AMultimethodApproachforCreatingNew BusinessModels

Interfaces, � 2002 INFORMSVol. 32, No. 1, January–February 2002, pp. 20–34

0092-2102/02/3201/0020$05.001526-551X electronic ISSN

A Multimethod Approach for Creating NewBusiness Models: The General Motors

OnStar Project

Vince BarabbaGeneral Motors Corporation, Corporate Strategy and Knowledge Development, 400 Renaissance Center, P.O. Box 400,

Detroit, Michigan 48265Chet Huber • Fred Cooke

General Motors Corporation, OnStar Headquarters, 1400 Stephenson Highway, Troy, Michigan 48083Nick Pudar

General Motors Corporation, Corporate Strategy and Knowledge Development, 400 Renaissance Center, P.O. Box 400,Detroit, Michigan 48265

Jim SmithGeneral Motors Corporation, OnStar Headquarters, 1400 Stephenson Highway, Troy, Michigan 48083

Mark PaichDecisio, 320 West Cheyenne Road, Colorado Springs, Colorado 80906

[email protected] • [email protected] • [email protected] • [email protected]• [email protected] • [email protected]

We developed a multimethod modeling approach to evaluate strategic alternatives for GM’sOnStar communications system. We used dynamic modeling to address some decisions GMfaced in 1997, such as the company’s choice between incremental and aggressive marketingstrategies for OnStar. We used an integrated simulation model for analyzing the new tele-matics industry, consisting of six sectors: customer acquisition, customer choice, alliances,customer service, financial dynamics, and dealer behavior. The modeling effort had importantfinancial, organizational, and societal results. The OnStar business now has two million sub-scribers, an 80 percent market share of the emerging telematics market, and has been valuedat between $4 and $10 billion. The OnStar project set the stage for a broader GM initiative inservice businesses that ultimately could yield billions in incremental earnings.Most important,OnStar has saved many lives that otherwise would have been lost in vehicle accidents.(Industries: communications. Transportation: automotive.)

G eneral Motors (GM) assembled a project teamconsisting of the authors of this paper to develop

its OnStar telematics business. Telematics is the pro-vision of communications services to cars, includingcrash notification, navigation, Internet access, and traf-fic information. We used a multimethod modeling ap-proach to design the OnStar business model and toanalyze the fundamental strategic decisions GM facedin 1997. We explain the modeling process and some

specifics of the model that we used to analyze the stra-tegic choices, and we present the financial, organiza-tional, and social impacts the project created.OnStar is GM’s two-way vehicle communication sys-

tem that provides a variety of services that enhancesafety, security, entertainment, and productivity (Figure1). The vehicle communicates with either an automatedsystem, called the virtual advisor, or with a human ad-visor through a cell-phone connection. Two-way

BARABBA ET AL.The General Motors OnStar Project

InterfacesVol. 32, No. 1, January–February 2002 21

Figure 1: OnStar provides safety and security, Internet, and communications services.

communication enables safety and security services,such as crash notification, in which the call center isnotified if the vehicle crashes or the airbag is deployed.A built-in global positioning system (GPS) precisely lo-cates the vehicle and, if necessary, the call center dis-patches emergency services to the accident scene.Two-way communication facilitates a variety of

other services that provide information and enhancethe user’s productivity. For example, users can obtainreal-time traffic information through the virtual advi-sor in most major cities. In addition, they can accesscontent, such as theWall Street Journal, and have it readto them. Many other services have either been imple-mented or are currently in development.OnStar began in 1994 as a promising communica-

tions technology. A GM engineering group proposedProject Beacon to test the application of advanced com-munications technology in GM vehicles. To test the

concept, in 1996, GM made OnStar available as an op-tion on some Cadillac models. The services at that timewere limited to safety and security and a few otherfeatures, such as remote door unlocking. The OnStarsystem required complicated installation by the dealer,costing about $1,300. The high cost and hassle of in-stallation limited customer acceptance, but market re-search showed that buyers found the system extremelyvaluable and that the customer retention rate was veryhigh. Some senior GM executives believed that withthe appropriate strategy, OnStar could become an im-portant product.

