Managing a dispersed Product development process potx

14 FOCUSING THE DESIGN TEAM BY IDENTIFYING STRATEGIC CUSTOMER NEEDS.... Managing a Dispersed Product Development ProcessDahan and Hauser Figure 1: Tradeoffs in New Product Development ba

For more information,

please visit our website at http://ebusiness.mit.edu

A research and education initiative at the MIT Sloan School of Management

Managing a Dispersed Product Development

Process Paper 103

Ely Dahan John R Hauser

October 2000

ByEly DahanandJohn R Hauser

October 2000

for the Handbook of Marketing

Barton Weitz and Robin Wensley, Editors

Ely Dahan is an Assistant Professor of Marketing at the Sloan School of Management, M.I.T., 38Memorial Drive, E56-323, Cambridge MA 02142 He can be reached at 617 253-0492, 617 258-

7597 (fax), or edahan@mit.edu

John Hauser is the Kirin Professor of Marketing, Sloan School of Management, M.I.T., 38Memorial Drive, E56-314, Cambridge, MA 02142 He can be reached at 617 253-2929, 617258-7597 (fax), or jhauser@mit.edu

This research was supported by the Center for Innovation in Product Development at M.I.T

Managing a Dispersed Product Development Process

Table of Contents

THE CHALLENGE OF A DISPERSED PRODUCT DEVELOPMENT PROCESS 1

PRODUCT DEVELOPMENT – END TO END 2

AN INTEGRATED PROCESS 2

PRODUCT DEVELOPMENT AS AN END-TO-END PROCESS 4

THE PRODUCT DEVELOPMENT FUNNEL, STAGE-GATE, AND PLATFORMS 5

THE FUZZY FRONT END: OPPORTUNITY IDENTIFICATION AND IDEA GENERATION 7

SURVEYS AND INTERVIEWS 8

EXPERIENTIAL INTERVIEWS 9

THE KANO MODEL: DELIGHTING CUSTOMERS 10

THE INNOVATOR’S DILEMMA AND DISRUPTIVE TECHNOLOGIES 11

EMPATHIC DESIGN AND USER OBSERVATION 12

UNDERLYING MEANINGS AND VALUES 13

KANSEI ANALYSIS AND THE MIND OF THE MARKET 13

BENEFIT CHAINS 14

FOCUSING THE DESIGN TEAM BY IDENTIFYING STRATEGIC CUSTOMER NEEDS 15

TEAM-BASED NEEDS-GROUPING METHODS: AFFINITY DIAGRAMS AND K-J ANALYSIS 16 CUSTOMER-BASED NEEDS-GROUPING METHODS: THE VOICE OF THE CUSTOMER 16

NEW WEB-BASED METHODS FOR THE FUZZY FRONT END 16

IDEATION BASED ON CUSTOMER NEEDS (AND OTHER INPUTS) 17

OVERCOMING MENTAL BLOCKS 17

TRIZ (THEORY OF INVENTIVE PROBLEM SOLVING) 18

INVENTIVE TEMPLATES 18

SUMMARY OF METHODS FOR THE FUZZY FRONT-END 18

DESIGNING AND ENGINEERING CONCEPTS AND PRODUCTS 19

LEAD USERS 19

EMPLOYEE FEEDBACK: KAIZEN AND TEIAN 20

SET-BASED DESIGN AND MODULARITY 21

PUGH CONCEPT SELECTION 21

VALUE ENGINEERING 21

QUALITY FUNCTION DEPLOYMENT AND THE HOUSE OF QUALITY 22

TRADEOFFS AMONG NEEDS AND FEATURES: CONJOINT ANALYSIS 23

NEW WEB-BASED METHODS FOR DESIGNING AND ENGINEERING PRODUCT CONCEPTS 29 SUMMARY OF METHODS FOR DESIGNING AND ENGINEERING CONCEPTS AND PRODUCTS 32

PROTOTYPING AND TESTING CONCEPTS AND PRODUCTS 32

TARGET COSTING: DESIGN FOR MANUFACTURING AND ASSEMBLY (DFMA) 33

RAPID PROTOTYPING METHODS 33

PARALLEL CONCEPT TESTING OF MULTIPLE DESIGNS 33

INTERNET-BASED RAPID CONCEPT TESTING 34

AUTOMATED, DISTRIBUTED PD SERVICE EXCHANGE SYSTEMS 34

INFORMATION ACCELERATION 35

PRETEST MARKET AND PRELAUNCH FORECASTING 36

MASS CUSTOMIZATION AND POSTPONEMENT 37

SUMMARY OF PROTOTYPING AND TESTING CONCEPTS AND PRODUCTS 38

ENTERPRISE STRATEGY 38

THE CHALLENGE OF DEVELOPING AN EFFECTIVE PRODUCT DEVELOPMENT ORGANIZATION 38

BOUNDARY OBJECTS 39

COMMUNITIES OF PRACTICE 39

RELATIONAL CONTRACTS 40

BALANCED INCENTIVES 40

DYNAMIC PLANNING 40

DEPLOYMENT OF CAPABILITIES WITH WEB-BASED TOOLS 40

PROCESS STUDIES OF THE ANTECEDENTS OF PRODUCT DEVELOPMENT SUCCESS 40

ADJUSTING PRIORITIES TO MAXIMIZE PROFIT 43

A VISION OF THE FUTURE OF PRODUCT DEVELOPMENT 45

The Challenge of a Dispersed

Product Development Process

New product developm ent has a long

his tory in marketi ng incl udi ng research on

cus tom er preferences (Green and Wind 1975,

Green and Sri ni vas an 1990, S ri nivas an and

S hocker 1973), product pos it ioning and

s egmentati on (C urrim 1981, Green and Kri eger

1989a, 1989b, Green and R ao 1972, Haus er

and Koppel m an 1979), product forecas ti ng

(Bass 1969, J am i es on and Bas s 1989, Kalwani

and S i lk 1982, Mahaj an and Wind 1986, 1988,

M cF adden 1970, Morri s on 1979), and tes t

m arket ing (Urban 1970, Urban, Hauser and

R obert s 1990) The appli cat ions have been

m any and vari ed and have led t o a deeper

unders tandi ng of how to gather and use

i nform at ion about the cus t om er in t he des ign,

t es ti ng, l aunch, and managem ent of new

products Many integrati ve texts on product

devel opm ent from a m arket i ng pers pecti ve

have been publi s hed to review the i s sues , t he

m et hods, and the appl icat i ons (Dolan 1993,

Lehmann and Winer 1994, M oore and

P es sem ier 1993, Urban and Haus er 1993, Wi nd

1982)

M arket ing, wi th it s focus on t he cus tomer,

has had great s ucces s Tools s uch as conj oi nt

analys is , voi ce-of-t he-cus tomer anal ys is ,

perceptual mappi ng, int ent ion scali ng, port fol io

opt im i zati on, and li fecycl e forecas t ing are now

i n com mon use Fi rm s t hat cont inuousl y and

effici entl y generate new products t hat are in

t une wit h their end cus tom ers’ needs and want s

are m ore l i kely to t hri ve (Gri ffi n and P age

1996) Di rect com municat i on wi th cust om ers

all ows firm s to learn from cus t om ers and tail or

products t o t hei r requi rem ents

In paral lel wit h t he devel opment of

prescripti ve tools , res earchers have s tudied the

correl at es of new product success i denti fyi ng

com municat i on between m arket ing and

engineering as one of t he most im portant

factors in success (C ooper 1984a, 1984b,

C ooper and Kl ei nschm i dt 1987, Doughert y

1989, Griffin and Hauser 1996, Souder 1987, 1988) As a res ul t, organizat i onal process t ool s

s uch as cross -functi on teams (Kuczm ars ki

1992, Souder 1980), quali t y funct iondeployment (Haus er and Cl aus ing 1988), andco-locat ion (Al l en 1986) were devel oped topromot e the s haring of ideas and the clos e

i nt egrat ion of engineering deci si ons wit hcus tom er needs P roces s ori ent ed t ext books now rout inely cons ider marketi ng is s ues and

t he need t o i nt egrat e engi neeri ng wi th t he

m arket ing funct i on (M cGrat h 1996, Ul ri ch andEppinger 2000)

As we move into the 21s t cent ury, newchall enges and opport unit i es are ari si ng driven

by gl obal market s, gl obal competi ti on, t heglobal dis persi on of engi neeri ng tal ent, and theadvent of new i nform ati on and com municat i on

t echnologi es such as el ect roni c m ai l , the worl wide web, and i ncreas ed el ectroni c bandwi dt h The new vi s ion of product devel opment is that

d-of a highl y dis aggregat ed process wi th peopleand organi zat ions spread throughout the world(Holm es 1999) At the s am e t im e products arebecom i ng i ncreas ing com pl ex wi t h typical elect ro-mechani cal product s requi ri ng cl ose t o

a m il l ion engineering deci si ons t o bri ng them to

m arket (Eppinger, Whi tney, S mi t h and Gebala

1994, Eppi nger 1998) Even software products

s uch as Mi crosoft Word or Nets cape requi redis aggregat ed, but coordi nat ed process es

i nvol ving hundreds of developers (C usumanoand S elby 1995, Cusum ano and Yoffie 1998)

C om pet it ive pres sures m ean t hat t im e t o market has become as key to new product success as

m arket ing’s ori ent at i on on cus t om er needs andcus tom er s ati sfact ion (Sm i th and Rei nert s en1998) Because product s are m arket ed

t hroughout the world, firm s face the t radeoffbet ween st andardizat i on for cos t reducti on andvariet y for s at i sfyi ng a broad set of cus tomers Thi s has expanded the need for marketi ng to

l ook beyond t he si ngl e product to focus on theproduct pl atform (Moore, Louvi ere and Verma1999)

In thi s chapt er we l ook at t he st at e of the

art i n res earch that address es thes e new

chall enges for the m arket i ng comm uni ty We

begin wi th an overvi ew of the int egrat ed

end-t o-end producend-t devel opm enend-t proces s indicaend-ti ng

m arket ing’s rol e i n addres si ng the chall enges of

devel opi ng profi tabl e products (and pl at forms )

The remainder of t he chapt er addres s es s pecifi c

res earch chal lenges rel at i ng t o t he end-t o-end

proces s We organize t he remai ni ng sect i ons

around t he vari ous s t ages of development

recognizing t hat , in pract ice, thes e s tages are

oft en it erati ve and/ or int egrat ed Speci fi cal ly

we address , i n order, t he st rat egic end-t o-end

product devel opm ent proces s, t he fuzzy front

end of cus t om er opportuni t y ident ifi cati on and

i dea generati on, t he process of det ail ed desi gn

and engi neeri ng of product s and proces ses , the

t es ti ng phase where concepts and product s are

protot yped and tes ted, and t he enterpris e and

organi zati onal strat egy necess ary for success

We cl ose wi th a vi si on of the fut ure of res earch

of product devel opment

Product Development – End to

End

In the l at e 1980s and earl y 1990s a

m arket ing focus on product developm ent

s tres s ed cust om er sat is facti on Res earchers in

m arket ing bel ieved t hat t he key t o succes s was

a bet t er unders t andi ng of the voi ce of t he

cus tom er and a bet ter abi l it y to li nk that voi ce

t o the engi neeri ng deci si ons t hat are made in

l aunching a product F or exam ple, Menezes

(1994) document s a case where Xerox moved

from a focus on ROA and m arket share t o a

focus on cust om er sat is facti on Im portant

res earch duri ng that peri od included new ways

t o underst and t he voi ce of t he cust omer (Griffin

and Haus er 1993), new ways t o devel op

opt im al product profi les in the cont ext of

com pet it ion (Green and Kri eger 1989a, 1991),

m ore effici ent preference meas urements

(Srini vasan 1988), and the abi l it y to handl e

l arger, more com pl ex cust omer inform at ion

(Wi nd, Green, S hiffl et, and Scarbrough 1989)

