PERSPECTIVE Third-Generation New Product Processes

J PROD INNOV MANAG
1994;11:3-14
3
0000
PERSPECTIVE
Third-Generation New Product Processes
Robert G. Cooper
New product processes-formal "stage-gate" systems for driving new product projects from idea
through to launch-have been widely adopted in the
last decade, and have generally had a strong and
positive impact on firms' new product efforts. While
these Second-Generation roadmaps represent a major
improvement over the NASA-based first generation
process of the 1960s, they too have weaknesses: too
time consuming and too many time wasters, too
bureaucratic, and no provision for focus. H ere, Robert
Cooper speculates about the nature of an emerging
next generation ofnew product processes.H e proposes
fundamental changes to today's "stage-gate" systems
thaI revolve around four Fs: they will be fluid and
adaptable; they will incorporate fuzzy gates which are
both situational and conditional; they will pro vide for
much sharper focus of resources and management of
the portfolio ofprojects; and they will be much more
flexible than today' s process. The end results should
pro vide companies with a much more efficient roadmap, bringing products to markerfaster and improving
their use of scarce resources.But pitfalls are never far
away in our evolution towards these Third-Generation
Processes.
lntroduction
Fonnal new product processes have had a profound
impact on way that some finns' new product programs
are managed, controlled and measured:
New Product Process: a formal blueprint,roadmap,
templateor thoughtprocessfor driving a newproduct
projectfrom the idea stagethrough to markerlaunch
and beyond.
Such processeshave also led to very positive results,
according to managers in these firms [7]. But are
today's new product processes appropriate modeIs to
lead us into the next century? Probably not. So now is
the time to be thinking ahead to how today's new
product process must evolve: what the next generation
process willlook like.
SomeBackground
Why all this talk and concemaboutprocess? You may
haveheardwordsto this effect:
"All work is a process.If you want better results at the
end -the output-then focus on the process that
delivered the results. Any process CaDmanaged to be
more effective."
Address con-espondence to Robert G. Cooper, Ph.D. Faculty of
Business, McMaster University, 1280 Main Street West, Hamilton, Ontario
L8S 4M4, Canada.
@ 1994ElsevierScienceInc.
655 Avenueof lhe Arnerica.,.NewYork. NY 10010
This apparentlyobvious and simple tenethas had
more impact on managementthoughtand practice in
the 1990sthan perhapsany other. Indeed,fifiy years
from today,. some book entitled the History ol
ManagementThought will probably record that the
1990swas a period of preoccupationwith process.
0737-6781/94/$7.00
4
J PRODINNOV MANAG
1994:11:3-14
R. G. COOPER
development: DuPont, 3M, HP, Procter & Gamble
(P&G), Northem Telecom, ICI-UK, mM, Dow ChemicaI, Polaroid, Black & Decker, Exxon Chemicals,
Coming and many others.
What are fonnal or stage-gatenew product systems?
Perhapsyou would recognize them better by Olle of the
many different names they are called: PDP (Product
Delivery Process),NPP (New Product Process),Gating
System, and Product Launch System. I prefer the name
stage-gate systems because it is generic: it describes
exactly what the systemis-a stage-and-gateprocess.
Stage-gate systems break product innovation into a
predetennined set of stages,each stage consisting of a
set of prescribed, cross-functional and parallel activities
(see top halt of Figure l). The entrance to each stageis
a gate: these gates control the process and serve as the
quality control and Go/Kill/Hold/Recycle check points.
Typically, there are about tour to six stages and gates.
Not a New Concept
New product development is no exception. While
most corporate attempts at "process improvement"
have been directed at operations and production-for
example, on the application of TQM methods to the
factory floor or to office procedures-a number of
leading companies have successfully implemented
formal new product processes aimed at improving the
effectiveness and efficiency of the product development cycle.
In the 1960s and 1970s, managers were alerted to
the fact that all was not well in new product
development. Beginning with early Conference Board
studies in 1964, we were warned that new products
surfer high failure rates, and that most of the causes of
new product failure were preventable [10,11,13]. The
Booz-Allen studies of the early 1980s revealed that
half our development resources go to new products
that are failures [I]. And claims that new products fail
at astronomical rates-as high as 90% failure ratesbeganto appear in the popular press (Merle Crawford
wrote a classic article setting the record straight on
some of these outrageous claims [8]).
A formal process for guiding product innovation is
ODe solution to correct to ~hat ails new product
programs. Many leading fInns have developed a
systematic stage-gate process~a r-oadmapfrom idea
to launch consisting of discrete stages, each stage
preceded by a Go/Kill decision point or gate. These
firms include recognized giants in the field of product
Today'sstage-gateprocesseshave their roots in much
earlier modeIs. lndeed OUTstage-gatenew product
systems-the Olleswe have becomefamiliar with in
the 1980sand early I 990s-are really second-generation modeis.
The first-generation scheme for product development was developed by NASA
in the I960s.
NASA's PPP(phasedproject planning), which today
is often called Phased Review Process, was an
elaborate and detailed scheme for working. with
contractorsand suppliers on various spaceprojects.
The U .S. military also adopted the approach for
weapons developmentwith its suppliers, and so a
number of corporationsended up using the Phased
ReviewProcess.