Fundamental Decisions in 1997In 1997, GM faced fundamental strategic decisionswith respect to OnStar. First, GM had to decidewhether to view OnStar as a car feature or as a service


Interfaces22 Vol. 32, No. 1, January–February 2002

Figure 2: In 1997, GM faced fundamental strategic decisions for OnStar.

business. The default and safe strategy was to marketOnStar as a car feature that would improve vehiclesafety and security. GM had decades of experiencewith new car features and had well-developedmodelsfor pricing and packaging vehicle options.An alternate strategywas to viewOnStar as a service

business that could contribute greatly to GM’s profits(Figure 2). The OnStar business would collect monthlysubscription fees in exchange for a portfolio of services.The business would be responsible for acquiring, de-veloping, and retaining customers and would providecustomer service. In contrast with the vehicle businessin which GM interacts with customers infrequently,the service business would put customers in touchwith GM with every use of OnStar.GM would have to choose between an evolutionary

and a revolutionary strategy. If it decided to create anOnStar service business with the evolutionary option,GM would take a cautious approach to the telematicsmarket. In 1997, the telematics category barely existed,and no vehicle manufacturer had invested in it ag-gressively. GM faced almost complete uncertaintyabout technological approaches, major competitors,

and what competitive and complementary technolo-gies would emerge among the Internet, digital cellularservices, and hand-held devices, such as PDAs.A reasonable approach to such pervasive uncertainty

would be to postpone any major investments until thepicture became clearer. GM could run a portfolio ofsmall technological experiments to develop and pre-serve its options until the direction of the market be-came clearer. Once the situation was better defined, GMcould be a fast follower that could profit from the mis-takes of bleeding-edge competitors. In the Internet field,for example, the blood of failed first movers, such asWordPerfect, CPM, and VisiCalc, and various Internetcompanies is splattered all over the battlefield.Specifically, if GM took an overly aggressive ap-

proach, it might make large investments in vehicle-communications hardware and infrastructure andthen fail to recover the costs because of low customerdemand. A good example of the failure of an aggres-sive strategy in communications is Iridium’s launch ofsatellites that cost several billion dollars. The systemnever attracted many subscribers, and Motorola wasforced to write off most of its investment. On its face,



the conservative strategy is sensible, and indeed, bothFord and DCX have taken a cautious approach; neitherwill launch a telematics service until 2003.On the contrary, in some cases in the automobile in-

dustry, the go-slow approach has served firms badly.Auto companies missed the opportunity to be majorplayers in the cell-phone business even though 60 to 80percent of cellular minutes are consumed in cars andeven though some auto companies understood the po-tential for cell phones years before demand exploded.The cell-phone analogy is well known at GM. A secondexample of go-slow failure was Detroit’s sluggishnessin responding to consumer demand for small, energy-efficient, high-quality vehicles in the 1970s.GM’s second option was to choose the revolutionary

strategy of preempting themarket. In this strategy, GMwould move quickly to build a large installed base andgain the cost advantages and network externalities thatcome from being the earliest and biggest player. Net-work externalities result from a process throughwhich, as more people adopt a service, the service be-comes more valuable to both existing and potentialcustomers. There is evidence that the get-big-fast strat-egy can be very successful. The literature on first-mover advantages, while mixed, provides examples inwhich preemptive strikes have generated big payoffs(Gurumunrthy et al. 1995). In addition, economic anal-ysis demonstrates that properly managed network ef-fects can create a long-lived competitive advantage(Shapiro and Varian 1999).GM faced a strategic choice: should it follow an evo-

lutionary wait-and-see strategy or should it make arevolutionary move to develop and preempt the tele-matics market? The choice was difficult because themarket did not exist and data were scarce. In addition,GM hadminimal experience in subscriber service busi-nesses, so its senior management had limited intuitionabout which direction would be better. The preemp-tive strategy would require a huge investment; failureof the preemptive strategy would be costly. On theother hand, its success would bring large returns.

Modeling an Industry That DidNot ExistGM formed a project team (this paper’s authors) toconsider alternative strategies for OnStar. GM makes

important strategic decisions through the dialogue de-cision process (DDP). DDP consists of four stages toreaching agreement on decisions: framing, alterna-tives, analysis, and connection (Figure 3). At each step,the project team interacts with the decision board thatis responsible for actually making the decision andcommitting resources.Dynamic modeling can be a part of each stage. For

example, in the alternatives phase, analysts often use

In 1997, the telematics category barelyexisted.