At the s am e t im e t he qual i ty m ovement focus edproduct devel opm ent engineering on improvedrel iabil it y t hrough conti nuous im provement

s uch as Kai zen met hods (Im ai 1986), st at i st icalquali t y control (Dem i ng 1986), modi fiedexperi ment al des ign (Taguchi 1987), anddes ign for manufacturing (Boot hroyd andDewhurst 1994) There were many success es

i ncluding a t urnaround of the maj or US aut om obi le manufacturers M any engi neers cam e to bel ieve that the key t o s ucces s was abet ter qual it y product

Als o during t hat t im e bot h m arket ing andengineering real ized that ti me to m arket wascri ti cal Marketi ng saw the phenom enon as

t hat of rewards to earl y ent rants (Gol der andTel li s 1993, Urban, Carter, Gas ki n, and Mucha1986) whil e engi neeri ng s aw, am ong other

t hi ngs , the l os t profit s due t o del ays (S mi th and

R ei nerts en 1998) B oth cust om er sat is facti onand t i me-t o-m arket became panaceas that, ifonl y the fi rm coul d achieve them, woul dguarantee succes s and profit abi li ty

An Integrated Process

Today, both industry and academiaview successful product development as anintegrated process that must overcome manytradeoffs, as depicted in Figure 1 Customersatisfaction, time to market, and cost reductionthrough total quality management are allimportant, but none is viewed as the onlyguarantee of success

Managing a Dispersed Product Development Process

Dahan and Hauser

Figure 1: Tradeoffs in New Product

Development (based on Smith and

Reinertsen 1998)

All else equal, a product will be more

profitable if it delivers customer benefits

better, is faster to market, costs less to

produce, and costs less to develop Figure 1

puts research on product-development tools

and methods into perspective Research

should be directed to assure (1) that the firm is

operating on the efficient frontier with respect

to each of these strategic goals and (2) that thefirm is making the best tradeoffs among thesegoals

R es earch m ust recogni ze t hat t here are

t radeoffs along the effici ent front i er Forexampl e, i f we focus on j ust t wo of the manygoals of product developm ent , then the effi ci ent front i er i n F igure 2 suggest s that there are

t radeoffs bet ween cus tomer s at i sfact ion andreuse A firm can become too com mi t ted toeit her F or example, t he si gni fi cant reuse ofcom ponents , s oft ware and des igns may get theproduct to the market fas t er and reducedevel opm ent cos t s (e g , Wit ter, Cl aus ing,Laufenberg, and de Andrade 1994), but t hefirm may s acrifi ce t he abi li ty to s ati sfycus tom er needs and m ay mi s s out on ways toreduce product cos ts Si m il arl y, qual it yfunct i on depl oym ent (QF D) may be aneffect ive means to deli ver cus t om er benefit s, but s ome appl icati ons are too cum bersomereduci ng t i me t o m arket and increas i ngdevel opm ent cos t

Figure 2: Quantifying the Tradeoffs in Product Development

0 2

4 6

8 10 0

2 4 6 8 10

Platform Reuse

Customer Satisfaction

Profit ($M)

benefits at the right costs Such modifications

include just-in-time QFD (Tessler and Klein

1993), turbo QFD (Smith and Reinertsen

1998), and simplified QFD (McGrath 1996)

Reuse and QFD are just examples As we

review various product develop tools and

methods, the reader should keep in mind that

the tools work together to enable the firm to

make the appropriate tradeoffs among the four

strategic goals in Figure 1

In order t o m ake t hes e tradeoffseffect ivel y, mos t fi rms now vi ew product devel opm ent (PD) as an end-t o-end proces s

t hat draws on m arket i ng, engineering,

m anufact uri ng, and human devel opm ent

F igure 3 is one representati on of an end-to-endproces s Figure 3 is m odi fi ed from a proces sused at Xerox and advocat ed by the Center forInnovati on in P roduct Developm ent (S eeri ng1998) It summ ari zes m any of the forces onproduct devel opm ent and hi ghli ght s

opport unit i es for res earch

Figure 3: Product Development – End to End

F rom our pers pecti ve, t he fi ve forces in

red on t he outer s quare of F igure 3 pres ent t he

ext ernal chal lenges to the P D team All acti ons

are cont ingent on these forces For exam pl e,

s peed to m arket mi ght be more cri ti cal i n t he

highl y com pet it i ve worl d of Int ernet s oft ware

R at her t han 3-year pl anni ng cycles, such fi rm s

m ight adopt 3-year hori zons wi t h adapt ive

i mplem entat ion strat egi es that are reviewed

m onthl y or even weekl y (C usumano and Yoffie

Managing a Dispersed Product Development Process

Dahan and Hauser

1998) The descripti ons i n t he seven blue

rectangl es indi cat e act ions that mus t be taken

F or exam pl e, the firm m us t have a s t rategy for

deali ng wi t h technol ogy (“Technol ogy

S trat egy”) and employ m et hods to underst and

t he benefi t s provi ded t o cus tom ers by

com pet it ive product offeri ngs, ident ify gaps

where benefit s are demanded but not suppl ied,

and unders t and how compet i ti on wi ll respond

(“C om pet it i ve P osi ti oni ng”), whil e “Suppl y

C hain Managem ent ” hel ps t he fi rm (and

ext ended enterpris e) incl ude s uppli ers i n

devel opi ng product s to meet cus tomer needs

In thi s chapt er we revi ew thos e act i ons that are

of great es t i nt erest to a marketi ng audi ence,

nam el y t hos e in the four sol id rect angles

In-bound marketi ng (“Voi ce of t he Cust omer,

C onjoi nt Anal ys i s, et c ”) provi des the wi ndow

on the cus t om er The m yri ad pers pecti ves from

m arket ing, engi neeri ng, desi gn and

m anufact uri ng t hat m ust be i nt egrat ed for

s ucces sful PD m ani fes t thems el ves i n t he form

of a ‘Core Cros s -F uncti onal Team ” “Hum an

R es ources” are import ant, incl udi ng the need to

unders tand the context and cul t ure of the

organi zati on and t he need to develop hum an

capabi li ti es through trai ning, informati on

t echnology, and comm uni ti es of pract ice

(Wenger 1998) “Marketi ng, Engi neeri ng, and

P roces s Tools ” enabl e t he end-t o-end P D

proces s to be both m ore effi ci ent and more

t he s t ages , but the des cri pt ion of PD as a st agedproces s is fairl y uni vers al The key

m anagement ideas are (1) that it is much less expens ive to screen products i n t he earl y s tages

t han in the l at er st ages and (2) that each st agecan i m prove t he product and it s pos i ti oni ng s o

t hat the l i keli hood of success increas es Si m pl ecal cul at ions in Urban and Haus er dem onst rat e

t hat such a s taged proces s i s likel y t o reducedevel opm ent cos t s si gni fi cantl y This s t agedproces s is best summ ari zed by Cooper (1990)who l abels the proces s st age-gate Fi gure 4

s um marizes a typical st age-gat e proces sadapt ed to the struct ure of thi s paper St age-gat e provi des di scipl ine through a series ofgat es in which mem bers of the PD team areasked to j ust ify t he deci s ion to move to the next

s tage – lat er s t ages dram ati cal ly i ncreas e thefunds and effort s inves ted i n thi s get ti ng aproduct to market success ful ly

The funnel in F i gure 3 al so il lus trat es the

concept of pi pel ine management Oft en the bes t

s trat egy for a firm is to have suffi ci ent ly m any

paral l el proj ect s so that it can launch products

t o the m arket at t he most profi tabl e pace

R es earch chal lenges include the ques ti ons of

how m any parall el project s are neces sary, t he

t radeoffs bet ween more paral lel proj ects and

fas ter t im e for each proj ect , and t he num ber of

concepts t hat are needed in each st age of a

paral l el proj ect t o produce the right pace of

product int roducti on F igure 3 does not capture

expli cit ly the import ant characteri s ti c of real

P D process es that st ages oft en overl ap For

exampl e, wi th new met hods of us er desi gn and

rapid prot otypi ng, i t i s pos si ble t o t es t concepts

earli er in the des ign and engi neeri ng st age or t o

s creen i deas more effecti vel y in the concept

s tage Fi gure 3 al so does not capture expli ci t ly

t he fact t hat t he ent ire proces s is it erati ve

(al though we have tri ed t o i ll ust rat e that wi t h

t he feedback arrows in Fi gure 4) F or example,

i f a product does not t es t wel l , it mi ght be

cycled back for furt her development and

ret es t ed In fact , many firms now tal k about a

“spiral process ” i n whi ch the product orconcepts m oves through a series of tight er and

t ight er st ages (e g , C us umano and Sel by 1995).The s m al l ovals in t he end-t o-end P Dproces s (F i gure 3) are ei ther individual products or product platform s In many

i ndus t ri es , i ncl uding com plex elect

ro-m echanical product s, soft ware, andpharm aceut i cals , firm s have found t hat i t i s

m ore profi t able to develop product platform s

A plat form is a set of com mon elements s haredacros s products in t he pl atform fam i ly Forexampl e, Hewl et t P ackard’s ent i re l i ne of i nk-

j et printers is based on a rel ati vel y few pri cartri dge platform s By s haring elem ents , t heproduct can be devel oped more qui ckl y andwit h lower cost Pl atform s mi ght al so l owerproducti on cost s and inventory cost s andprovi de a bas is for flexi ble m anufacturi ng On

nter-t he cusnter-t om er si de, pl anter-t forms enable a fi rm nter-tocus tom ize features i n a proces s t hat has becom eknow as mas s cus tomi zat ion (Gonzalez-Zugas t i, Ot to, and B aker 1998, Meyer andAlvin Lehnerd 1997, Sanderson and Uzum eri

Managing a Dispersed Product Development Process

Dahan and Hauser

1996, Ul ri ch and S teve Eppinger 1995,

B al dwi n and C lark 2000 )

F inal l y, t he ri ght s i de of t he end-t o-end

proces s in Fi gure 3 il lus trat es the growing

t rends t oward m etrics -bas ed managem ent of

P D As the proces s becom es more di s pers ed

among vari ous functi ons , various teams ,

various suppl iers, and throughout t he world and

as product s become m ore complex, there i s a

great er need to balance t op-managem ent

control wi t h the empowerm ent of s el

f-m anaged, cros s-funct i onal teaf-m s To achi eve

t hi s bal ance, fi rm s are t urning t o a m et

rics-bas ed approach in whi ch t eam s are m eas ured

on st rat egi c indicat ors s uch as cus t om er

s at is facti on, t i me t o m arket , producti on cost ,

and development cost If the wei ght s on thes e

m et ri cs are s et properl y, then the teams , act i ng

i n their own bes t int eres t s, wi ll t ake t he act ions

and m ake t he decis ions that lead to the great est

s hort - and long-term profi t (B aker, Gi bbons ,

and M urphy 1999a, 1999b, Gibbons 1997)

Thi s com pl etes our m arket i ng overvi ew of

t he end-to-end product devel opm ent proces s

The i m port ant l ess on, t hat we hope to il l us trate

t hroughout the rem ai nder of thi s chapt er, i s that

t he proces s depends upon all of i ts el em ent s

Alt hough t he det ai led i mpl em ent at ion of each

element varies depending upon technology,

com pet it ion, cus tomers, and suppl iers, a fi rm is

m ore effect ive if it underst ands al l of these

elements and can m anage t hem effect i vely

We now exam ine res earch opport uni ti es

wit hi n each s tage of the PD process by

beginning wit h the fuzzy front end of

opport unit y i denti fi cat ion and idea generat ion

The Fuzzy Front End: Opportunity

Identification and Idea Generation

P erhaps the highes t leverage point in

product devel opm ent is the front end whi ch

defines what the product wil l be, repres ent ed

by the opening of the funnel i n F igure 3 Thi s

decis i on balances the firm ’s core s t rengt hs

versus com pet it i on wi th t he dem and ofpot ent ial cus tom ers R el evant topi cs include

t echnology st rat egy and readiness , cus tom er

i nput , and newer, vi rtual -cust omer met hods

B ecaus e thi s is a marketi ng handbook, we wi ll focus more of t his s ect ion on the obtaini ng

i nform at ion t o sat is fy cus tomer needs and on

i dea generati on We recom mend that readers

i nt erest ed in t echnol ogy readi nes s review

R ouss el, S aad, and Ericks on (1991) or

M cGrat h (1996)