The Phased Review Process broke development
into discretephases.There were review points at the
end of each phase: runding for the next phasewas
conditional on the fact that certain prerequisiteshad
beenmet-typically that all taskshad beensatisfactorily completed for the previous phase. Thus the
methodwas morea measurement
and control methodology, designed to ensure that the project was
proceedingas it shouldand that every facet of it was
completed on time (if not on time, then at least
completed!). The process was very engineeringdriven, however: it applied strictly to the physical
designand developmentof the product (for example,
Marketingpeoplewere not part ofthe scheme),and it
c
THIRD-GENERATlON
NEW PRODUcr
PROCESSES
J PROD INNOV MANAG
5
1994;11:3-14
Build
Business
Case
Prellminary
Investlgation
8~
Test &
Development
c
--0
Gate 1
Validate
Gate 3
~~~
Gate 5
Today's stage-gate process.
Figure 1. A comparison of the secondgenerationStage-GateProcesswilli what
the Third-Generation Process might
look like.
Gale I
Gate 2
Gate 3
Gate 4
Gate 5
Tomorrow's Third Generation Process wilh overlapping, fluid slages and "IuZZY"ar condllional Go declsions
at gates.
.
was designed to deal solely with technical fisks (but
not business risks).
History has given mixed reviews to this firstgeneration process. There were SOfie positives, and
indeed SOfie units at Hewlett Packard still uge a
modified version of this Phased Review Process. The
systemdid bring discipline to an otherwise chaotic, ad
hoc activity; it did reduce technical risks; and it did
ensure completion of tasks. But the Phased Review
Processwas cumbersome: most notably, the laborious
check-offs of dozens of tasks at each review point; it
was slow: projects could be held up in a queue for a
managementreview, orworse yetbe put "on hold" at
a review point awaiting the completion of ODebehindschedule task [15]; it was too narrow: it only dealt
with the development phase (rather than the entire
process from idea to launch); and it was too
functional: it was narrowly focused on the technical or
engineering side of the project--on technical fisks
rather than on business risks.
Sofie pundits were dead against the process. ODe
executive exclaimed that the bureaucracy inherent in
the system meant that "NASA's Phased Review
Process managed to double the development time of
every project it was used on!" On the other band, that
generation of managers and engineers did manage to
put a man on the moon in less than a decade using the
system-a feat that we have yet to repeat. So the
PhasedReview Process could~nothave been all bad!
~
The Second-GenerationProcess
Today's stage-gatemodeIs resemble somewhatthe
Phased Review Process of the 1960s. They, too,
consist of identifiable and discrete stagespreceded by
review points or "gates" [2,7], but that is where the
similarities end. Over the last few decades, many
successfactors have been uncovered:factors that
separate successful projects and firms from less
successfulones; or practices that result in shorter times
to market. These lessons, gleaned from personal
experience or from researchinvestigations-for exampIe, studies of why products fail, or what distinguishes
winners from losers-have been built into Dur newer
second-generation stage-gate systems.
.First, the stage-gatesystemis very much crossfunctional.Note that the PhasedReviewProcess
was largely an engineeringmethodology. By
contrast, today, each stage involves activities
from many differentdepartmentsin the corporation. No stageis "owned" by anyane function:
for example,there is no "Marketing Stage" or
"Manufacturing Stage." Rather, at eachstage,
players from all functions-Marketing, R&D,
Engineering,Manufacturing,and so on-are on
the field togetherand are active players on the
project team. The nature of activities virtually
lorces the useol a cross-functionalproject team
approach,and also does much to reduce the
infIuenceand roadblockscreatedby functional
fiefdoms.
Marketing and Manufacturingare now integral
parts of the product development process.
Remember,the PhasedReview Processtended
to be solely an engineeringor technicalscheme,
and all but excludedMarketingandManufactur-
)
6
J PROD INNOV MANAG
R.G. COOPER
1994;11:3-14
ing people. By contrast, today's stage-gate
systemdemandssubstantialbusinessand marketing homework together with a thorough
manufacturingassessment
before a project receivesthe greenlight into Development.
Here's how it worked in Rohm and Haas's
Biocides Division before the introduction of
their stage-gateprocess.
Marketing/Sales initiated the project via a
specific new product request; often the product specifications (specs) were ill-defined,
however, and based on a superficial scoping of
market needs. The project was then "handed
off" to the R&D Department to begin "their
phase," Development, while Sales/Marketing
waited impatiently for a finished product.
With vague, often poorly defined product
specs and minimal Marketing/Sales involvement during this phase, the R&D effort
usually went in circles: the final product was
delivered months late and did not quite meet
customer' sneeds. Assigning the blame then
began, with R&D accused of being slow and
unresponsive, and Marketing/Sales of failing
to understand their marketplace. It was anything but a team or cross-functional effort!
By contrast, today, R&H's new stage-gate
processrequires a cross-functionaland team
effort, with R&D, Sales/Marketing,and Manufacturingpeople all run-time playerson the
projectteam.
The decision points or gates are also crossfunctional. In the old days, ane function-for
example,the Engineeringor Marketing Director-might have approvedthe project on their
own. While this meantthat the projectreceived
the neededsupportand resourcesfrom the ane
function, it did not ensure the buy in and
commitment of the other functions. Today's
stage-gatesystemseescross-functionaldecision
making,wherethe varioussenior managerswho
own the neededresourcessit togetherat a gate
meetingand togetherdecideon and committo a
project.
.Stage-gate systems are more holistic: they
capturethe entire processfrom idea through to
launch,and not just the middle stage,Development.
.Stage-gate
systems have built in much more
emphasis on up-front homework or pre-development work. In the Phased Review Process, this
homework was just assumed to have occurred.
