 33 35 34
47 46 570 29 27 27 35 lb 14 17 19 17 ?0 24 29 23 22 i.k 51 46 5d
18
37 592 15 it, 65 60 6a
ta 21 15 14 13 24 20 13 It 25 29
20 33 30 27 29
27 583 12 15 13 26 19 21 17 14 30 24
36 33 41 3* 35 :>! 53 47 4b
49 11 11 668 t<: 41
bO 15 19 13 27 25 19 19 36 30 40 41 36 to 51 57 41
1 824 17 30 4>) 4?
51 17 27 28 31 35 24 to 38 5b 47 48 b7 Ii. 77 49
Benefit Structure Analysis 31

ings themselves may suggest other similar or re- related to the various uses for which each product
lated benefits that were not covered specifically in was seen as being suitable. This provided "struc-
the study that should be considered. ture" for a particular type of consumer product or
With this information the planner can: service market. The household cleaning products
• See how many general types of products study utilizes this basic idea in relating products
should be on the market. to uses, benefits wanted/received and deficiencies
to products and to uses (see Exhibit 5), product
• Qjnsider possible additional uses for existing characteristics desired/received and deficiencies
products (i.e., a product designed for one task to products and uses, plus many additional rela-
in a particular cluster should also be suitable tionships within the complete benefit matrix.
for other tasks in that cluster—or could be
modified to do so). BSA also draws on the work of Haley, who was
not the first to be concemed with the benefits
• Conceive new products that deliver the exact which people wanted from products/services but
combinations of benefits wanted for a single who best showed how markets could and should
cleaning task (or for a group of related tasks). he segmented on the basis of benefits, since "the
• Reposition existing products by stressing ad- benefits which people are seeking in consuming a
ditional benefits within a benefit cluster that given product are the basic reasons for the exis-
are already offered by a company's own tence of true market segmentation."'
. .|>roducts, etc. . ,. .. „_., ., Finally,-BSA is related to the various product
•' • •'".* • " •' • " • • • y''^'V''tr^. positioning models proposed by Green.*' Johnson/
Additional Anaiyses . ' '•'•'•-.'•'.'•'' *'.
and others. These efforts utilize quantitative tech-
niques to identify the principal types of product
There are many additional analyses that might characteristics/benefits that are used by consum-
be useful fw the market planner. For example, ers in perceiving or ex'aluating competing products
those respondents who most wanted a particular within a specified product type. This is similar to
cleaning benefit or characteristic can be sorted the Benefit and Deficiency Structure graph shown
out and compared to the total sample of respon- in Exhibit 4. (In the household cleaning products
dents in terms of demographics, at what time of study products were plotted on the first three
day, and other questionnaire data not included in benefit segment dimensions; this 3-dimensional
previous analyses. One firm asked for mental and configuration was not shown due to space lim-
physical "feeling tone," plus limited psychograph- itations.)
ics.
Another useful analysis consists of isolating Managerial Considerations
those respondents who most wanted a particular
benefit (or benefit cluster, obtained from factor Benefit Structure Analysis does not directly
scores) and determining the following averages compete with any of the above models. It was
for each product characteristic: (I) average developed especially for locating new product op-
wanted score; (2) average received score; (3) aver- portunitities in very broad product or service cat-
age positive deficiency score; (4) average negative egories, such as an entirely new type of food (e.g..
deficiency score. Comparisons among these aver- Instant Breakfast) or banking service (e.g., bank
age scores show which characteristics are particu- credit card) or alcoholic beverage (e.g.. Malcolm
larly crucial for a given benefit or benefit cluster. Hereford's 30-proof Cows). In such cases the mar-
(This also can be done by correlation analysis, as ket planner cannot be at all sure where his best
reported earlier.) opportunities might lie and needs a technique to
The limit of possible useful analyses is based analyze many different types of loosely related
largely upon the imagination of the research products/services in a single study. BSA enables
analyst or market planner. him to spot weaknesses and opportunities in each
general type of product or type of use, oppor-
Summary tunities that might well be overlooked on an a
priori basis.
Benefit Structure Analysis (BSA) was developed For example, it was pointed out earlier that one
out of a confluence of ideas from several earlier BSA study identified a type of beverage product
new product development models. Its greatest the was used primarily for a "pick-up" but that
legacy is from the Market Structure Study pro- failed to deliver this benefit sufficiently. In this
cess proposed in the early 196O's by Stefflre.** In particular case the client company: (1) decided it
particular, his Item-Use matrix showed how an would not develop and introduce a product of this
array of existing products could be systematically same type that provided more pick-up; (2) spent
32 Journal of Marketing, October 1976