models to suggest hybrid strategies that combine theoriginal alternatives. In the analysis phase, they usedynamic models to calculate ranges for importantbusiness variables, such as cash flow andmarket share.The OnStar case was difficult tomodel. In the vehicle

business, GM has decades of experience and plentifulhistorical data. Modelers can build on a wealth of pre-vious analyses and examples of best practice. TheOnStar business was very different in that the tele-matics market did not exist and no one had experienceor historical data.To cope with the inherent uncertainty, we needed a

modeling process that would allow integration of vari-ous methods and data sources. Our method had to beflexible enough to absorb a wide variety of inputsbased on judgment, historical analogies, market re-search, and other sources. Our chosen modeling ap-proach integrated concepts and techniques from sev-eral management sciences approaches:(1) System dynamics was useful in two important

ways. First, the stock-flow structure used in systemdy-namics provided a good foundation for the physics ofcustomer migration from one state to another. For ex-ample, the model included structures that tracked theflow of customers from unawareness, to adoption, tochurn (loss of customers), and to possible resubscrip-tion. Second, the feedback perspective used in systemdynamics was very helpful in modeling the networkexternalities that are crucial to the telematics market.(2) Conjoint analysis was used to calibrate the con-

sumer choice relationships that govern consumeradoption of OnStar.(3) Dynamic optimization was used to assess the



Figure 3: Dynamic modeling is a part of the GM strategic decision process.

correct magnitudes for price and for spending onmarketing.(4) Diffusion models and the literature on previous

studies of product diffusion helped us to understandthe impact of market spending, word of mouth, price,and product innovation on OnStar adoption.(5) Concepts from lifetime customer-value analysis

helped us to analyze the impact of churn and customerrecapture.(6) The real-options approachwas useful for decom-

posing the decision to expand OnStar into a set ofsmaller, less costly steps. As it implemented each step,GM had the option of taking the next step if the out-come of previous decisions was favorable.

The Integrated Simulation ModelA simulation model was our core tool in the OnStarstrategy project (Figure 4). It had six key sectors: cus-tomer acquisition, customer choice, alliances, customer

service (whichwe discuss inmore detail), finances, anddealer behavior. The financial sector calculated finan-cial metrics, such as revenue, cash flow, and profit. Thedealer-behavior sector dealt with issues of how dealersmade customers aware of OnStar and how much saleseffort they expended. The sectors interact over time togenerate time series for such important business vari-ables as market share and cash flow.

Customer Acquisition and RetentionThe customer acquisition and retention sector of thesimulation model describes the inflows and outflowsof OnStar subscribers. The model builds on conceptsfrom the literature on lifetime customer value by ex-plicitly representing the causal mechanisms responsi-ble for subscriber acquisition and churn. Since the ac-cumulated number of subscribers produces monthlyrevenue directly through the monthly subscription fee,we could use the model to evaluate the financial



Figure 4: The simulation model of the telematics industry included multiple, interacting sectors.

impact of decisions that would affect GM’s acquisitionand retention of subscribers.The detailed model structure begins with the simple

relationship that determines the number of newsubscribers:

NS � TR *V, (1)

whereNS is the number of new subscribers addeddur-ing a time period, TR is the take rate (a fraction be-tween zero and one), and V is the number of new ve-hicles on which OnStar is available. V can include bothGM and non-GM vehicles, depending on GM’s policytoward alliances with other vehicle manufacturers.The modeling of the take rate (TR) begins with ideas

developed in the product diffusion literature. The takerate is the product of customer awareness and cus-tomer choice:

TR � A*C, (2)

where A is the fraction of new car buyers who havetop-of-mind awareness of OnStar, and C is the fractionof new-car buyers who choose to subscribe to OnStarconditional on awareness. The product-diffusion lit-erature suggests that the two factors that drive aware-ness are the coefficients of internal and external influ-ence. The coefficient of internal influence usuallyrepresents the effect of word-of-mouth communicationon sales. We believed, and the data have subsequentlyconfirmed, that word-of-mouthwould be an importantfactor in generating awareness for OnStar.Several published metastudies of estimated diffu-

sion models were helpful in calibrating the word-of-mouth effect (Mahajan et al. 1995, Sultan et al. 1995).We had no direct evidence about the magnitude of theeffect in the telematics market, but we used published



studies to construct a range of estimates that wethought were reasonable. The ranges were the basis forextensive sensitivity analysis on word-of-mouth andother uncertain parameters.The coefficient of external influence represents the

effect of the firm’s marketing effort on product adop-tion. In several product-diffusion studies, researchershave modeled the external effect as an increasing func-tion of spending on marketing. We used a similar ap-proach and used the metastudies to calculate reason-able bounds for the magnitude of the effect. We also