By contrast, in both Exxon Chemicals' system
and the Procter & Gamble scheme,there are two
homework phasesto tighten up the front end of
the process befare Development even begins: a
preliminary scoping stage, and a more detailed
investigation stage. This is fairly typical. The
result is that the needed homework gets done: the
market research, competitive analysis, the concept tests, the manufacturing assessment,and the
business/financial analysis. These are vital activities, as many investigations into new product
success-and-failurehave shown, yet tao aften are
omitted altogether or handled in a poor fashion
[5,6].
.ODe result of this homework phase is a much
stronger marker orientation: the customer becomes an integral facet of the product development process, and the entire project focuses on
delighting the customer (as opposed to building
a monument to the scientist or engineer). A
second positive result is much sharper product
definition prior to the beginning of Development. This is what Crawford calls the protocol
for the project-an all-party agreement on the
product definition, benefits to be delivered,
positioning, and product requirements-an ingredient that we have found central to success
[5,6,9].
.Another stage-gateimprovementis the notion of
parallel or concurrentprocessing.While Phase
Review Processesdid not demand sequential
problemsolving, they surelyencouragedit. The
Developmentphasewas subdividedinto many
mini-phases,eachdone in sequence.The analogy is that of a relay race, with the batonbeing
passedfrom runnerto runner.
Today's new product modeIs, by comparison,
feature parallel processing [17]: activities are
undertaken concurrently (rather than sequentially) and so more activities are undertakenin an
elapsed period of time. Note that the play is
much more complex using a parallel play
lliIRD-GENERATION
NEW PRODUCT PROCES SES
J PROD INNOV MANAG
1994;11:3-14
scheme(versusa seriesor relay raceapproach),
hencethe need for a disciplined game plan or
template.
.Sharper
decision points with clear Go/Kill
criteria is yet another improvement of the
modem stage-gate system. As ODeexecutive at a
major auto maker exclaimed:
"In OUTold scheme, we only asked two
questions at the review points: 'is it on budget
and is it on time?' If the answers were 'yes,'
the project was given the go-ahead. It could
have been the worst project in the world from
a business standpoint, but it nevet got stopped
as long as it was on time and on budget!"
In short, the old PhasedReview Processwas more
a control mechanism: it made sure that the project
was unfolding as it should. But review points
rarely addressed the wisdom of continuing with
the project from a business perspective. Today's
stage-gate systems feature tough gates with
rigorous criteria and metrics--<:riteria that focus
on quantitative, financial measures, but also
measures of qualitative business issues, such as
sustainable or product advantage, synergy, and
market attractiveness.
The results of implementing this second-generation
or stage-gate approach to new products generally
appear to have been positive. Olle study of leading
firms, including 3M, IBM, Northem Telecom and
others, looked at 21 divisions that had stage-gatetype
systems in place [7]. Here, managers wholeheartedly
endorsedthe new product scheme, rating it most often
as having a "highly positive impact."1 In rank order,
specific improvements that the formal new product
processyielded were as folIows:
1. Much bettercross-functiona1
teamwork.
2. !..essrecyc1ingand rework-far less "going back
and doing it again."
3. Earlierdetectionof failures-they areeitherkil led
outright or stepsare takento avertdisaster.
4. Better launch-marketing planningis integral to
mostfirms' new productprocesses.
5. Shorter elapsed time due to better homework,
more multifunctional inputs, sharpermarket and
productdefinition, and lessrecycle.
Not a Panacea
No scheme is perfect. aften the problems lie not with
the scheme itself, but with how well or poorly it has
been implemented. Implementation deticiencies aside,
no system designed to improve product innovation is
without its inherent weaknesses.Some of the problems
with second-generation new product processesinclude
the following:2
ProjectsMust Wait at Each Gate Until All TasksHave
BeenCompleted
This is a control feature of stage-gate schemes to
ensure the satisfactory and quality execution of critical
tasks-an attempt to overcome the errors of omission
and commission that plague many flrms' new product
programs. In dealing with quality-of-execution problems, however, stage-gate methods ean slow down
project: for example, the project could be held up at a
gate-put "on hold"-for the sake of ODeactivity that
remains to be completed. This creates delays that caD
be most expensive. On the other band, the missing
activity may be a critical one that yields vital
information-for
example, the results of the field
trials-which is central to the decision to move the
project to the next stage.
Overlapping ol StagesIs All but Impossible
Stage-gate systems require the successful completion
of ODestage prior to embarking on the next. It is much
like some grade school systems: you must "pass" all
courses in grade 1 before getting into grade 2, and so
on through all the grades or stages. There are
exceptions, of course, but for the most part, overlapping of stages is discouraged by today's secondgeneration modeis.
Today's quest for speed, however, means that
overlapping of stages-for example, beginning the
final stage-Launch and Production Start-Up-before
the previous stage- Testing & Validation-is complete may be, in some cases,desirable. The analogy to
2 Some of these problems were inherited from the first-generation
I Opinions were sought on a 1-5 scale where 1 = no impact and 5 =
highly positive impact. The mean res!J<>nse
was 4.25 [6].
Phased Review
Process, thaI is, are common
second-generation pnJCesses.
to
both first-
and
8
J PROD INNOV MANAG
R. G. COOPER
1994;11:3-14
a schoolsystemis that you can beginyour senioryear
coursesbefore you have successfullycompletedyour
sophomoreyear: an acceptableroute perhaps,but not
for everystudent!