The fuzzy front end may be viewed

t hrough the l ens of uncer t ai n search That is ,

t he desi gn team must cons i der a m ul t it ude ofdes igns in order t o find an ideal s oluti on at the

i nt ers ecti on of cust omer preferences and fi rm capabi li ti es Once the fi rm has det ermi ned t he

s trat egi c val ue of developing a new product wit hi n a part icular cat egory, but before it can

s peci fy the det ail ed requi rements and features

of the des i gn, it mus t sel ect the m ore prom is i ngdes igns to devel op and tes t so as t o m eet devel opm ent -cos t , product i on-cost , cus tom er-

s at is facti on, and ti m e-to-market targets

P romi s ing des igns are t hos e that are t echni cal lydes irabl e, i e feas i bl e des igns that exploit thefirm’s com pet it i ve advant ages, and areatt racti ve to potent i al cust om ers Marketi ng’srol e at the fuzzy front end of PD i s t o reduceuncert ai nt y duri ng t he des ign team’s s earch forwinni ng product concept s by accurat elycapturing cus tom ers’ point s-of-vi ew andcom municat i ng cust om er preferences to thedes ign t eam In s om e cas es, t he process is

m ore direct , wi t h engineer/des i gners obs ervingand comm uni cati ng di rectl y wit h pot ent ial cus tom ers, poss i bl y facil i tated by market ingpersonnel The ease of comm uni cati on and

i nt eract ion over t he Internet has t he pot enti al to

i ncrease t he frequency and effect iveness of

s uch unfil t ered observati ons The proces s of

l is tening to cus tomers in order t o opt im i ze anew product i s iterat ive, as depi ct ed in Figure

5, consistent with the aforementioned “spiralprocess ”

Figure 5: Listening to Customers

Prioritize Needs

Gather Customer Voice

Recognizing the iterative nature of Figure

5, we begin this section by reviewing

t echni ques for gat hering raw data on cus t om er

needs These m ethods i ncl ude direct s urvey

m et hods wi t h whi ch m arket i ng researchers are

fam il i ar, but i ncl ude as wel l Kano’s m odel of

del ighti ng cust omers , t he concept of dis rupti ve

t echnologi es, m ethods t o get at underl yi ng

m eani ngs and val ues, methods for the “mi nd of

t he m arket , ” and benefi t chains We t hen

review m et hods for charact erizi ng and refining

cus tom er needs bas ed on apparent pat terns and

t hemes and methods for organizi ng needs and

i dent i fying m arket s egm ent s Needs mus t be

pri ori ti zed and many marketi ng methods are

qui te effecti ve In the fuzzy front end we us e

t he s i mpler and less cost l y met hods recogni zi ng

t hat any i nform ati on wi ll be refi ned i n the

des ign and prot otype phas e Thus , we save a

review of these “high-fideli ty” m et hods unt il

t he next s ect ion of thi s chapt er However, i n

t hi s secti on we do revi ew some of t he more

com mon m et hods of ideat ion We clos e thi s

s ecti on by exam i ni ng how the Internet is

changi ng t he way we view the proces s of

i dent i fying and meas uri ng cust omer needs

Surveys and Interviews

There are m any cha lleng es when

a tt em p ting to ca pt ure t he voice of t he

custom er, m ea sure preference, a nd

p redict new p rod uct p urcha se b eha vior

During t he fuzzy f ront end t he methods

m ust recog nize t ha t: (1 ) custom ers m ay still be f orm ing t heir preferences a nd

m ay change their opinions by t he tim e

a ct ua l p rod ucts ship , ( 2) it m a y be

d if ficult f or cust om ers t o exp ress t heir

t rue p references ( e.g deg ree of pricesensit iv it y ) due t o socia l norm s, ( 3 ) the

q uest ioning p rocess itself can beint rusiv e, so it is b est t o use m ult ip le,convergent methods, a nd ( 4 )

inf orm at ion g at herers m ay “f ilt er” t he

v oice of t he customer t hroug h t heirown b iases Resea rchers hav e

d ev eloped a nd v a lida t ed m ult ip le

m et hod s in at tem pt s t o ad d ress theseissues

M ahaj an and Wind (1992) s urveyed fi rms about techniques t hey used t o ident i fy cust om erneeds They found t hat 68% of fi rm s usedfocus groups, and 42% used l im i ted productrol l-out s In addi ti on, m any fi rm s used formal concept tes ts , conjoi nt anal ys i s, and Quali ty

F unct i on Depl oym ent (QF D) The st udy als o

s ugges ted the foll owi ng i m provement s forcus tom er research:

• M ore quant i tati ve approaches

• M ore effici ent in-depth probing

• Great er accuracy and vali dit y

• S im pl er and bet t er cust om er feedback

• Great er cus tomer i nvolvem ent

• M ore effect ive use of l ead users andfield sales peopl e

• M et hods that address a long-term, funct i onal l y-int egrat ed s t rategyThe new met hods we revi ew at tem pt t oaddres s many of t hes e concerns However,res earch i s s ti l l underway Each m ethod has its

l im it ati ons and the val ue of t he informat iondepends on the quali t y of executi on of t heres earch

Managing a Dispersed Product Development Process

Dahan and Hauser

Experiential Interviews

F or evol ut i onary des i gns target ed at an

exi st i ng or fam i li ar cust omer bas e, focus

groups (Cal der 1977, Fern 1982), provi de

val uable i nform ati on However, focus groups

are s ubj ect t o social norm s wi t hi n the group

and often focus on i nter-s ubject int eract ions

and t hus m i ss m any of t he cust omer needs that

are hard t o art i culat e or which t he cust omer

cannot expres s effect ivel y i n a group set ti ng

Thus, many fi rm s are turni ng t o experi ent ial

i nt erviews in which the needs and desi res of

cus tom ers are expl ored in one-on-one

i nt erviews in which the cust om er des cribes hi s

or her experi ence wi t h the product class The

i nt erviewer probes deeply into the underl yi ng,

m ore stabl e and long-term probl em s that the

cus tom er i s t ryi ng t o s ol ve Res earch by

Gri ffi n and Haus er (1993) indi cat es that ten to

t went y experi ent ial int erviews per market

s egment el i ci t the vast m ajori t y of cust omer

needs Quali tat ivel y rich i nt erviews at thecus tom er’s locat ion are m ost effect i ve, but expens ive to conduct One chal lenge i s to li m it

t he s ess ion l ength, usual l y to an hour or l es s ,

t hat engages, but does not i nconveni ence, t heparti cipant In s el ect ing i nt erview candidat es, a

s el ect ion mat ri x t hat s egm ents the market according to type-of-us e and cust om er source,

l ike the one in Tabl e 1, ens ures that a divers it y

of cus tomers is cont act ed (B urchi ll andHepner-B rodie 1997, Hepner-B rodie 2000) The key concept is a represent ati ve rather than

a random s ample in which the P D t eam gat hers

i nform at ion from all the rel evant s egm ent s andfrom cus tom ers wit h varyi ng perspect ives oncurrent and fut ure needs In addit i on, if thereare m ult ipl e decis ion m akers , say doct ors , lab

t echni ci ans , and pat i ents for a m edi cal

i ns trument , t hen each t ype of decis i on m akerneeds to be cons ul ted S ee examples i n Hauser(1993)

Table 1: Customer Selection Matrix for Coffee Makers

Market S egment C ur rent

M ul ti ple m embers of the P D t eam s hould

review t he trans cript s For exam pl e, Gri ffin

and Haus er (1993) suggest that each team

m em ber recognizes approxi m at el y hal f t he

needs in a trans cript and that mult i pl e team

m em bers are very effect ive at ident i fying m ore

t han 95% of t he needs Because non-verbal

com municat i on i s cri t ical , m any firm s now

videot ape int erviews in addi ti on to trans cribi ng

t hem S uch i nt erviews, often dis tri buted on

C Ds t o t eam m em bers, have becom e known as

t he “F ace of the C us t om er ” For exam pl e,

heari ng a cus tom er s ay “I us e Windows on my

not ebook and need an accurat e, buil t -i n

poi nt i ng devi ce that does n’t require m e to move

m y hands from t he keyboard” carri es more

i nform at ion t han a fi lt ered sum mary of “Goodpoi nt i ng devi ce is i m port ant ” S eei ng t he us er

s truggle wi th exis ti ng poi nt ing devi ces is even

m ore persuasi ve Many fi rms now include theact ual des i gn-engi neers i n t he intervi ewi ngproces s when the proces s is cos t-effecti ve (c f ,Leonard-Barton, Wi ls on, and Doyle 1993) However, i n com plex products where the P D

t eam is oft en over 400 engineers and otherprofes si onals , key m embers obs erve the

i nt erviews and use m ethods , such as the bas ed Face of t he Cus tomer, to carry t hi s

video-i nform at video-ion t o the P D t eam video-i n a form t hat video-i t video-is

used effect ivel y

The Kano Model: Delighting Customers

“Cust omer needs ” are often verbal

s tatem ents of benefi t s that cus tomers gai n from

t he product or servi ce For exam pl e, a

cus tom er m i ght want a s afer car, a com put er

m onit or that takes up l es s s pace on the des k, or

a port able computer that makes si x hours on an

airpl ane m ore pl easant However, i n order to

des ign a product , the P D team mus t map t hes e

needs into product feat ures One wi dely-us ed

m et hod i s the Kano m odel

The Kano m odel characteri zes product

features accordi ng t o t hei r rel at ionship to

cus tom er expect ati ons (Cl aus ing 1994) as in

F igure 6 Som e features addres s “must have”

needs Such needs are us ual ly met by current

t echnology and any new product must sati s fy

t hese needs However, i t is di ffi cul t to

differenti ate a product by i ncreasi ng the

s at is facti on of thes e needs because they are

already sat is fi ed wel l by the com pet it ive s et of

exi st i ng product s In other words , the

com pet it ive equi li bri um has di ctated t hat all

viabl e products address t hes e needs If the PD

t eam does not m eet t he needs t hen t he product

wil l eli ci t cus t om er di ss ati sfact ion and lose

s al es For exam pl e, an automobil e mus t havefour properly i nfl at ed ti res t hat do not comeapart on t he road Ford’s recent problem s wi t h

F ires t one tires suggest t hey di d not addres s

t hese “m us t have” needs However, there areopport unit i es t o s ave cos t i f new creati ve

t echnology can addres s these needs as wel l orbet ter wit h l ower cos t For exam pl e, Hauser(1993) gives an exam ple where a bas i c need for

m edical ins trum ent s – pri nti ng pati ent records –was m et wi t h a new t echnol ogy (paral lel port toconnect to the physi cian’s offi ce printer) thatwas s i gnifi cant l y les s expensi ve that exi st ing

t echnology (bui l t-in therm al printers), yet m et

t he need bett er

Other needs are “m ore t he bett er”