We believed word-of-mouth wouldbe an important factor in generatingawareness.

consulted with internal GM marketing experts andoutside advertising experts to estimate how muchOnStar would have to spend to reach different levelsof awareness. Consistent with previous studies, the re-lationship between marketing spending and aware-ness exhibited diminishing returns.The choice variable is the conditional probability

that a new-car buyer will subscribe given that OnStaris available on the vehicle. The probability of thebuyer’s choosing OnStar depends on the utility ofOnStar relative to the utility of alternative uses of themoney and the utility of any available competing sys-tems. We modeled the probability of choice with thelogit-choice function that is commonly used in mar-keting studies (Meyer and Johnson 1995). We derivedthe utilities we used to calibrate the logit choice modelfrom a consumer research study.The choice function included the effects of network

externalities. Writers on the economics of new-productdiffusion make a strong case for the importance of net-work externalities or positive-feedback effects (Shapiroand Varian 1999). Positive-feedback effects are the in-creases in the value of the service for all existing usersas additional users adopt the service. We were awareof these effects and asked an outside consulting firmto search for examples relevant to the OnStar situation.Its research turned up several examples that were use-ful primarily for identifying potentially importantpositive-feedback processes. The examples were not

useful, however, for projecting the number of OnStarsubscribers or estimating revenue and cost flows.We believed that three positive-feedback processes

could be important in the telematics industry. Eachprocess could create positive feedback, but their im-plementation required managerial attention and effort.The first process concerned creation of new OnStar ap-plications. In 1997, OnStar was limited to the coresafety and security features. Although these featureswere extremely valuable, the market research showedthey would not be enough to drive widespread pene-tration. From the beginning, we recognized that GMcould never find and build internally the myriad ap-plications that OnStar would need. Alliances with im-portant new-economy and old-economy playerswould be crucial.We considered the factors that would make GM an

attractive partner to prospective content partners andthe factors that would make it economic for GM toinvest in partnerships. Examples of potential contentpartners included Fidelity Investments and severalvehicle-insurance companies. For both old- and new-economy firms, partnerships become much more at-tractive as the installed base of subscribers grows. Theclassic example of how applications partnerships cancreate positive feedback is Microsoft Windows. Appli-cations developers wrote applications for Windowsthat made Windows more valuable. Increased valueincreased the number of customers buying Windows,which, in turn, made developing further applicationsmore attractive.We hypothesized that a similar process could occur

with OnStar. The economic dynamics of a recently an-nounced OnStar alliance for providing real-time per-sonalized traffic information demonstrates this pro-cess. Market research revealed that consumers wantpersonalized traffic information and that providing itcould be the telematics killer application. Traffic infor-mation requires both GM and its partner to make ma-jor up-front investments. Providing the traffic infor-mation feature is economically unattractive with asmall installed base because the average cost per sub-scriber would be much too high. The economics be-come very attractive as the installed base reaches sev-eral million. The alliance mechanism forms a positivefeedback because the availability of the traffic service



makes OnStar more attractive to car buyers. In turn,additional subscribers make OnStar alliances more lu-crative for all involved.Quantifying the magnitude of the positive-feedback

processes was a challenge. We were tempted to con-clude that we could not accurately quantify thesemechanisms and to leave them out of the formal mod-eling. We decided to try to quantify the alliance feed-back, because historical examples suggested that theycould be critical to OnStar’s development, and becausethe strength of the positive feedback is affected bymany other variables, such as pricing and spending onmarketing. Omitting the mechanisms from the modelwould greatly distort the effects of alternative pricingand marketing-investment policies.To quantify these effects, we considered a long list

of potential services and partners. We used GM man-agers’ judgment and financial data to estimate how the

The AMIC standards could be adouble-edged sword for OnStar.

number of subscribers would influence the economicsof the different services. We used a combination ofmarket research and judgment to estimate how con-sumers would value additional services. We used sen-sitivity analysis to determine how the system wouldreact to different values of uncertain parameters.The second positive-feedback process concerned the

dynamics of third-party sales of OnStar. Third-partysales would be installations of OnStar through stereostores; electronics retailers, such as Circuit City andBest Buy; and discount retailers, such asWalMart. Pre-vious research reported in the marketing literature iso-lated a positive-feedback process in which productswith high sales and large market shares receive moredisplay space and attention at retailers, which, in turn,further increases market share and sales (Reibstein andFarris 1995).Third-party distribution is the tool for reaching the