Projects Must Go Through All Gates and Stages
,.
I
I
S-omecompanies' systems are a littie rigid in that they
i~quire strict adherenceto the scheme.For example, all
projects most go through eachstageand eachgate. The
argument in favor of this discipline is that in the ..heat
ofbattle" and under pressure from top management, it
is all too easy to start skipping stages,steps,activities,
and gates in the interest of saving a littie time. It is
these very same omissions and shortcuts that invariably lead to product failures.
an the other band, while folIowing the stage-gate
process in a religious fashion may be appropriate for
large, higher-risk projects, this may not be true for all
projects. For example, for smaller projects, lower fisk
initiatives or close-to-home projects, this blind adherence to the system creates unnecessary work and
Some New Product ProcessesAre Spelled Out in Far
Too Much Detail
Sofie modeIs, designed willi the hest of intention s, are
simply too detailed. For example, ffiM's "red book"
new product scheme spelIs out every minute detail in
eleven 100se-Ieafbinders; by contrast, Polaroid's is
quite a manageable document, user-friendly and
easy-to-read. P&G's model is so simple that it is
displayed on a PC-computer, complete with pop-up
windows should the user want greater detail. ODe
problem with too much detail is that the voluminous
manual is rarely read and the system is never totally
understood; moreover the system invariably is perceived as "too bureaucratic," so that it never gets
universally accepted and implemented. A second and
more subtle deficiency is that detailed procedures get
followed blindly and rotely: too much direction takes
away the need for thinking, the result being that people
simply go through the motions-a "paint by numbers" approach.
delay.
SomeNew Product Processes Tend to be Bureaucratic
TheSystemDoes Not Lead to Project Prioritization
and Focus
Sadly some managerssee any systemas an opportunity
to impose more paperwork, more meetings, and more
red tape. Along with the positive comments, our
research also revealed these concems with regard to
Stage-gatesystemspay linIe attentionto the resource
bureaucracy [7]:
allocation question: they do not provide for the
prioritization of projects and the focusing of scarce
"(The new product process has made) a positive
resourceson these top-priority projects.While gates
contribution ...a littIe bureaucratic, but otherwise
introduce tough Go/Kill criteria which cull out bad
good."
projects,thesecriteria rend to be absolutestandards:
"It has created SOfietime-consuming steps, but the
"Does this project meet certain hurdles?" There is
overall effect is good."
linIe provision for a comparisonacrossprojects-to
look at the entire set of projects at eachgate in ordeT
The Third-Generation Process
to selectthe hest subsetor portfolio of projects. In
today's systems, what invariably happens is that What then is the solutionto thesesix deficienciesof
"Go" projects simply get addedto the list of active stage-gatesystems?The Third-GenerationNew ProdOlles,the result being too many projects and not
uct Processis alreadyevolvingfrom today'sstage-gate
enoughresourcesto do anyOlle projectwell!3
systems.Its particular emphasisis on efficiency: on
A relatedthomy issueis the difficulty in reconciling speeding up an already eifective second-generation
resourcedecisions made in real time (for example, stage-gateprocessand on moreefficient allocation o/
makinga Go decision on a project at a gate)versus developmentresources.
those made in calendar time (the annualbudgeting
Procter & Gamble first introduced a fonnal new
decision).
3 Although in some finns, priorities among projects are set, but often
outside the stage-gate system: either informally, via political means or
through an annual budgeting process.
product process in 1964, years ahead of most other
finns. It served the company well:. the 1960s and
1970s saw many noteworthy new products from P&G.
B Y the late 1970s, however, the process had fallen
from favor: it was perceived as too bureaucratic and
THIRD-GENERATION
NEW PRODUcr
PROCESSES
J PR()D INNOV MANAG
9
1994;11:3-14
cumbersome, and product development reverted to no
system at all. Then everything went wrong and Procter
last hundreds of millions of dollars on new product
disasters through the early-to-mid 1980s.
A thorough audit and review of past winners and
losers led to the conclusion that a formal new product
process was indeed the correct way to go. So a new
process--essentially a second-g~neration processwas designed and rolled out in the late 1980s. While
more streamlined than the 1960s version, sadly it tao
was far tao rigid and inflexible, and management
resisted its adoption.
It was only when P&G reintroduced in 1991 their
"Product Launch Model"-a
mave towards a ThirdGeneration Process4-that the organization bought in.
The Third-GenerationProcessrepresentsa precarious balance betweenthe need for thoroughnessof
action and complete infonnation versusthe need to
mave quickly; but it is still very much a systemand
still requiresdiscipline (that is, we are not retreating
back to a world of no process,no systemand ad hoc
decisions!).The Third Generationhas four fundamental Fs:
l. Fluidity-it is fluid and adaptable,with overlapping and fluid stagesfor greaterspeed.
2. Fuzzygates-it featuresconditionalGo decisions
(ratherthan absoluteanes), which are dependent
on the situation.
3. Focused-it builds in prioritization methodsthat
look at the entireportfolio of projects(ratherthan
ane projectat a time) andfocusesresourceson the
"best bets."
4. Flexible-it is not a rigid stage-and-gate
system:
each project is unique and has its own routing
throughthe process.
Consider eachof thesetour vital elementsin more
depth:
Fluid and Adaptable
To enable shorter cycle times, the Third-Generation
Process will be much more fluid and adaptable.