(somet im es call ed li near sat is fiers ) When new

t echnology or i m proved ideas i ncreas e theamount by whi ch thes e needs are s at i sfied,cus tom er s ati sfact ion i ncreases , but us ual lywit h dim ini shing ret urns S uch needs areusual l y rel evant when t echnology is advanci ngrapidl y, s uch as i s the case wi th t he speed of acom put er proces s or In order to st ay on top of

t he m arket , a comput er manufact urer must always be devel opi ng more powerful and easi er

t o us e com put ers

Figure 6: Kano Taxonomy of Customer Needs

Feature Level

Satisfaction Level

Feature Level

Satisfaction Level

Feature Level

Satisfaction Level

MUST HAVE MORE the BETTER DELIGHTER

F inal l y, a speci al cl as s of needs are those

of whi ch cust om ers have di fficult y art iculati ng

or rarel y expect t o have ful fi l led When

features are included i n a product to sat is fy

s uch cus tom er needs, often unexpect edl y,

cus tom ers experi ence “del i ght!” S ources of

cus tom er deli ght can becom e st rong mot ivators

for i nit ial purchase and for cust om er

s at is facti on aft er t he sal e Exampl es i ncl udecom pl ement ary frui t bas ket s in hotel room s,

s oftware t hat anti ci pat es your next move, aut om obi les t hat rarely need s ervice, and others

s uch as those i n Tabl e 2

Table 2: Examples of Kano Feature Types

Must H ave Mor e the B etter D elighter

C ar (4) R eli abl e Ti res Gas Mi leage Free Loaner

N otebook A C Adapter H ard D ri ve Capacity B ui lt-in D V D Pl ayer

S oftware C ompatibil i ty P rocessi ng Speed A uto-Fil l-In

It is important to remember that the

Kano model is dynamic Today’s “delighter”

features become tomorrows “must have”

features For example, a graphical user

interface (GUI) and multi-processing were

once “delighter” features, but today they are

“must have” features for any desktop

computer operating system However, the

basic underlying customer needs of an

effective and easy to use operating system

remain Antilock braking systems and

premium sound systems were once delighter

features for high-end cars, but today are “must

have” features for any brand competing in the

high-end segment New “delighter” features

include automatic mapping and location

systems, satellite-based emergency road

service, side-view mirrors than dim

automatically, night-vision warning systems,

and Internet access However, the basic

underlying needs of safety and comfortable

transportation remain Really successful

products are frequently due to newly identified

“delighter” features that address those basic

customer needs in innovative ways The

dynamic nature of Kano’s model suggests a

need for ongoing measurement of customer

expectations over a product’s lifecycle

The Innovator’s Dilemma and Disruptive Technologies

The Kano model cautions us that asnew technology develops, today’s excitementneeds can become tomorrow’s must-haveneeds However, customers also evolve astechnology evolves For example, as morecomputer users purchase laptop computersrather than desktop computers, their needs forreduced size and weight increase dramatically

If a disk-drive manufacturer is focused only

on desktop computers, it may missopportunities when customers move to laptopcomputers

Bower and Christensen (1995) andChristensen (1998) formalize this concept andpoint out that listening to one’s current

customers, while consistent with the firm’sfinancial goals, may enable entrants with newtechnologies to eventually displace

incumbents This may happen even thoughthe new technologies are not initially aseffective as the incumbent technologies on

“more-the-better” features For example, theinitial small hard drives for portables were not

as fast nor could they store as muchinformation as the larger disk drives used indesktop computers But an emerging class ofportable users demanded them because theywere smaller and lighter in weight

Eventually, the smaller drives caught up on

the “more-the-better” features (storage, speed)

and dominated the market because they won

on the “delighter” features (size and weight)

Firms fall into disruptive-technology

traps because current customers may not

appreciate the new benefits of a new

technology because it does not perform as

well on the traditional attributes they value

However, new users, not well known to the

firm, may value the new attributes more

highly and forgive the below-average

performance on traditional attributes In

addition, new entrants may be willing to settle

for lower sales and profits than incumbents in

order to gain a foothold in the market

Eventually, as the new technology achieves

higher performance on traditional attributes,

the incumbent’s old customers begin

switching and the firm loses its leadership

position To avoid the disruptive-technology

trap and to stay on top of the needs of all

customers, Christensen proposes that

incumbent firms partner with (or develop)

independent, “entrepreneurial” entities to

explore disruptive technologies Christensen

proposes that the entrepreneurial entities be

held to less stringent short-term financial and

performance objectives so that they might

focus on long-term performance by satisfying

the “delighter” needs of the new and growing

markets

Empathic Design and User

Observation

M any firms real i ze t hat no m at t er how

refined the res earch methodology and no mat ter

how m uch data i s col l ected, som e ins ight s can

onl y be gai ned by obs ervi ng cus tomers in thei r

nat ural habit at (Leonard Barton, Wi l son, and

Doyle 1993) This i s part icul arl y true when

cus tom er needs are di fficult t o verbal ize or are

not obvi ous The technique of em pat hi c des ign

requi res t hat m embers of the desi gn team

i mm ers e thems el ves i n t he cust omer

environm ent for a period long enough t o abs orb

t he problem s and feel ings experienced by us ers

If a product is inconveni ent , ineffi ci ent , or

i nadequate, t he desi gner gai ns fi rs t -handexperi ence wi th the probl em Empat hic

m et hods are part icul arl y effect ive atdet erm ining t he ergonom ic as pects of aproduct The em pathi c met hods can be carri edout by m em ber of t he PD t eam or m arket ingprofes si onals , but i n eit her case, rich media

s houl d be used to capture the users experience

s o that it can be shared wit h the enti re PD t eam Int ui t , makers of Qui cken, t he leadi ngpersonal fi nanci al s oft ware package on t he

m arket , pi oneered the “Fol low-M e-Hom e”program in which Int uit em pl oyees observepurchasers in t hei r hom es from the mom ent

t hey open the box to the tim e they haveQui cken funct ioning properly (C as e 1991) Usi ng em pat hi c des ign and us er observati on, Int ui t has st eadil y improved Quicken’s ease-of-use wi th feat ures such as auto fi ll -in of

accounts and payees, on-s creen checks andregis t ers that look like their paper counterpart s, and push butt ons t o aut om ate comm on tasks

M ore import antl y, Int ui t took res ponsi bi l it y for

t he enti re process of producing checks

i ncluding worki ng to im prove printers andpri nt er dri vers even thos e t hes e were made by

t hi rd part i es Em pat hi c des ign highli ght ed t hat cus tom ers were not buyi ng Intui t’s products even though t he probl em was not Int uit ’s

t echni cal res ponsi bi l it y Intui t recogni zed t hat i t could lose it ’s share of the s oft ware market if it did not sol ve t he pri nt er manufacturers’

s oftware and hardware problems Int ui t’s focus

on cus tomer needs has kept t he company on t op

of a highl y com pet it i ve, ever-changi ng

m arket pl ace

Hol tzblatt and Beyer (1993) devel oped a

t echni que known as cont ext ual inqui ry inwhi ch a mem ber of the des i gn t eam conduct s

an ext ensi ve int ervi ew wi t h a cus tom er at his orher s i te whil e the cust om er perform s real t as ks The i ntervi ewer can int errupt at any t im e t o askquest i ons about the why’s and how’s of t hecus tom er’s acti ons The res ul t s of thes e

Managing a Dispersed Product Development Process

Dahan and Hauser

contextual inqui ri es are shared wit h other

des ign t eam m em bers and s ynt hes ized us ing

affini ty di agram s (descri bed below) and work

flow chart s

Underlying Meanings and Values

In addit ion t o exploring cus tom ers’ st at ed

needs , P D teams often s eek t o unders tand the

uns tat ed or “unart iculated” needs One met hod

t o get at unart i culat ed needs is wi t h in-dept h

experi enti al int ervi ews t hat s eek t o get

cus tom ers to express such needs (In ot her

words , unarti cul at ed needs are real l y jus t

diffi cul t-t o-art icul ate needs ) In addi t ion, some

firms have expl ored ant hropological methods

C ul tural anthropol ogy (cf Levi n 1992) i s

t he s t udy of hi dden meani ngs underl ying

products , or meani ngs whi ch are s ought , but

l eft unm et The approach is broader t han

psychology-based m ot i vati onal res earch i n t hat

i t account s for cust omers ’ s oci al values , not just

emoti onal needs Is s ues such a com pany’s

environm ent al i m pact and minori ty hi ri ng

record gai n s ignificance

How does cult ural ant hropology affect

product des ign? The key is consi st ency wit h

t he s oci al si gni fi cance of t he product For

exampl e, i f cus t om ers buy zero em is s ion

elect ric vehi cl es because of t hei r concern about

t he envi ronment , t hey m ay object to a des ign

wit h bat teries whi ch produce t oxi c was te

Zal tm an’s (1997) M et aphor El ici tati on

Techni que (ZM ET) s ugges ts that the

underl yi ng values and m eanings , whi ch dri ve

cus tom ers towards speci fi c product choice

decis i ons, may be uncovered through a process

of vi s ual sel f-expres si on ZM ET requi res

parti cipant s to provi de pi ct ures and i mages t hat

capture what they seek in the product cat egory

B ecaus e ZM ET al l ows the research st i muli to

be control l ed by t he respondent s, t hey can

expres s their feel ings, product m eanings and

att it udes

Kansei Analysis and the Mind of the Market

A s el ect group of product s , es pecial ly

“hi gh-touch” consumer durabl es such as aut om obi les and pers onal inform at ionappli ances , are purchas ed as m uch for theemoti onal res ponses they evoke as for thefunct i on t hey provide F or such product s ,

m easuring cus tom ers’ true feel i ngs towardpot ent ial des igns, es peci all y their look and feel,

m ay prove inval uable

Kansei anal ys is may be des cribed as the

“Li e Det ect or” of cus tomer res earch

m et hodol ogi es Most techniques of lis tening to

t he cust om er as s um e that res pondent s provideans wers that accurat ely refl ect t hei r preferencesand percept ions But for vari ous reas ons s uch

as social press ure, vanit y, or even inaccurat e

s el f-percepti on, furt her probi ng is neces sary Kansei anal ys is seeks t hes e true preferences by

m easuring non-verbal respons es to product

s ti mul i, m uch i n t he same way that gal vanic

s ki n res ponse, voi ce st res s, and breat hi ng rat eare recorded in li e det ect or t est ing Exam pl es

of ot her non-verbal res ponses that can be

m easured are facial mus cl e cont ract i ons andeye m ovement and dil ati on By meas uri ng

t hese subt l e physi ol ogi cal res ponses whi l e acus tom er vi ews or int eract s wi t h a new product ,

t he P D t eam gauges t he cus tomer’s feel ingsand at ti tudes A gri mace duri ng sharp s t eeri ng

m ight indi cat e poor res ponse i n a car, whil evis ual focus on a parti cul ar coffee makerprotot ype might reveal a preference for theout ward appearance of t hat des i gn Bycorrel at ing t he non-verbal reacti ons ofcus tom ers wit h the s pecifi c st i muli that produced t hos e react i ons, cust omer preferences for a product ’s “l ook and feel ” can be

det erm ined Si m il arl y, by obs erving det ail edcli ck st ream dat a, s oft ware and web si tedes igners can opti mi ze the user i nt erface for

m axim um cus tomer s at i sfact ion

Kos sl yn, Zalt man, Thompson, Hurvi tz, and B raun (1999) des cribe a m ethod that

del ves even deeper i nto t he physi ol ogi cal

aspect s of cust omer res ponse m echani sm s, a

m et hod t hey t erm , “The Mi nd of the Market ”