200 million existing vehicles and the 70 percent of newcar buyers who buy competitors’ cars. The viability ofthird-party distribution depends on the cost and easeof retrofitting vehicles with OnStar hardware. Wealready knew that replicating the existing dealer-

installation procedure would be too costly and wouldgenerate few sales. Also it would be time-consumingand expensive to modify OnStar to interface with mul-tiple car electronics systems. GM acted to reduce thecost of third-party installation by sponsoring the Au-tomotive Multimedia Interface Consortium (AMIC).The AMIC is in the process of setting standards forconnecting to vehicle electrical systems. The standardswill enable any manufacturer of telematics systems toconnect to the vehicle electronics systems of any au-tomobile without compromising the integrity andsafety of any of the systems.The AMIC standards could be a double-edged

sword for OnStar. On one side, they enable businessesto connect OnStar to competitors’ cars. On the otherside, they allow businesses to connect future compet-itors’ telematics systems with GM cars. The AMICstandards increase the value of building a viable large-scale telematics system because the consumer valueinitially created for GM cars can be extended to otherplatforms.The third positive-feedback process concerns includ-

ing other vehicle manufacturers in the OnStar alliance.We believe that our project is one of the first pub-

lished applications to analyze the strategic implica-tions of network effects in a real-life situation. Al-though the importance of network effects is clear, expost, from many historical case studies, few new-product models actually incorporate them. Gupta et al.(1999, p. 327) wrote the following:

Network effects have attracted significant attention fromeconomists in recent years. However, marketing scientistshave been slow to respond to the growing importance of thisphenomena in new product adoption. For instance, most newproduct models in the marketing science literature assumethat new products are autonomous and that the adoption ofnew products is not affected by the presence or absence ofcomplementary products. These assumptions are being calledinto question in almost every durable product market in theNetwork Economy, where firms rarely act alone to create newproducts.

Customer ChoiceTo calibrate the customer-choice sector, GM commis-sioned a conjoint study to estimate how consumerswould respond to different subscription fees, initial



costs, and combinations of features (Reibstein andFarris 1995). The sample for the conjoint study was 621new-car buyers. In the conjoint study, researchers es-timated the utility of 13 potential attributes of theOnStar system, including route guidance, remote ve-hicle diagnostics, traffic information, initial price, andmonthly subscription fees.The study showed that consumers could be divided

into six market segments. The segments had differentutilities for the service attributes and prices. In the cus-tomer acquisition sector, we calculated take rates sep-arately for each market segment.We used the market study to calibrate the consumer-

choice decision. For example, we tested the impact ofdifferent attribute combinations and prices on long-term OnStar penetration and profitability. We also ex-perimented with alternative price trajectories, such asskim and penetration pricing. Skim pricing involvedinstalling OnStar on a few expensive GMmodels, suchas Cadillac, and charging a premium price for OnStar.Penetration pricing involved installing OnStar on allGM models and charging a low price that would max-imize the take rate.

Vehicle Manufacturer AlliancesThe option of offering OnStar to other vehicle manu-facturers emerged early in the project. Clearly, enrol-ling other manufacturers could be beneficial. First, in-creasing the vehicles in the alliance would create alarge OnStar installed base and strengthen positive-feedback processes. Second, GM could collect licensingfees for the use of OnStar. The disadvantage of makingOnStar generally available would be that GM wouldlose a competitive weapon for selling vehicles.We evaluated the option of offering OnStar to other

manufacturers by including an additional positive-feedback process in the model. Manufacturers had fouroptions in the telematics market: do nothing, start theirown services, join the OnStar coalition, or join anothercoalition. The probability that amanufacturer (other thanGM) would choose one of the options was given by