Activities will not be married to specific stagesas they
tend to be-.in today's approach" Rather, there will be
overlapping ol stages: some activities, normal ly done
in the next stage, will begin befo!e the current stage is
4 Note thai while P&G's new system has some of the elements of a
Third-Generation Process, it is not there quite yet. Further, the system is
quite new, and has yet to be ful ly tested.
completed;long 1ead-timeactivities might be brought
forward from ODestageto an earlier stage;and the
demarkationbetweenstageswill be morefluid.
The purchaseof productionequipmentor thedevelopment of sales force training modules are typica1ly
found in the final "Commercialization" stage. But
becausethey might takefar longerthanthe restof the
activities in this final stage, begin them partway
throughthe Testingstage.In this way, delayscan be
eliminatedand the time to marketreduced.
As Smith and Reinersten note, there should be a
continuous flow of activities that interlin~ overlap,
and allow for design to evolve. This activity overlapping relies heavily on the adept uge of partial
information and requires a close-knit team with lots of
face-to-face communication to be effective [16].
For example,P&G's ProductLaunch Moderis fluid
and adaptable: it allows bringing key activities
forward to a previousstageas well as overlappingof
stages:"it is acceptableto overlapthe beginningand
ends of succeedingphasesprovided that we understand the fisks involved and have agreed at the
previousdecisioncheckpoint."5
Several words of caution here: First, while certain
time-constraining activities caD be moved forward,
this does not apply to all activities nor does it apply all
the time (otherwise the notion of stages and gates, a
positive facet of the Second-Generation process, is
lost: note that stages are defined by the activities
within them!). Second, thesedeviations frpm the norm
should be made consciously at the gates and with full
recognition of the fisks involved. An example folIows:
Ordering productionequipmentbefore the productis
fully tested involves fisk: the productionroute may
yet change, pending confirmation of product test
results;or worse yet, the producttestsare so negative
that the project is cancelledaltogether.Look for an
"escapehatch" from decisions to move long lead
time activities forward. In this case of production
equipment,this escapehatchcould bethe inclusionof
a cancellation clause in the equipment purchase
contract,or provisionto usethe equipmentelsewhere
in the company.
.
5Quotationsource:ProductLaunch: Roadnzap
for Success,
Procter&
Gamble,Cincinnati,Ohio, October1991.
10
J PROD INNOV MANAG
1994;11:3-14
Third, make sure that activities are not moved too
far out of phase:you don't wantsituationswhereR&D
is down on the olle-yard line ready to score a
touchdown(stage5), while Marketingis still back on
their own thirty-yard line, and caD barely see the
goalposts(stagel)!
Fuzzy Gates
The entire new product process caD be viewed as a
series of steps,stages,and activities designed to gather
information neededfor the next decision point in order
to manage risk. Sometimes the information is missing,
however, resulting in a postponed decision and serious
delays at the gates. In order to minimize these time
delays, the Third-Generation Process will have
"fuzzy" gates. The term fuzzy is used, not in a
negative sense, but much in the way it is used in the
mathematics of "fuzzy sets"-where conditions are
not absolute (in traditional thinking, a switch is either
"off" or "on"), but caD assume varying states in
between. These fuzzy gates mean that Go decisions are
not necessaryabsolute or "solid green lights." Rather,
gates are conditional and situational. The tricky
balance is between timely decisions versus risk
management.
What does this fuzzy conditional/ situational feature
translate into?
.Conditional:
Projects will be given conditional
Go decisions in order to speed up the process.
This conditional Go is subject to a lask being
completed at a specified time in the future and
the information generated by that lask being
positive.
This type of conditional decision making is
contrary to today's absolute or black-and-white
schemes. To make a hard-and-fast Go/Kill
decision at today's gates means that all the
needed information must be on the table. By
contrast, a "Conditional Go" means that "on
the basis of the partial information available, the
project looks good. For flOWthe project is Go,
but the incomplete tasks must be finished and the
missing information delivered (and positive) by
XYZ date. Otherwise, the project will be halted
on XYZ date."
This conditional Go decision enablesthe project
to move ahead without 10ssof time. It also pulS
R. G. COOPER
pressure on the project leader and team to
completethe missingtasksin a timely fashion.
A word of warning, however: the pressures to
rationalize are considerable.For example, suppose
the Go decision is made, conditional upon a test
being completed and yielding specified results;
yet when completed several months into the next
stage, the results are marginally short of target.
Since so much progresshas already beenmade in
the next stage,the result is rationalized as "good
enough." This intense pressure to "move ahead
anyway" necessitates having a tough project
leader and a closely involved, disciplined gatekeeping group; otherwise the "Conditional Go"
becomesa perfunctory "Yes."
.Situational:
This conditional Go only applies to
some projects and to some (not all) missing
information and tasks. Note that certain key
"must-do" tasks and "must-have" information
must be completed and be available at the gate
review, If not, the decision is "hold," This begs
the question: which tasks and infonnation are
"must have," The answer is: "it is situationally
dependent-it depends on what the infonnation
is likely to be, the impact of both positive or
negative infonnation, and the cost of the delay."
An example foliows:
A new biocidal chemical product awaits the
"Go-to-Development" decision-Gate 3 in
Rohm & Haas's new product model. All
deliverablesbut one are available: results of
market studies, technical appraisals,patent
searches,manufacturingassessment,and financial analysis. ane task is incomplete,
however: a preliminary environmentaland
regulatory study. It will be another two
monthsbefore it is ready.
In today's model,the decisionwould be to wait the
two monthsand delayto Go-to-Development
decision.