Thi s work uti li zes brai n imagi ng of respondent s

viewi ng marketi ng st i muli , i n thi s cas e

aut om obi le deal ers hi p s cenarios , to as ses s

negat i ve and pos it ive reacti ons B y com paring

t heir resul ts t o t hos e of anot her research team

uti li zing more conventi onal market res earch

m et hods, t he accuracy of the brai n imagi ng

approach was val idat ed, at l eas t di recti onall y

Benefit Chains

B enefi t chains focus on w hy cust omers

have a part icul ar need that is not yet addres s ed

by exi st ing products For exam pl e, whil e a

focus group or Kano analys is m i ght det erm ine

t hat cus tom ers want small er, l i ghter-wei ght

not ebook comput ers t hat perform fas t er, the

underl yi ng values dri vi ng thos e needs may not

be so obvi ous Are cus tom ers so am bit ious that

t hey want to accom pl i sh t wice the am ount ofwork (notebook performance) and workeverywhere they go (l ight wei ght )? Or could i t

be that cus tomers seek more lei sure ti me (i e ,

l es s tim e worki ng), and prefer to do t hei r workout si de of the office? The underl yi ng val uesdri vi ng those needs might di ffer dramati cal lyand t he di fference i n underl yi ng val ues might

i mply di fferent product -development solut ions The workaholi c not ebook comput er us er mi ght need more features and bat tery li fe whil e t he

l ei sure-seeker might need ease-of-l earni ng and

l ow-price F igure 7 il lus trat es a benefi t chainfor coffee makers Here, the user’s work ethi c

l eads to a desi re for eit her a di gi t al t i mer wit haut o-s hutoff (or anot her sol ut i on s uch as Int ernet cont rol ) that hel ps t he us er sat is fy hi s

or her cul t ural work-et hi c val ues

Figure 7: Benefit Chain Structure for a Coffee Maker

Cultural, Social

and Personal

VALUES

Desired BENEFITS CONSEQUENCE

Product Feature Products

CATEGORIZED

by Valued Attributes

Customer NEEDS Produces a for Particular

AttributesThe Benefit Chain: How VALUES lead to NEEDS

m or ning”

“ I would like

m y coffee

r eady when I awake”

Five coffee

m aker s featur e digital tim er s

“ I want the

m odel with digital tim er and auto-

s hutoff.”

R el at ed met hods incl ude a Means -End

C hain model of cus tom er choi ce behavior

Gut man (1982) and a val ue-syst ems model

(Rokeach 1973) Thes e aut hors vi ew cust omer

needs as a chai n reacti on begi nni ng wi th the

cul tural , social , and pers onal values hel d by the

i ndivi dual The underl yi ng val ues hel d by

cus tom ers then gui de thei r choi ces toward

products t hat produce des i red benefi ts Since

t here are num erous choi ces for a gi ven product ,

peopl e cat egori ze them int o set s or cl as s es ,

t hereby si m pl ifying the deci si on Thecat egori es creat ed by each cus t om er are

i nfluenced by hi s or her val ues Whil e the

l ei sure seeker may categorize not ebookcom put ers bas ed on price, the workahol ic mayconsi der m achines grouped accordi ng toperformance

Managing a Dispersed Product Development Process

Dahan and Hauser

Gut man and Reynolds’ (1979) technique

for m eas uri ng s uch benefi t chai n begins wit h

Kel ly’s (1955) repert ory gri d techni que Aft er

res pondent s have drawn di s ti nct ions wi thi n a

s et of t hree product s (by determi ni ng

s im il ari ti es bet ween two products and

differences wit h a t hird), t hey are as ked whi ch

att ri but e they prefer They are then as ked why

t hey prefer t hat att ribut e at higher and hi gher

l evel s unt i l som e core val ues are reached This

t echni que is som et im es referred t o as ladderi ng

It is il lus trat ed in Fi gure 8

Figure 8: Laddering Example for a

Notebook Computer

Why Ask Why?: The Laddering Technique

Feel Better

Notebook Computer Needs to

weigh less than three pounds

Easier to take Places

Do Hard Things

Less Taxing

Get Less Tired

Use More Often

Feel Stronger Accomplish More

Why should your notebook computer be lightweight?

Focusing the Design Team by

Identifying Strategic Customer Needs

Traditional surveys and interviews,

experiential interviews, Kano analysis,

disruptive-technology analysis, empathic

design, the study of meanings, Kansei

analysis, and benefit chains all identify

customer needs and desires which, if fulfilled,

lead to successful new products However,

these methods are sometimes too effective

producing not just a few needs, but rather

hundreds of customer needs Even for a

simple product such as a coffee maker, it is

not uncommon to generate a list of 100-200

customer needs For complex products such

as copiers and automobiles, such lengthy listsmight be generated for subsystems (interior,exterior, drive train, electronics, climatecontrol) But the needs are not allindependent

To proceed further in idea generation,the PD team needs focus This focus isprovided by recognizing that the needs can begrouped into strategic, tactical, and detailedneeds If we call the raw output of the variousneeds-generation methods “detailed needs,”then we often find that the 100-200 detailedneeds can be arranged into groups of 20-30tactical needs For example, detailedstatements by customers of a software packageabout the on-line help systems, “wizards,” on-line manuals, documentation, telephonesupport, and Internet support might all begrouped together as a need by the customer

“to get help easily and effectively when I needit.” The detailed needs help the PD teamcreate technology and other solutions toaddress the tactical need However, thetactical need is sufficiently general so that the

PD team might develop totally new ways ofmeeting that need such as communities ofpractice within large customers The tacticalneeds might also be grouped into 5-10strategic needs such as “easy to use,” “doesthe job well,” “easy to learn,” etc Thestrategic needs help the team develop conceptsthat stretch the product space and open up newpositioning strategies

Later in the PD process (Figure 3) the

PD team needs to decide on which strategicneed to focus or which features best fulfill astrategic need In later sections we reviewmethods to prioritize these needs (andfeatures) However, in the fuzzy front end it ismore important that we get the grouping right,that is, it is more important that the rightstrategic and tactical groups be identified.This is because, at this stage of the process,the PD team wants to generate a larger number

of potential product concepts, each of which

might stress one or a few strategic needs The

PD team also wants to explore new ideas to

address these strategic needs by solving the

relevant tactical needs in new and creative

ways In this chapter we review the two most

common methods of grouping needs

Team-based Needs-Grouping Methods:

Affinity Diagrams and K-J analysis

The J apanes e ant hropologi s t Ji ro Kawakit a

(denot ed K-J by the Japanese cust om of l ast

nam e first ) developed a m ethod of s ynt hes izing

l arge am ounts of dat a, includi ng voi ce of t he

cus tom er data, int o manageable chunks bas ed

on themes that emerged from the dat a

t hems elves (M izuno 1988) The K-J met hod

uses a t eam approach to develop af fi ni t y

diagr ams in which each voi ce-of-t he-cus tomer

s tatem ent is grouped wi th ot her s im i lar

s tatem ents The K-J technique requi res an

open mind from each parti cipant , encourages

creat i vi ty, and avoi ds cri ti ci s m of “s trange”

i deas The K-J method cl aim s to be based on

s ti mul at ing t he ri ght -brai n creat ive and

emoti onal centers of thought rather than relyi ng

on pure cause-and-effect logic

Typicall y, each data el em ent , preferably in

t he original language of the cust om er, i s

recorded onto a card or P ost -i t  note The

cards are wel l shuffl ed t o eli m inat e any

pre-exi st i ng order bias and are then grouped based

on feeli ngs rat her t han l ogi c The im press ion

or im age gi ven by each cus tomer s tat em ent

s ugges ts t he group t o whi ch that card has t he

great est affi ni t y rat her than any pre-conceived

cat egory When a few cards are grouped, they

are l abeled wit h a descri pti on that capt ures the

ess ence of thei r m eaning C ard groups are then

ass em bled int o a l arger di agram wit h

rel at i onshi ps between t he groups of cards

i ndicated The end res ul t i s a diagram showi ng

t he t op fi ve to ten cus tom er needs, relat ions hips

bet ween needs , and detail ed cus tomer voi ce

dat a expres si ng thos e needs

Customer-based Needs-Grouping Methods: the Voice of the Customer

Whi le affi ni ty di agram s and K-J analys is m ethods have proven t o be powerful i n

m any appli cat ions, t hey can al s o suffer when

t he t eam i s t oo em bedded in it s corporat ecul ture For exam pl e, Gri ffin and Hauser(1993) com pared affi nit y diagrams developed

by PD team s wit h affi ni ty di agram s devel oped

by act ual cus tom ers In bot h cas es , t he team

m em bers grouped cust omer needs the way t hefirm normal ly buil ds the product Cus tom ers

i ns tead grouped the needs by t he way t hey use

t he product Griffi n and Haus er al s o appli edhierarchical cl ust eri ng (Green, C arm one, and

F ox 1969, Rao and Kat z 1971) t o needsgat hered from a larger sam pl e of cus tomers Here each cus tom er does a relat ivel y s im ple

s ort of needs i nto a sm al l num ber of pil es Thehierarchy of st rat egi c, t act ical, and det ai ledneeds comes from t he st at i st ical analysi s This

m et hod, cal led bot h the Voice of the C us t om erand Vocalys t, has proven effect ive in li t eral l yhundreds of appl icat i ons Alt hough we know

of no head-to-head compari son cus tom affini ty and voi ce-of-t he-cust omer met hods, cus tom er-based met hods seem to provi deusabl e s truct ures than team-bas ed m ethods and

er-t hi s deer-t ai l l eads er-to more creaer-t ive sol uer-t i ons

New Web-based Methods for the Fuzzy Front End

Information pump The methods

reviewed above provide a breadth of means toidentify customer needs, whether they arearticulated or unarticulated, individual-specific

or bound in the culture, verbal or non-verbal,etc Recently, the Internet has made itpossible for groups of customers tocommunicate directly and iteratively with oneanother and, together, produce a set of needsthat might not have been identified any otherway The “Information Pump” is a novelmethod of objectively evaluating the qualityand consistency of respondents’ comments, inwhich “virtual focus group” participants opine

Managing a Dispersed Product Development Process

Dahan and Hauser

on a common stimulus such as a new product

concept (Prelec 2000) The quality of the

participants’ comments is judged by an

impartial expert and by the other respondents

Two keys to the effectiveness of the

Information Pump method are the use of

real-time Bayesian updating and the

implementation of an incentive-compatible

reward system for the respondents A benefit

of using the Information pump methodology is

that, in addition to collecting objective ratings

of individual opinions and comments about a

new product concept, one gains insight into

the overall quality of each respondent

“Listening in” to customers on the

Internet The Internet also provides the means

to identify customer needs by passively

observing customer purchase behavior on a

web site By organizing the web site by

agendas based on features or customer needs,

a virtual engineer can listen in and observe

how customers process attributes and, in

particular, when they search for attributes,

features, or needs that cannot be satisfied by

any extant product Urban (2000)

demonstrates this indirect method of capturing

unmet customer needs by observing customer

interactions with an Internet-based sales

recommendation system for trucks While the

virtual salesperson attempts to identify the

ideal, current-model truck for each

respondent, a virtual design engineer notes

which product attributes leave the customer

the most unsatisfied The virtual engineer

then “interviews” the customer to better

understand the unmet needs and how to best

resolve the inherent tradeoffs that prevent

those needs from being met

Ideation Based on Customer Needs

(and Other Inputs)

Once the PD team has identified and

grouped customer needs it must generate ideas

on how to address those needs (Goldenberg,

Lehmann, and Mazursky 1999) In the next

section (on designing and engineering

concepts) we discuss formal methods such asQFD by which the PD team can systematicallygenerate effective concepts But not allconcepts can be generated systematically.Sometimes the PD team needs crazy andbizarre solutions which, when refined, solvethe customers’ needs in new and creativeways A wide variety of ideation methodshave been proposed including brainstorming(Arnold 1962), morphological analysis (Ayres1969), group sessions (Prince 1970), forcedrelationships (Osborn 1963), systemsapproaches (Campbell 1985), variedperspectives (De Bono 1995), archivalanalysis (Altschuler 1985, 1996), andinventive templates (Goldenberg, Mazursky,and Solomon 1999a, 1999b) In this chapter

we review the three most recent proposals andrefer the reader to the references for the moretraditional ideation methods