P � f(S, V, T, C, F, M), (3)

where S is the number of subscribers for a specific ser-vice (OnStar, Ford, and so forth), V is the number of

vehicles that offer a service, T is the take rate for aservice, C is the estimated investment cost of settingup a telematics service, F is the fee charged by the co-alition for using the service, andM is the manufacturerthat sponsors the coalition. In the model, we assumedthat the probability of choosing an existing service in-creased with increases of S, V, T, and C. A large baseof subscribers (S) and a high take rate (T) demonstratethat the service is successful and will be more success-ful in the future. Other manufacturers would prefer toenroll in a successful coalition. In addition, a high takerate shows that consumers want the service and thatmanufacturers without a service are at a competitivedisadvantage. If the sum of the Vs across coalitions islarge, most vehicles offer telematics services and theholdouts are under pressure to join one of the coali-tions. A high cost of establishing a service (C) makes it

The analysis showed that the cost-focused strategy would cause theeffort to fail.

disadvantageous to create a new service and more ad-vantageous, especially for small manufacturers, to joina coalition. High fees (F) for participating in a coalitionreduce the probability that an outside manufacturerwill join. Finally, some manufacturers will be very hes-itant to partner with other manufacturers, such as Fordwith GM, so that the identity of a coalition sponsor (M)influences the probability of an alliance.Original equipment manufacturers (OEMs) benefit

greatly by joining an OnStar coalition. First, replicatingthe GM system would be very costly for auto manu-facturers with much lower volumes than GM, espe-cially if GM were able to exploit positive feedback andadd high-value services. Second, assuming that con-sumers find telematics services attractive (market re-search supports the conclusion), if several competingOEMs were to join the coalition, the holdout compet-itor could lose precious market share in the vehiclebusiness. Finally, if OnStar were to build a crediblethird-party distribution system with the AMIC stan-dard, OnStar would have access to the competitors’cars even if they didn’t join the coalition. OEMs couldconclude that their best interests lie in joining the co-alition and cutting the best deal possible, instead of



letting GM capture their customers through theaftermarket.For each major vehicle OEM, the team considered

the costs and benefits of partnering with OnStar fromthe perspective of that competitor. Our reasoning pro-cess was similar to that of estimating competitor pay-offs in a game-theory analysis. During each time pe-riod, the model calculated the probability that amanufacturer would choose one of the four options;once a manufacturer chose to join a coalition or to startits own service, it could not change its decision.The alliance decision structure creates another

positive-feedback process. Additional partners in-crease the value of the system through multiple mech-anisms, such as word-of-mouth and additional appli-cations. In turn, a more valuable system attracts newsubscribers and additional partners. We did not be-lieve that these processes would be so strong as to cre-ate a single system for the whole vehicle industry. Weacknowledged that some competitive automakerswould be so averse to a GM-sponsored system thatthey would never join an alliance.

Customer ServiceThe customer-service sector represented demands forcustomer service and the acquisition and retention ofservice capacity. Poor customer service could restrainthe long-run growth of OnStar. Rapid subscribergrowth increases the demand on the call centers.OnStar must be able to match customer-service capac-ity to demand or, beyond a point, the quality of itscustomer service will deteriorate. Common sense sug-gests and the literature on customer service confirmsthat customers’ poor experiences with service reducethe attractiveness of the service, reduce the firm’s ac-quisition of new subscribers, increase churn, and gen-erate negative word-of-mouth.A firm can minimize the negative effects of inade-

quate customer-service capacity by choosing the rightcustomer-service policy. Often, firms run their call cen-ters with a cost mentality. Their objective is to mini-mize the cost of the call center by paying low wages,limiting the time spent per call, and always running atclose to full utilization. The model-based analysisshowed that the cost-focused strategy would cause the

entire OnStar effort to fail. OnStar depends on a staffof intelligent, well-trained service personnel to provideexcellent service during difficult situations, such as carwrecks and serious illness. It takes time to recruit andtrain people who are up to the task. Consequently,OnStar has adopted a policy of purposely overstaffingthe call center in order to build customer-service ca-pacity in advance of expected demand. The overstaff-ing policy gives service employees opportunities tolearn systems and scenarios before they have fullcustomer-service responsibilities. This policy is theonly one that is consistent with the strategy of buildinga large installed base, and we estimate that it will payfor itself several times over in terms of lower churnrates and positive word-of-mouth.