In the Third-GenerationProcess,however, the decisionis not automatic;ratherit depends:management
weighs the odds of a negativeenvironmentaVregulatory assessment,
its impacton theproject,thefmancial
loss (two months of developmentwork) versus the
costsof beingtwo monthslaterto the market.In SOfie
cases,the decisionwill be to move aheadanyway-a
Conditional Go-conditional on a positive environmentalresult in two monthstime.
THIRD-GENERATION
NEW PRODUCT PROCESSES
J PROD INNOV MANAG
1994;11:3-14
Focused
Too many new product programs guffer from a lack of
focus: too few resources dissipated across too many
projects, many of questionable merit! Tomorrow's
new product process will look much more at the
totality of projects in the pipeline: it will emphasis the
portfolio decision rather than simply a decision on a
single project at each gate. This means that, not only
will a project be judged against a standard s~t of
criteria as it is today, but the issue of the "optimal set
of projects, given OUtlimited resources" will enter the
.The Strategic Decision Group method (SDG) is
a tinancially based, two-dimensional grid which
looks at the expected commerciai vaiue of the
project to the company and also its probabiiity of
success[12]. Projects are plotted on the grid and
are categorized into Olle of four quadrants:
pearls, oysters, bread-and-butter, and white elephants. The portfolio is reviewed at each gate
meeting, and prioritization decisions are made to
yield an appropriate portfolio or balance øf
projects.
Go/Kill/Hold decision. Thus the project will also be
compared to other projects in the pipe and against
company historic norms. Additionally, the real time
versus calendar time-resource commitment decisions
will be directly addressed. The result will be better
prioritization and sharper focus.
ODe solution will be the introduction of New
Product Information and Tracking Systemsinto the
gate meetings. Information systems are being developed today to track the progress of all projects and
their anticipated and combined resource requirements.6 The problem is made more complex by
recognizing that forecasts well into the future mus t be
made for these resource requirements (there's no senge
starting a project today if there'll be no resources next
year to finish it) and that there is a good likelihood that
some projects currently in the pipe may be cancelled
by next year (and thus unexpectedly free up resources).
.The
Arthur D. Littie approach argues that
tinancial or NPV methods used to generate
portfolio maps may be not only meaningless but
could be also harmful [14]. ADL considers a
number of keyquaiitative characteristics of each
project: tit with corporate strategy; inventive
merit and strategic importance; durability of the
competitive advantage; tinancial reward; competitive impact of technologies; probabilities of
success; and investment required. The ADL
method plots all projects on a number of
two-dimensional portfolio grids, with various
combinations of these qualitative characteristics
as the grid dimensions. (The number of plots
seems endless, however; but as firms gain
experience with the method, they willlikely find
which plots have the greatest power and value
for them, and concentrate on these few).
Developing an information system to forecast re-
Each model has its strengths and weaknesses;and
no doubt these portfolio modeIs, so badly needed,wiIl
be greatly improved in the Third-Generation Process.
A useful element of both these modeIs is the
consideration of two key characteristics of each
project: its value to the companyand its likelihood o/
success. The message it that it is vital to begin
estimating both elements early in a project' s li/e: ODe
side benefit of portfolio modeIs is that they force Ug,in
the very first stage of the project, to make initial, albeit
rough estimates of market size, possible sales, and the
potential for profit as weIl as to afrive at realistic
assessmentsof the odds of technical and commercial
success. Too often, both elements are conveniently
overlooked until too late in the project.
sources committed to known projects as well as
resource availability as a function of time is no easy
task, but it is essential: only in this way are decision
makers able to visualize what resources will be
available to "new" projects, and what the impact of
approving more projects will be on current and future
resourcesavailabilities.
A second and complementary solution is the use of
portfolio models that enable managers to see in a
moment the nature, composition, and expected impact
of projects already in the pipe. For example, two
portfolio approaches are proposed by two different
consulting flrmS:
..
4
6The notionof an infonnation system -thaI tracks and displays projects
is clearly essential for the large finn or division, where the number of
projects is great. It also applies to the smaller firm: here the number of
projects is more manageable, hence the "system" may be a manual one, but
there should be a tracking system.
At the Royal Bank of Canada,one of the rive largest
banks in North America, a computer-basednew
productportfolio model is being developed.It relies
on the NewProdmodel-a scoringschemethat looks
at qualitative factors and yields a likelihood of
12
R. G. COOPER
J PROD INNOV MANAG
1994;11:3-14
commercial success [4]-in combination with a
financial spreadsheet.which determinesthe commercial valueof the projectto the firm (net presentvalue).
Thesetwo scores-likelihood of successandvalue to
the firm-are the vertical and horizontal axes on a
computer-generated
displayof a "bubble diagram" or
portfolio map,to be usedin gate decisionmeetings.
Flexible
The final facet of the Third-Generation Process is
easierto visualize, and indeed is built into same firms'
schemeseven today: flexibility. Sadly, certain companies' new product models are quite rigid and inflexible: "Thou shalt do the folIowing steps and activities." More progressive firms build flexibility into
their new product processes.The fule here is that not
all gates must be passed through; nor are all stages
essential; nor are all prescribed activities mandatory.
For example, in P&G's process, flexibility is the key
word: "You and your initiative teams decide how the
Model is used to meet the specific needs of your
projects and business while adhering to its proven
principles. "7
If the new product process is essentially a riskmanagementprocess (noted above), then the risk level,
the uncertainty, and the need for information define
what steps need to be done and which caDbe left out.