Overcoming Mental Blocks

De Bono (1995) outlines a method ofovercoming the mental blocks most of us havethat derive from our particular approaches toproblem solving Figure 9 depicts De Bono’ssix hats, representing the diverse perspectives

of potential members of a product designteam Typically, each participant in a newproduct debate feels most comfortablewearing one or two of the hats, frequentlyleading to conflict The “six hats” exercisesrequire team members to “wear the otherguy’s hats” so as to improve communicationsand foster creative exchange For example,one might ask members of the design team toreact to a novel situation such as, “A pill isinvented that makes people dislike the taste offatty foods,” from the perspective of each ofthe six hats By identifying the types ofthinking each team member engages in,participants gain insight into their ownproblem solving approaches as well as those

of others

Figure 9: De Bono’s Six Hats Method

TRIZ (Theory of Inventive Problem

Solving)

Altschuler (1985, 1996) developed a

technique for generating creative solutions to

technical problems by harnessing archival

knowledge, an early version of knowledge

management Specifically, Altschuler

reviewed tens of thousands of patents and

noticed that their genius was in applying

inventive principles to resolve tradeoffs

between a limited set of “competing” physical

properties (approximately 40 in number)

These solutions typically resulted in no

tradeoff being made at all, for example the

way aluminum cans are both lightweight and

strong by virtue of their cylindrical design

Altschuler organized the patents according to

the fundamental tradeoffs they resolved, and

created tables so that future designers could

apply the inventive principles to similar

problems More recently, others have

advanced Altschuler’s work into other

domains of science and technology

Marketing’s role in applying a method such as

TRIZ is to represent the customer’s voice in

comparing the multiple technical alternatives

generated

Inventive Templates

Goldenberg, Mazursky, and

Solomon (1999a, 1999b, 1999c) extend

Altschuler’s methods to propose thatideation is more effective when the PD teamfocuses on five templates – well-definedschemes that are derived from an historicalanalysis of new products The authorsdefine a template as a systematic changebetween an existing solution and a newsolution and provide a method by which the

PD team can make these changes in a series

of smaller steps called “operators:”

exclusion, inclusion, unlinking, linking,splitting, and joining For example, the

“attribute dependency” template operates onexisting solutions by first applying theinclusion and then the linking operators.The authors give an example of how a newcar concept was developed by creating adependency between color and the location

of a car’s parts Specifically, Volkswagen’s

“Polo Harlequin” features differentlycolored parts and has become quite popular

in Europe even though it was initiallyintended as an April Fools’ joke, Othertemplates include component control(inclusion and linking), replacement(splitting, excluding, including, and joining),displacement (splitting, excluding, andunlinking), and division (splitting andlinking)

Summary of Methods for the Fuzzy Front-end

The fuzzy front-end of the PDprocess is the least well defined, but,perhaps, the most important phase of theprocess Without good customer input andcreative ideas, the process is doomed fromthe start Customers just will not buyproducts that do not satisfy their needs It ishard to succeed in PD unless the PD teamhas ideas which help them to create newways to satisfy those needs Thus, it is notsurprising that there has been significantresearch to propose and test many differentways to identify customer needs and

Managing a Dispersed Product Development Process

Dahan and Hauser

generate creative ideas In this section we

have tried to review the most common

methods that are relevant to a marketing

audience They are rich and varied; each

has its own strengths and none should be

used alone For example, if PD team uses

just the Kano model it could become overly

focused on the product’s technological

features and miss underlying psychological

or social needs On the other hand, a pure

focus on the mind of the customer could

cause the team to miss the obvious solutions

that will ultimately dominate the market

Good practice suggests that the PD team

consider a variety of approaches to

customer-need identification and use them

in parallel If the final output is subjected to

a rigorous needs-grouping method such as

affinity diagrams or customer sorts, then the

PD team will be able to assure that the ideas

it creates solve one or more strategic

customer needs

Once the PD team knows the

strategic needs, it needs some ideas Once

again there are a variety of methods Our

own experience suggests that different teams

will be comfortable with different

approaches Some teams prefer the formal

systems approaches, others need the wilder

approaches that “take a vacation from the

problem,” and still others prefer to just work

alone The organization (see enterprise

strategy below) must be conducive to these

myriad approaches While such creativity is

lauded by most PD teams, the organizational

challenges and frustrations of dealing with

truly creative people frequently preempt the

benefits (Staw 1995) However, if idea

generation is successful, the teams will

suggest large numbers of initial ideas that

can later be systematically engineered into

effective concepts, prototypes, and products

We now turn to more systematic methods by

which these ideas are shaped into concepts

and products

Designing and Engineering Concepts and Products

R et urning to the P D funnel at the cent er of

t he P D proces s in Fi gure 3, we see t hat t he

m any ideas creat ed i n opport uni ty i denti ficat i onare funnel ed to a sm all er set of concept s t hat are wi nnowed st i ll furt her t o a viable s et ofproducts or plat form s In t hi s s ect ion weaddres s concept generat ion and the des ign andengineering proces ses t hat develop these vi abl eproducts We begi n wit h met hods such as leaduser analys is , Kai zen and Teian anal ys is , s et -bas ed desi gn, and Pugh concept selecti on Each of these m ethods bui l ds on t he ideat ionand cust om er-needs unders t andi ng of the fuzzyfront end

Lead Users

S om et i mes the best i deas com e fromout si de the firm and, i n parti cul ar, from cus tom ers thems elves In some categories t heaverage cus tomer can recogni ze and appreciatenew s oluti ons t o t hei r bas ic needs and i n othercat egori es it i s m ore difficul t M ore

i mport antl y, PD team s are often embedded in

t heir corporate cult ure and vi ew PD through t he

l ens of their current products For exam pl e, com put er s oft ware des igners oft en focus oncurrent markets rather than em erging m arket s Older readers m i ght rem em ber when

M ul ti M at e and WordSt ar dom inat ed the wordproces si ng market These packages were bas ed

on the older Wang word process ors t hat wereused mai nl y by office s taff whose prim aryfunct i on was word process i ng When wordproces si ng moved t o the m ainst ream, ot herpackages , such as WordP erfect, recogni zed

t hose us ers who word process ed as t hey wrot e

Of cours e, WordP erfect it s el f did not move fas tenough t o Windows-bas ed product ivit y s ui t es and, in turn, l ost t he market to Mi crosoft’sWord As wit h other di srupt ive t echnologies,

t he evol ut i on of word proces si ng software

s ugges ts t hat fi rm s need to act ivel y s ol i ci t input from forward-looki ng cust omers

Lead user analys is recogni zes that a s mal l

s et of t oday’s users face tomorrow’s problems

These us ers oft en face new problems and oft en

devel op their own sol ut ions In som e cas es

t hese us ers are a very speci al i zed market , but i n

m any cas es they anti cipat e t he needs of the

l arger m arket For exam pl e, aut om obi le

m anufact urers foll ow NASC AR racing

carefull y becaus e the raci ng t eam s face new

chall enges and oft en invent new s ol uti ons t hat

can l ater apply to a more general m arket

C om put er proj ect ion sys tem s manufact urers

m onit or hi gh-end uses , such as NASA’s fl i ght

s im ul ators , becaus e as technol ogy advances the

probl ems faced by si m ul at or us ers wi ll s ugges t

s ol ut i ons for broader m arket s such as vi deo

gam ers

Von Hi ppel (1984) suggest s t hat s om e of

t he best s ources of ins ight int o us er needs and

pot ent ial product prototypes are these “l ead

users , ” cus tomers whose s t rong product needs

i n the pres ent fores hadow the needs of t he

general marketpl ace in the fut ure Thes e

peopl e are “ahead of thei r t im e” and badl y need

good ans wers now t o sol ve speci fi c probl ems

t hat mos t peopl e won’t experience for som e

t im e to com e Von Hi ppel describes how to

i dent i fy l ead us ers, and then how t o i ncorporate

t heir insi ght s int o the product des i gn proces s i n

a five-s tep proces s:

1 Ident i fy a new market t rend or product

opport unit y (e g great er computer

portabil it y, zero em i ss ion vehi cl es , etc )

2 Define m eas ures of potent i al benefi t as

t hey rel at e t o cus tom er needs

3 S el ect “lead us ers ” who are “ahead of their

t im e” and who wi ll benefi t t he most from a

good sol ut i on (e.g power us ers )

4 Ext ract informat ion from the “l ead users ”

about thei r needs and pot ent ial s ol uti ons

and generat e product concept s that embed

t hese solut ions

5 Tes t the concept s wi t h the broader market

t o forecas t t he im pl i cati ons of l ead userneeds as t hey appl y to the m arket i ngeneral

Urban and von Hi ppel (1988) appli ed this

t echni que to com puter-aided des ign (CAD)

s ys tem s Alt hough t he convent i onal wi sdom of

t he C AD devel opers was that the s ys t em s were

m uch too complex for us ers t o modify, Urbanand von Hi ppel found that lead us ers who faceddiffi cul t probl ems had not onl y m odi fi ed thei r

s ys tem s, but had generated s ignificant

i mprovem ent s For exam pl e, des igners ofcom pl ex, i ntegrated circui ts developed 3-dim ens ional C AD syst ems t hat coul d deal wit hcurved s urfaces , m ul t iple layers, and non-

s urface-mount ed component s When 3-D CAD

s oftware packages were devel oped bas ed on

t hese lead-us ers ’ sol ut ions, t hey were hi ghlyrat ed by t he more general market

Employee Feedback: Kaizen and Teian

Another source of ins ight into ways inwhi ch to addres s cus t om er needs bet t er i s t hecom pany’s own work force In his writ ings on

K ai zen, t he Japanes e concept of cont i nuous

i mprovem ent , Mas aaki (1986) explains t hat each employee i s res ponsi ble for bot h

m ai nt aining t he st at us quo and dest roying i t Thi s refers t o the noti on that em pl oyees must fol low cert ai n standards, but als o eli mi nat ewas te and contri bute to i nnovat ion One way inwhi ch em pl oyees can contri bute is by m aki ngfrequent s ugges t ions on product and process

i mprovem ent s through a sys tem the J apanes e

cal l T ei an See K ai zen T ei an 1 by t he Japan

Hum an Relat ions As sociati on (1992) Ofcours e, the s cope of such an em pl oyee

s ugges ti on syst em covers more than jus tcus tom er needs, but the es sence of conti nuous

i mprovem ent i s meeti ng cus tomer needs moreeffect ivel y

Managing a Dispersed Product Development Process

Dahan and Hauser

Set-based design and Modularity

In addit ion t o get ti ng ideas from l ead us ers

and from t he product i on em pl oyees , the P D

t eam can purs ue syst emati c m et hods such as

s et -based des ign This m ethod generat es

m ul ti ple desi gn opti ons by breaki ng a product

i nt o small er subsyst ems , standardizi ng t he

i nt erfaces between t hos e subsys tems , and

generati ng one or more des ign opt ions for each

key s ubs ys t em s Given int erchangeabil it y

bet ween the s ubs ys tem s, m ult ipl e des ign

s ol ut i ons becom e avai labl e t o the fi rm , lim it ed

onl y by the num ber of com binat i ons of

s ubsys tems that are feasi ble

Ward, Li ker and Sobek (1996), Sobek,

Ward, and Liker (1999), and Li ker, Ward, and

C ri st i ano (1996) des cri be a set -bas ed des ign

proces s in which t he freezing of the final

choice of subsys tems is delayed unt i l the

product is cl os er to launch The fi rm can then

check the pul se of a dynam ic m arket in order to

opt im i ze t he fi nal choi ce of m odular des i gns,

t hereby exploit i ng t he fl exi bi l it y inherent i n t he

s et -based approach Baldwin and Cl ark (2000)

further charact eri ze fl exi bi li t y due t o product

and proces s m odulari t y as form s of real opt ions

and demons t rate the pot ent iall y high val ue of

hol di ng such opt ions

Pugh Concept Selection

When mass customization is not

prevalent in an industry, the firm must narrow

from a broad array of possible design

solutions to a few critical solutions

(sometimes just one) Pugh (1996) develops a

method of winnowing multiple new product

concepts which he terms “controlled

convergence.” In essence, Pugh suggests that

each member of the design team

independently generate conceptual solutions to

the design problem The competing ideas are

then compared to a standard datum, selected

for its typicality in the product category, and

are evaluated as being better than, equal to, or

inferior to the datum on the key dimensions

that will contribute to product success Thegroup proceeds to eliminate weaker ideas, butalso attempts to cull the advantages of eachconcept and incorporate it into the remainingones before discarding it In this way, the