Model Analysis of the FactoryInstallation DecisionGM’s most important decision was whether to factory-install OnStar in new vehicles. Factory installation hasseveral advantages. First, factory installation elimi-nates the $1,300, two-day dealer-installation processthat greatly reduced the take rate. Second, factory in-stallation eliminates the need for dealers to convince

OnStar adopted a policy of purposelyoverstaffing the call center.

buyers to purchase OnStar as an option. The disad-vantage of factory installation is that the hardware issomewhat expensive and, if a new vehicle buyer doesnot subscribe to OnStar, GM gains no revenue fromthe factory installation.Factory installation directly affects cost by eliminat-

ing the cost of dealer installation (Figure 5). Factoryinstallation also increases the number of cars withOnStar available and, for a given take rate, the numberof OnStar subscribers. Increases in the number of sub-scribers strengthen the positive-feedback processesthat influence alliances and the development of appli-cations. Specifically, additional subscribers increasethe attractiveness of creating OnStar applications andthe attractiveness for other OEM’s of joining the On-Star alliance. Additional applications and a larger al-liance both increase the number of new subscribers.



We compared the number of subscribers for 100-percent factory installation with a baseline in whichOnStar was a car option (Figure 6). Although the dif-ference between the two scenarios is fairly small dur-ing the first two years, the positive-feedback processescause a dramatic gap in the eighth through 10th years.Profits are low during the first three years in thefactory-installation case because of the high initial costof installing OnStar in all GM vehicles. In subsequentyears, revenue and profit are much higher in thefactory-installation case than in the base case becausetotal subscription fees increase and the average costper subscriber drops.The simulation model was also an essential tool for

sensitivity analysis. We varied uncertain parametersone by one and in combinations to determine whetherdoing so would change the best strategic option. Fi-nally, we ran Monte Carlo simulations that varied allthe uncertain parameters together. We found that thesuperiority of the factory installation strategy was ro-bust against almost all parameter variation (Figure 7).

Financial, Organizational, andSocietal ImpactsIn late 1997, the project team recommended a very ag-gressive strategy that included factory installation onall GM vehicles, recruitment of other manufacturersinto the OnStar system, and making the first year ofservice free. In addition, we recommended that GMaggressively pursue alliances with content partners,such as a major cell-phone networks, Fidelity Invest-ments, and Dow Jones Publications. GM’s senior man-agers agreed to the aggressive strategy, and imple-mentation is proceeding today. Many people arefamiliar with OnStar through GM’s well-known Bat-man advertising campaign.Vince Barabba, GM’s senior executive for strategy,

and Ron Zarrella, the president of GM at the time, be-lieve that the modeling process greatly influencedGM’s strategic choices for OnStar. Vince qualifies hisbelief in the utility of models by proclaiming Barabba’sLaw: “Never say the model says.” Vince’s point is that

Figure 5: The positive feedback loops that drive OEM alliances and applications amplify subscriber growth.



people, not models, have points of view that areformed by experience and can be clarified and im-proved by simulation models. Especially for strategicchoices, models can never be more than aids in reach-ing sound judgments.

Financial ImpactsThrough 2001, the implementation of the OnStar busi-ness strategy has progressed very much as expected.As of fall 2001, OnStar has more than two million sub-scribers and an 80 percent share of the telematics mar-ket. The number of subscribers is growing rapidly. Noother vehicle manufacturer is expected to launch acompetitive system until 2003 at the earliest. The larg-est competitor is a small independent firm that hasabout 80,000 subscribers.Competitors will have difficulty replicatingOnStar’s

business. OnStar has benefited from learning during

the past four years. Creating a system like OnStar ismuch more complex than putting a modem and GPSin the car and contractingwith a call center. The systemmust be capable of handling millions of subscribersand never fail because of the high stakes involved inauto accidents and emergencies. Many dot-com firmshave learned how difficult it is to expand mission-critical IT systems to accommodate millions of users.The system must be flexible enough to add new ser-vices quickly without disrupting existing features.GM has forged alliances with vehicle manufacturers

that account for over 50 percent of vehicle sales. GM’spartners include Toyota, Honda, VW-Audi, and Su-baru. Honda and Toyota are initiating factory instal-lation on their Acura and Lexus models, respectively.GM has also enrolled an impressive list of content

partners. WestWood One, a prominent radio com-pany, has partnered with OnStar to provide real-timetraffic information. Dow Jones provides theWall Street

Figure 6: The factory installation strategy generated far more subscribers than the incremental base case strategy.