As we evolve towards the Third-Generation Process,
we will see more flexibility:
.In
any project, stages,gates, and activities can be
omitted or bypassed. The system is a guide or
roadmap-not a bureaucratic book of rules-and
deviations or detours are made consciously and
deliberately, and with fu" awarenessof the facts,
consequences,
and risks. The decisionto skip
stages or activities is determined by the fisk and
the need for information, and, in part, dependson
what stages or activities have been done (or
omitted) upstreamand what results were achieved.
For example, if a concept test was omitted (or
gave so-so results), then customertestsduring the
Development phase using rapid or partial prototypes are likely to become essential.
The point is that decisions to skip over, delete, or
shift activities or gates are not ad hoc, arbitrary,
7Quotationsource:seefI)Otnote5 on page9.
or made for the wrong reasons;rather, they are
decidedthoughtfully, and agreedto by the team
and gatekeepersat the precedinggate.
.In
a more formal approach, managementdefines
two or three categories of projects, based on
project scope, investment, and fisk level. These
range from sales developments or product
modifications (relatively simple, short time
frame and low-risk projects) through to major
projects involving heavy expenditures and high
risks. And appropriate routes are decided for
each type of project, willi lower fisk projects
typically omitting some stages and gates (or
combining gates and stages).
Today's growing professionalism in the field on
new product managementpoints to a move towards a
more flexible new product process. A rigid structure
based on a hook of rules is most appropriate at the
beginning-when the organization has littIe experience in the effective management of product development. As ODeexecutive put it: "It's much like leaming
to drive an automobile. FolIowing a rigorous driver
instruction course, the inexperienced driver needsrules
and rigidly follows them-signals at every torn or lane
change, stays exactly within the speed limit. With
experience, the professional driver does some things
automaticaIly; and he learns when certain rules can be
broken without additional fisk in order to speed up the
driving." In the same way, the organization gains
experience in effective product development by relying on a fairly rigid new product process-the
second-generation process; as professionalism develops, it caD then move towards a more flexible
process.
A Fifth F: Fallibility
The font Fs of the Third-Generation Process have been
highlighted above. All are positive. The fifth F is a
possible negative consequence: failure or t'allibility!
This Third-Generation Process introduces much more
freedom and discretion to project leaders, teams, and
senior managers (who are the gatekeepersor decision
makers). With freedom and discretion, of course,
comes fisk: the odds of making mistakes go up. This
new process is more delicate, sophisticated, and
sensitive, thus requiring a more experienced, professional managementapproach. Sofie of the possible fail
points have been noted throughout the discussion
above (and in the implications below). Using the
.
THIRD-GENERATION
NEW PRODUCT PROCESSES
J PROD INNOV MANAG
1994;11:3-14
automobiledriver analogy, with more discretionover
the rules and increasedspeedcomesan increasedfisk
of disaster.
The Implications of the Third-Generation New
Product Process
What are the implications of this Third-Generation
Process? Today's second-generation systems have
created dramatic improvements in the effectivenessol
the process: for example, higher success rates, and
fewer foul-ups. Tomorrow' s process will provide more
efficiency-speeding up the process and better focus.
Here are some specific implications:
l. The system becomes a "smart system"-much
more tailored to the needs and specifics of each
project. But it is still very much a system.
2. Conditional, situational, and focused decisions
means that decision making will be much more
complex and sophisticated, and hence will require
more sophisticated and thoughtful decision makers (the gatekeepers). Indeed, there will be much
pressure on senior managers who own the resourcesto upgrade their decision-making abilities.
(Note that some of these same managers are
already having great difficulties coping with the
decision-making demands in today's secondgeneration processes.8)
3. Combine conditional/situational gate decisions
with a fluid and adaptable process and the result is
overlapping, fluid, and difficult-to-define stages.
That is, the next stage begins before the previous
Olle has ended; it becomes less obvious when Olle
stage has been completed or when the next Olle
starts; and it mayeven be difficult to determine
what stagethe project is in! Figure 1 presentswhat
fluid stages and conditional gates might look like
in contrast to today' s well-defined stages-andgates system.
A ShiIt in Authority
A final and much more subtle implication of the
built-in flexibility, adaptabili~, and conditionalityof
the Third-GenerationProcessis the shift and clouding
"For example, initial screening and the "Go to Devek>pment" decision
were found to be two ofthe worst handled facets ofthe new product process
[S].
of decision-making roles (and authority) between the
project team members and senior management. In
second-generation processes, these roles are very
clear: senior managementtends the gates, reviews the
projects in detail, makes the Go/Kill decisions, and
allocates the resources, at least for the more critical
gates. Meanwhile, the project team and leader handle
most decisions between the gates with regard to their
project (i.e., they are not "micro-managed"): they
work to make good on their deliverables to each gate
and to pass the gate criteria.
With more flexibility, however, decision-making
authority will shift away from senior managementand
more toward the team and leader (and invariably
become more clouded). This shift is consistent with the
concept of a self-managed team:
The team maps the project through the process
and recommends when detours from the process
are to be made, assessesthe fisk, and advisability
of these detours, and all but makes the decision
to deviate. Senior management, who is not as
close to the project as the team and leader are, is
most often left in the fole of a tacit approval.
.Similarly,
the team is expected to recommend
when stagesshould be overlapped, when activitie s from one stage should be brought forward
into a previous one, and assessand manage the
fisk of these overlaps. Once again, senior
managementasks the right questions, but is often
left in simply a sanctioning fole.