“winning” concept incorporates many of thebest ideas of all of the other concepts

Marketing’s role in this process is to identifythe key customer criteria on which conceptswill be based and to ensure that each concept

is evaluated with customer preferences inmind

Using inputs from the ideationprocesses, lead-user analysis, set-baseddesign, and Pugh concept selection the PDteam outputs a smaller set of high-potentialproduct concepts Following the PD funnel,the PD team then focuses on these conceptsand develops each to their greatest potential.This means linking engineering solutions tocustomer needs and vice versa

Value Engineering

F rom a cus t om er’s poi nt -of-view, aproduct consi st s of a bundle of feat ures andbenefi ts result i ng from i t s us e, whi le from t hefirm’s pers pect i ve, the product cons is ts of abundl e of parts and proces s that res ul t in it s

m anufact ure At t he point when cos t andfeasi bil it y t radeoffs are made by t he fi rm, i t i s

i mport ant to connect the cus tom er and fi rmperspect ives Ulrich and Eppi nger (2000)des cri be one met hod of doi ng s o known as val ue engi neeri ng, which rel at es the i mport ancecus tom ers place on each functi on perform ed by

a product to the cos t of the part s contri buti ng to

t hat funct i on A key pri nci pl e underl yi ng val ueengineering i s that the m arginal cos t of eachpart of a product shoul d be in proport ion t o its

m argi nal cont ri but ion t o cus tom er value To

i mplem ent val ue engi neeri ng the t eam m us t know (1) t he val ue pl aced by cust om ers oneach funct i on and (2) t he cost of t he parts and

m anufact uri ng t o provide that funct i on Weaddres s (1) bel ow as it i s m os t rel evant to t he

m arket ing audience For great er det ai l on (2)

s ee Ul ri ch and Eppinger (2000)

Quality Function Deployment and the

House of Quality

Val ue engi neeri ng requi res t hat we link

cus tom er needs to product solut ions so t hat t he

P D team can m ake i nt ell igent t radeoffs and,

perhaps, fi nd creati ve sol ut ions that do not

requi re tradeoffs Quali t y Funct ion

Deployment (QFD) and it s more-recent

progeny (Tess ler and Kl ei n 1993, McGrath

1996, Sm it h and Reinert sen 1998) provi de the

m eans to m ake t his l i nkage QF D i ts elf i s a s et

of process es that li nk cus tomer needs al l t he

way t hrough t o producti on requi rements

Alt hough t he ful l QF D proces s is som et im es

used, most notably i n J apan, i t i s the fi rs t

m at ri x of QFD, cal led t he Hous e of Quali t y(HOQ), t hat i s used mos t oft en The dri vingforce behi nd the HOQ is t he short , accurate,rel evant l i st of key cust omer needs ident ified i n

t he fuzzy front -end and s t ruct ured int o

s trat egi c, tact i cal, and det ai l ed needs In theHOQ, these needs are relat ed t o productatt ri but es which are then eval uat ed as t o howwel l they meet cus tom er needs P roductatt ri but es are “benchmarked” agai ns t com pet it ors ’ features i n their abil i ty t o m eet cus tom er needs and t he HOQ i s used tocom pare the benchm arking on features t obenchm arki ng on cust omer needs Fi nal ly the

t ot al product i s eval uated by the abil it y of its features t o m eet cus t om er needs m ore

effect ivel y and at l ower cos ts than competi ti veproducts

Figure 10: The House of Quality

The House of Quality

How do product features affect each other?

How well does each competitor meet each need?

How well does each feature meet each need?

Cost of each feature

Target level of each feature

How do competitors rate

The HOQ provides and organizes the

information that the PD team needs to refine

each concept It has proven effective in a

variety of applications including consumer

frequently purchased goods, consumer

durables, consumer services,

business-to-business products, and business-to-business-to-business-to-business

services A further advantage of the HOQ and

related techniques is that it enhances

communication among PD team members

(Griffin and Hauser 1992) This is becoming

even more important as PD teams become

more dispersed and global

Marketing’s primary input to the HOQ

includes identifying customer needs (fuzzy

front end), measuring how products fulfill

those needs (e.g., Green, Tull, and Albaum

1988, Churchill 1998, Moore and Pessemier

1993, Urban and Hauser 1993), and

understanding the tradeoffs among customer

needs and among potential product features

Ultimately, the HOQ method translates

customer priorities, as captured by a

prioritized list of needs, into

engineering/design priorities by identifying

those product features that contribute the most

to satisfying customers better than competitive

offerings

Tradeoffs Among Needs and Features:

Conjoint Analysis

Aft er cust omer needs are ident i fi ed and

grouped, after cri ti cal feat ures are i denti fi ed

and l i nked to cust om er needs , and after high

pot ent ial concepts are devel oped, t he PD

t eam’s next s tep i s to focus on t hos e features

and concept s that are m os t l ikely t o i mprove

cus tom er s ati sfact ion and lead to profit abl e

products Devel oping m et hods to measure

s uch tradeoffs among cust omer needs and/ or

features i s , arguabl y, one of the m ost s t udied

probl ems i n m arket ing res earch We have been

abl e to ident ify alm ost 150 art icles publ is hed i n

t he t op marketi ng journal s on conjoi nt anal ys i s

i n the l as t t wenty years In thi s secti on we

review s om e of the basi c ideas See als o

reviews by Green (1984), Green and Sri ni vas an(1978, 1990) Als o, because t hey cont inue to

be us ed by PD t eam s, we i ncl ude i n our revi ew

s el f-expli cat ed methods s uch as t hos e reviewed

“good for the envi ronment , ” and

“inexpensi ve ” The team now want s toevaluate a seri es of product concept s, each ofwhi ch st ret ches one of the five s trategi ccus tom er needs Conj oint analys is , appli ed tocus tom er needs, is t he general method to

m easure the cus t om ers ’ tradeoffs am ong t hos eneeds By ident ifyi ng and quanti fyi ng t he

t radeoffs, perhaps by cus t om er segm ent ,conjoi nt anal ys i s hel ps t o focus the P D team on

t hose concept s that have the hi ghes t pot ent ial

C onjoi nt anal ys i s can als o be appli ed to product features ; for exam pl e, the m aker of an camera

m ight want to know how hi ghl y cus tom ersval ue such feat ures as 1-s tep vs 2-st ep pi ct ure

t aking, st yli ng covers, automat ic vs manualfocus i ng, and automat ic vs control l able

l ight i ng Conj oint analys is can tel l the P D teamwhi ch of t he features i s mos t highl y val ued (bywhi ch segm ent ) and can as s ociat e a wil li ngnes s

t o pay for thos e feat ures

Camera Example We begin by

i ll us t rati ng conjoint anal ys is wi th the mos tcom mon t ype of appli cat ion – providi ngpreferences wit h res pect to product s (or product concepts ) in whi ch t he experim ent er has varied

t he feat ures (or needs ful fi ll m ent) of t heproducts s yst em ati cal ly We t hen revi ew ot her

t ypes of conj oi nt analysi s and suggest newforms that are now feas ibl e wi t h st ate-of-t he-art

i nform at ion and comm uni cat ion technology

Suppose that we have identified a set

of features for a new camera from acombination of sources including experientialinterviews, empathic design, Kansei analysis,and the information Pump In general, this

feature list will be quite extensive, but for the

purpose of this chapter we will illustrate the

feature list with a reduced set of five features.1

We might conclude that customers have needs

that can be addressed through various levels of

the five product features in Figure 11

F or exam pl e, one product permut at ion

would cost $25, weigh 16oz., have automat ic

l ight cont rol , produce 3-i nch square pict ures ,

and require t he us er to focus manual ly In al l

t here are 4 x 4 x 2 x 3 x 2 = 192 perm ut ati ons ,

each of whi ch m i ght be a viabl e product In

pri nci pl e, we coul d ask a sampl e of cust omers

t o evaluat e each of the 192 pot enti al product s ,

but t his woul d be an extremely unwi eldy tas k

As the num ber of pot ent ial products increas es

t he t ask becomes qui t e burdens ome (Green,

C arrol l, and Gol dberg 1981, Green, Gol dberg,

and M ont em ayor 1981, Malhotra 1986) and the

quali t y of the dat a degrades (B at es on,

R ei bs t ei n, and Bouldi ng 1987, Huber, Wit t ink,

F iedl er, and Mi l ler 1993, Moore and Semenik

1988) We must al so be concerned wi th bi as es

t hat can result when the num ber of level s vari es

acros s att ribut es (Wi tt ink 1989), potent i al ly

drawi ng more at t enti on and i mport ance to thos e

att ri but es wi th more level s

Factorial designs It would be even

m ore cum bersome if we asked cus tomers to

evaluate product s that varied on the t wenty-t wo

cam era feat ures in foot not e 1 – if thes e were a

m ix of t wo-level and three-l evel at t ri but es t his

would yi el d alm ost 400 mi l li on potenti al

1

Some attributes might include price, picture

quality, picture delivery, opening of the camera,

removable cover or not, picture taking process (1 vs 2

steps), light selection (3 settings vs feedback),

disposable camera, camera with cartoon characters,

metallic vs plastic camera, battery type: AA vs AAA,

picture size, color vs black & white, chemical vs.

digital picture vs both at the same time, zoom lens

vs regular, holster for the camera, picture has a

sticky backing, the film is decorated, the film

contains some advertising to reduce cost, panoramic

pictures, waterproof camera, manual control over the

picture, picture cutter included in the camera, etc.

products Ins tead we woul d l ike t o capture the

i nform at ion t hat cus t om ers woul d provi deabout tradeoffs am ong feat ures by as ki ng eachcus tom er t o eval uate a much sm all er number ofproducts For exampl e, for the five att ribut es in

F igure 1 1 we m ay not need t o ask eachcus tom er t o eval uate al l 192 feat urecom bi nat ions Ins tead, we coul d us e a m oreeffici ent experi ment al des ign known as

f ract i onal fact ori al desi gn If we as sum e that

all of t he at tri butes are independent we can

s im pl i fy t he num ber of com bi nat ions st il l further by us ing a s pecial fracti onal factori al

known as an ort hogonal ar ray For exampl e,

we mi ght us e one ort hogonal des ign cal led a

“hyper-greco-lat in-s quare” des i gn and as k eachcus tom er t o eval uate just 16 careful ly chos enproducts (A si m il ar experim ent al approach, known as Taguchi [1987] m ethods , is us ed indes cri bi ng reli abi li t y tes ti ng and stati s ti cal

s am pl i ng.) The actual det ai ls of a part i cularfract i onal fact ori al desi gn, t hat i s t he speci fi c

l evel s of pri ce, wei ght , light cont rol , picture

s ize, and focus i ng for the 16 pot ent ial products ,can be det erm ined us i ng l i st ings com pi led byAddel m an (1962) and Hahn and S hapiro(1966) or by us i ng comput er program s produced by S AS , S ys t at , Sawtooth S oft ware,