many hours of discussion with planning, labora- {i.e., groupings of similar individual benefits/
tory, and management personnel considering new characteristics) will often account for upwards of
types of pick-up beverages that would be entirely 70% of the variance in overall evaluations of
different from anything on the market; (3) decided products or frequency of use for a specific type of
to abandon the project entirely as being incom- product.
patible with the basic product technology the The rather simple principal line of questioning
company had had its primary experience with. In that is at the heart of Benefit Structure Analysis
the case of the sauce product that was considered (benefits/characteristics wanted/received and their
too fattening and staining, that particular com- deficiencies) can easily be included as part of
pany also decided not to develop and introduce a many consumer surveys that have some other
product of this type since it did not have re- primary objective. This is especially appropriate
sources to take on the existing well-established when a firm has conducted previous research on a
brands; however, it did utilize this information in particular product line and feels confident that it
reviewing its own sauce products and in planning understands the benefits/characteristics that are
entirely new types of sauces. important for its product. Sometimes only ten to
On the other hand, if the market planner is twenty benefits or characteristics need to be cov-
asked to work within a rather sp)ecific product ered.
category, such as white liquor beverages, soft No single study can provide all types of market
drinks, small cars, or be«r, he is probably better information needed by the planner. The technique
advised to utilize one or more of the models or presented here provides direction as to general
research processes proposed by Green, Johnson, areas of greatest opportunity for new product/
and others discussed above. In such cases the service development plus specific diagnostic in-
range of benefits and/or product characteristics formation useful for either modifying or reposi-
that have primary relevance is much narrower. tioning existing products. Through broad applica-
Most investigators, including the present author, tion it may prove to be a useful tool for market
have found that the two or three major benefits analysis and planning.

ENDNOTES
1. Paul E, Green, "Marketing Applications of MDS: As- 4. Stefflre, same as reference 2 above.
sessment and Out\ook." Joumal of Marketing, Vol. 39 No. 1 5. Russell I. Haley, "Benefit Segmentation: A Decision
(January 1975), pp. 24-31. Oriented Research Tool," Joumal of Marketing, Vol. 32 No. 3
2. Volney Stefflre, "New Products and New Enterprises," (July 1968), pg. 30.
Market Structure Studies, Inc., 1968. 6. Paul E. Green and Yuram Wind. Mulliatlrihute Deci-
3. Dr. Edward W. Forgy is a statistical consultant in Los sions m Marketing: A Measurement Approach (Hinsdale, Il-
Angeles, CA. His program is proprietary, but a somewhat linois: The Dryden Press, 1973).
similar procedure is described in "Some Eliciting and Com- 7. Richard Johnson, "Market Segmentation: A Strategic
putational Procedures for Descriptive Semantics," V. Stef- Management Tool." Joumat of Marketing Research. Vol. VIII
flre, P. Reich, and M. Wendell, in Explorations in Mathemat- (February 1971). pp. 13-18.
ical Anthropology, P. Kay, ed. (Cambridge: M.I.T. Press, 8. Benefit Structure Analysis was not actually used for
1971). any of these products, however.

MARKETING MEMO
What is a weed? A plant whose virtues have not yet been discovered.
—Ralph Waldo Emerson, Fortune of
the Republic