.Fuzzy
gates mean that the project team must
ofte" "make the call" as to when their project is
ready for a gate, even though not all activities
have been completed and some information is
missing. Additionally, the team becomesresponsible for seeing that these missing actions are
indeed completed folIowing the gate review.
The point is that the Third-Generation Process,
being more flexible and adaptive, is potentially a much
more complex process. Given that the project team is
more familiar with the details and intricacies of their
project than any group of senior managerscould ever
hope to be, senior managementmust increasingly rely
on the teamfor reasoned arguments and recommendations, and be prepared to accept these. This is not to
suggest that senior managementhas merely become a
rubber stamp to the needs and desires of the team:
senior management still owns the resources and is
14
J PROD INNOV MANAG
R. G. COOPBR
1994;11:3-14
ultimately responsible for the results achieved. But
there will be some shift in decision-makingauthority
to the team.
much more sophisticated and professional new product
managers! Today's growth in the professionalism of
new product management will surely accelerate the
evolution of the Third-Generation Process.
No SystemAt All
II
I :1
j,:
J.il
Taken to an extreme, would not this Third-Generation
PrQcessultimately lead to no system at all? We will
have come full circle-right back to the 1950s before
the Pirst-Generation system began.
Not at all. There is a big difference betweena system
willi flexibility, adaptability, conditionality, and fluidity and no system at all. No system at all is chaos-it
is like driving an automobile in new territory willi no
roadmap; but a system that is fluid, adaptable,
conditional, situational and flexible provides the
roadmap, where detours and deviations are possible,
and where they are clearly marked and consciously
decidedupon.
Conclusions
The design and implementation of a new product
process is undoubtedly ODe route toward a more
effective product innovation. Today's second-generation schemes, when properly implemented, have
yielded positive results and are certainly an order of
magnitude better than the Phased Review Process of
the l 960s. As the quest for efficiency and speed
continues, however, our processesmust adapt to these
new demands. Moreover, the trend to more professionalism in new product management means that SOfie
leading firms are now poised to move to a new,
higher-order process.
The Third-Generation Process is beginning to
evolve from today's systems: it will be flexible rather
than rigid; it will have fuzzy gates, permitting
conditional Go decisions that are situationally dependent (rather than today's black-and-white decisions and
the inevitable delays caused by waiting for "perfect
information "); it will be fluid and adaptable-with
stages overlapping each other; and it will be focused,
building in techniques that lead to project prioritization and sharper focus. The results should be
positive; but the implementation and operation of such
sophisticated and "smart" systems will also require
\&
The authorwould like to thank ProfessorThomasHustad,editorof
this journal, for his guidanceand assistancein the preparationof
this manuscript. Tom made many excellent suggestions for
improvement.Thanksarealso in orderfor Mr. Bob Davis,formerly
of Procter & Gamble, now an independentconsultant, who
provided manyinsights into a third-generationprocess.
References
1. BO<Jz-AlIen & Hamilton. New Product Management for the I 980s.
New York: Booz-AlIen & Hamilton Inc., 1982.
2. Cooper, R. G. Stage-gate systems: A new tool for managing new
products." Business Horizons, 33:3 (May-June 1990).
3. C(JOper,R. G. Winning at New Products: Accelerating the Process
from Idea to Launch. Cambridge, Mass: Addison-Wesley, 1993.
4. Cooper, R. G. The NewProd system: The induslry experience. Journal
of Product Innovation Management, 9: 113-127 (1992).
5. Cooper, R. G. and Kleinschmidt, E. J. New products: What separates
winners from losers. Journal of Product Innovation Management
4(3):169-184(1987).
6. CO<Jper,R. G. and Kleinschmidt, E. J. New Products: The Key Factors
in Success.Chicago: American Marketing Assoc., 1990 (monograph).
7. Cooper, R. G. and Kleinschmidt, E. J. New product processes at
leading industrial firms. Industrial Marketing Management 10(2): 137147 (May 1991).
8. Crawford, C. M. New product failure rates-facts
Research Management 9-13 (September 1979).
and fallacies.
9. Crawford, C. M. Protocol: New IDOl for pnJduct innovation. Journal of
Product Innovation Management 2:85-91 (1984).
10. Hopkins, D. S. New Product Winners and Losers. Conference Board
ReplJrt No. 773, 1980.
11. Hopkins, D. S. and Bailey, E. L. New product pressures. Conference
Board Record, No. 816-24 (1971).
12. Matheson, J. E., Menke M. M., and Derby, S. L. Improving the quality
of R&D decision: A synopsis of the SDG approach. Journal ofScience
Policy and Research Management (in Japanese), 4(4) (June 1989).
13. National Industrial Conference Board. Why new products fail. The
Conjerence Board Record. New York: NICB, 1964.
14. Roussel, P., Saad K. N., and Erickson, 1:J. Third Generation R&D,
Managing the Link to Corporate Strategy. Cambridge, MA: Harvard
Business School Press & Arthur D. LittIe Inc, 1991.
15. Smith P. G. and Reinertsen, D.G. Shortening the product development
cycle. Research & Technology Management, 44-49 (May-June 1992).
16. Smith P. G. and Reinertsen, D. G. Developingproducts
Time. New York: Van Nostrand Reinhold, 1991.
in Half the
17. Vttal, Bro. Speeding new ideas to market Fortune 62-66 (March
1987).