B rett on-Cl ark S oft ware, and ot hers

Respondents’ tasks The t ask by whi ch

cus tom ers expres s their eval uat ions of product svaries S ee Cat ti n and Wi tt ink (1989) for a

s urvey of indus t ry practi ce By far t he most com mon t as k i s to si m pl y ask t he res pondent s

t o rank order t he product profi les in terms ofpreference For exam pl e, each respondent

m ight order a s et of cards according t o his orher preferences for (or l i keli hood of buying) the

products depi ct ed Each card, known as a f ul

l-profi l e, des cri bes a product cons is ti ng of

differing level s of the key at t ri but es Ot her

t as ks incl ude as ki ng the res pondent to eval uat epai rs of profil es (S rinivasan and S hocker 1973,

B at es on, R eibst ein, and B oul di ng 1987) or

t radeoffs among at tri butes dis played t wo at a

t im e (Johns on 1974, Segal 1982) The tas k can

Managing a Dispersed Product Development Process

Dahan and Hauser

be rank order (Green and Wind 1975) or t he

cus tom er can provi de a scaled evaluati on

(Carm one, Green, and Jain 1978, Haus er and

S hugan 1980, Curri m, Weinberg, and Wit ti nk

1981, Leigh, MacKay, and Sum mers 1984,

S ri ni vas an and Park 1997) Al l form s of data

col lecti on appear to be reli abl e (B ateson,

R ei bs t ei n, and Bouldi ng 1987, Green and

S ri ni vas an 1990) and none seem to domi nat eeit her practi ce or t he academi c l it erature

Figure 11: Simplified Conjoint Attributes and Levels

P roduct Attribute A lter native Levels

P rice (P ) $15 $20 $25 $30Weight ( W) 16oz 20oz 32oz 64oz

Light contr ol ( L) A uto 1-s tep

P ictur e size (G ) pos tage stamp 3 inch s quare s tandard

F ocus ing ( O ) A uto manual

Part worth functions Once we have the

rat ings (or rankings ) for each product i n t he

experi ment al des ign, we repres ent t his

i nform at ion by a uti l it y funct i on, that is, a real

-val ued functi on of t he at t ri but e level s chosen

s uch that differences i n uti li t y represent

differences (or rank orders) i n preference

among the products If the att ribut es are

i ndependent , as is as sumed i n an ort hogonal

array, t hen t he ut il i ty of a product i s sim pl y t he

s um of t he uniat tributed uti li t ies of each of the

att ri but es If the att ri but es are speci fied by

dis crete levels as in Figure 11, then the uti li t y

of each of the level s of each of the feat ures is

cal led a part wort h That i s, the uti li t y of a

cam era t hat cos t s $25, wei ghs 16oz , has

aut om ati c light cont rol , produces 3-inch square

pictures , and requires the user t o focus

m anual ly woul d be equal t o t he part wort h of

$25, plus the part wort h of 16 oz., pl us the part

worth of automat ic l i ght control, pl us t he partworth of 3-inch square pi ctures , pl us the part worth of m anual focus (We can als o repres ent uti li t y as the product of the part worths – i t i sonl y separabi li t y that is im pl i ed by

i ndependence ) Part wort hs are i ll ust rat ed i n

F igure 12a However, if the att ri but es areconti nuous we m i ght als o repres ent the ut il it y

by a more conti nuous funct ion An ideal point

m odel for att ri but es where m ore i s not al ways bet ter (e g., pi ct ure s ize, Fi gure 12b) is one

s uch conti nuous funct ion and a vect or model where more is bett er (e.g , quali ty of pi ct ure,

F igure 12c) is anot her s uch cont inuousfunct i on Decreas ing functi ons (e g., price), concave funct ions (e g , clari t y of sound), andant i-i deal point (e g., t he tem perat ure of tea)

m odel s are al so poss i bl e S ee the dis cus si on in

P ekel m an and Sen (1979)

In som e cas es t he at t ri but es m i ght not be

i ndependent F or example, t he desi re of the

cus tom er for manual focus i ng m i ght depend

upon the qual it y of image that can be produced

wit h the camera’s lens and fil m In t hi s cas e,

conjoi nt anal ys es us e m ore com plex

experi ment al des igns that al low i nt eract i ons

and es ti mat e ut i li ty funct ions that cannot be

repres ented s im ply as t he sum (or product ) of

t he part wort hs See Green and Devi ta (1975),

Akaah and Korgaonkar (1983), and Johns on,

M eyer, and Ghos h (1989)

Estimation In order t o est i mate the uti li t y

or part worth functi ons we m us t decompos e the

overal l preference rati ng (or ranki ng) i nto t he

uti li t ies of the att ribut es (wi th or wit hout

i nt eract ions) Earl y appl icat i ons ass um ed that

onl y the rank-order inform at ion was relevant

and us ed m onotonic anal ys i s of vari ance

(monanova) to es ti mat e the part wort hs (Green

and S rinivasan 1978) Al t ernat ivel y, li near

programm ing provides accurat e est im ates

bas ed on a crit eri on of m ean absolut e or

direct ional errors (S ri ni vas an and Shocker

1973, Jain, Aci t o, M alhot ra, and Mahaj an

1979, Malhotra 1982) However, many

res earchers have dis covered that cus tomers can

provi de val id preference dat a whi ch has strong

m et ri c properti es (Huber 1975, Haus er and

S hugan 1980, Curri m, Weinberg, and Wit ti nk

1981) whil e other res earchers have found that

t reat i ng rank data as i f it were met ri c provi des

accurate es ti mat es (C armone, Green, and Jai n1978) Thus, m any appl icati ons use OLSregres si on or ot her met ri c m et hods If ris k is

i mport ant in the des i gn deci si on, t hen vonNeumann-Morgens t ern uti li t y es t im at i on can

be us ed by as ki ng res pondent s to evaluat eproduct profi les i n whi ch one or more of theatt ri but es is uncert ain (Hauser and Urban 1979,Eli as hberg and Hauser 1985, Farquhar 1977,Kahn and M eyer 1991)

Another com mon dat a col lecti onprocedure sim pl y pres ents the cus tom er wi th

s et s of al t ernat ive products profil es chosenfrom an experim ent al desi gn and asks t hecus tom er t o s el ect t he product he or s he prefers from each set of product profi l es This method, known as choi ce-based conj oi nt anal ysi s, us es aquant al choice model such as a logi t or probi t

m odel to es ti mat e the part wort hs from t hechoice dat a S ee El rod, Louvi ere, and Davy(1992), Carroll and Green (1995), Haai jer,Wedel , Vri ens , and Wans beek (1998), Oppewal, Louviere, and Ti m merm ans (1994), and Orme (1999)

Compositional methods C onjoi nt

analys is i s usuall y thought of as adecom pos it i onal technique in which partworths are es ti m at ed by as ki ng respondent s toevaluate potent i al product s (or reduced set s ofatt ri but es of t hos e products ) However, there i sals o a l ong t radit ion i n market ing of

Managing a Dispersed Product Development Process

Dahan and Hauser

com pos it ional m ethods i n whi ch the cus tom er

i s as ked t o s pecify direct ly t he im portance or

part worth of a need or feat ure Thes e met hods,

als o known as expect ancy-val ue or s

elf-expli cat ed methods , were reviewed by Wil kie

and P ess em i er i n 1973 In t hes e met hods , t he

cus tom er rates each product on each need

(rates ) and eval uates t he im portance of each

need (weights ) The ut il i ty of a product i s then

t he s um of the wei ght s ti m es t he rat es , sum med

over cus tom er needs M ore recent ly,

S ri ni vas an (1988), S rinivasan and Wyner

(1988), and B uckli n and S rinivasan (1991) use

a m et hod call ed Casem ap i n whi ch they have

m odifi ed s elf-expl icated met hods for part wort h

est im ati on Aft er an init i al conj uncti ve phas e

(described below), C asemap asks cus t om ers t o

s peci fy the i mport ance of each feat ure or need

and t he rel at ive val ue of each level of each

feature or need (rel ati ve to a base level of that

feature or need) The part worth i s t hen t he

product of thes e t wo values Such sel

f-expli cat ed methods have t he advantage of

bei ng a rel at ively easy t ask for the cus t om er –

for exam pl e, Cas em ap can be com pl et ed over

t he t elephone and does not require that a

s am pl e of cus tom ers com e to a central locat ion

C as em ap and other sel f-expli cat ed m ethods

have proven t o be accurat e and reli abl e

(Bates on, Rei bs t ei n, and Bouldi ng 1987, Akaah

and Korgaonkar 1983, Green and Hels en 1989,

Hoepfl and Huber 1970, Huber, Wit ti nk,

F iedl er, and Mi l ler 1993, Leigh, MacKay and

S um mers 1984)

C on ju n ct ive proces ses St andard

conjoi nt anal ys i s as ks res pondent s to evaluat e

products t hat vary on all level s of al l of the

features or needs But m any researchers have

hypot hes ized that not all level s of al l features or

needs are accept able to respondents For

exampl e, a cust omer who want s a cam era for a

weddi ng mi ght not accept a pos t age-s tamp si ze

picture; no l evels of t he ot her att ribut es wi l l

com pensate the cus tom er enough to m ake hi m

or her buy a weddi ng camera that produces

s uch small pi ct ures When att ribut es have such

m inim um acceptable l evels (at least a 3-i nch

s quare pict ure) the cus tom er i s s ai d t o fol low a

conjunct ive proces s (Ei nhorn 1971, Grether and

Wil de 1984) C asemap and ot her m et hods(Sawt oot h 1996) have modi fied conjoi nt analys is t o account for s uch conj uncti veproces ses by fi rst as ki ng the res pondent to

s peci fy those l evels of t he at t ri but es t hat areunaccept abl e Pri or speci fi cat ion of

unaccept abl e level s improves predict ion if donecarefull y However, Green, Kri eger, and

B ansal (1988) and Kl ein (1986) caut i on t hat i f

t he ques ti ons are not pre-test ed and

i mplem ented carefull y, the res pondent wi l lfal sel y rej ect att ri but e level s t hat he or shewould have later accept ed

Reducing respondent burden Whi l e the

bas ic concept of conj oi nt anal ysi s is bot hpowerful and si m pl e, a key i mpl em ent at ionbarri er has been t he respondent s’ t ask

Alt hough we can reduce the five att ribut es in

F igure 11 to an ort hogonal array of 16 profi l es ,

t hi s is not always t he cas e For exam pl e, Wi nd, Green, S hi ffl et , and Scarbrough (1989) report a

s ucces sful appl i cati on of conj oint analys is t o

t he desi gn of M arriot t’s Court yard Hot el s i nwhi ch there were 50 factors at a tot al of 167

l evel s C l earl y, we cannot as k res pondents t oevaluate even a reduced experi m ental des i gnfor s uch problem s As a res ul t , res earchers andpract i ti oners have focused on means to reduce

t he burden on respondents We have al readyreviewed t radeoff analysi s (e g., J ohnson 1974)

i n whi ch respondents compare t wo at t ri but es at

a t im e, choice-bas ed conj oint analys is i n whi chres pondent s choose am ong set s of profi les (e g.,

C arrol l and Green 1995), and t he el i mi nat ion ofunaccept abl e at t ri but e level s (e g , S ri nivas an1988) (M alhot ra [1986] als o propos es a

m et hod t hat s creens unaccept abl e profi les )There has als o been much effort all ocated t overy effici ent experi ment al des igns (e.g ,Kuhfel d, Tobi as , and Garrett 1994) Wereview here t hree ot her m ethods for reducingres pondent burden: (1) met hods that mi x

i ndivi dual and market -l evel dat a (hybridconjoi nt ), (2) met hods that em ploy mul ti -st agedat a col lecti on in which the fi rs t tas k is