Journal, which is read to car occupants by the OnStarvirtual advisor. The virtual advisor downloads the re-quested content and reads it to the driver through thevehicle’s sound system. The virtual advisor also allowsusers to trade securities from the vehicle through Fi-delity Investments. OnStar receives approximately fiveproposals per week from prospective service partners.OnStar is on track for financial success. In early No-

vember 2001, Ron Zarrella, then president of GM, fore-cast that OnStar would break even by 2003 and wouldgenerate major cash flows thereafter. Investment bank-ers have valued OnStar at between $4 and $12 billionif it were to be spun off as an independent business(Merrill Lynch 2000, Lapidus 2000, Martiliotti 2000).

Organizational ImpactsThe OnStar project contributed to creating a new en-terprise mental model for GM. The company has hada mental model of the vehicle business as a transaction

in which GM creates value by the transaction of sellinga vehicle. This mental model, or theory of the business,served GM well for many years, but intense competi-tion has reduced its effectiveness in recent years.The new mental model augments the transactions

revenue with a stream of revenue from service busi-nesses like OnStar (Figure 8). Examples of other servicebusinesses are enhanced vehicle service and repair,and vehicle insurance. The annuity revenue and cashflow from services should greatly improve GM’s mar-ket value. Investment banking studies have estimatedthat GM could generate several billion of extra earn-ings annually from greater participation in servicebusinesses. In addition, the service and vehicle busi-nesses form a positive-feedback loop in which healthyservice businesses make the vehicle business morevaluable, and a healthy vehicle business makes the ser-vice businesses more valuable. OnStar was the firststep and a critical component in the process of con-necting the vehicle business and service businesses.

Figure 7: The simulation model was used to generate multiple scenarios for subscribers and other importantvariables.



Societal BenefitsThe first societal benefit of the OnStar project is thecreation of the new telematics business. The telematicsindustry did not exist before GM implemented itsstrategy. Today, Wall Street analysts project that theindustry will grow to $12 billion over the next 10 years.In addition, such services as real-time traffic informa-tion will increase productivity by helping to reducetraffic congestion.By far, OnStar’s most important contribution is sav-

ing lives. OnStar answers thousands of emergencycalls each month and has often made the differencebetween life and death. Studies in emergencymedicine

show that reducing response time by only fiveminutescan enormously increase survival probabilities. On-Star’s impact on the lives of thousands of people canbe partly measured by the amazing letters that GMreceives every day. These letters tell remarkable storiesof how rapid emergency response has prevented deathand serious injury. As more manufacturers make tele-matics services available, survival rates should in-crease further.

AcknowledgmentsThe authors thank Lyle Wallis for his ongoing help in developingand expressing the concepts in this paper. They also thank JohnSterman and Nelson Repenning of MIT for their ideas and the use

Figure 8: The new GM enterprise mental model increases market value by managing the dynamic interactionbetween the vehicle and service businesses.



of their classes as test beds for our presentation. Bill Stripling pro-vided valuable coaching assistance in developing the Edelman pre-sentation. Rebecca English Paich was very helpful in editing the pa-per and improving the clarity of the writing.

ReferencesGupta, Sachin, Dipak C. Jain, Mohanbir S. Sawhney. 1999. Modeling

the evolution of markets with indirect network externalities:Anapplication to digital television. Marketing Sci. 18(3) 396–416.

Gurumurthy, Kalyanaram, William T. Robinson, Glen L. Urban.1995. Order of market entry: Established empirical generaliza-tions, emerging empirical generalizations and future research.Marketing Sci. 14(3) Part 2, G212–G221.

Lapidus, Gary. 2000. Gentlemen start your engines. Goldman Sachs,January.

Mahajan, Vijay, Eitan Muller, Frank M. Bass. 1995. Diffusion of new

products: Empirical generalizations and managerial uses.Mar-keting Sci. 14(3) Part 2, G79–G88.

Martiliotti, Domenic. 2000. Can GM finally deliver the product? BearStearns Equity Research, December.

Merrill Lynch Equity Research. 2000. General Motors Corp. April.Meyer, Robert, Eric J. Johnson. 1995. Empirical generalizations in the

modeling of consumer choice.Marketing Sci. 14(3) Part 2, G180–G189.

Reibstein, David J., Paul W. Farris. 1995. Market share and distri-bution: A generalization, a speculation, and some implications.Marketing Sci. 14(3) Part 2, G190–G202.

Shapiro, Carl, Hal Varian. 1999. Information Rules. Harvard BusinessSchool Press, Cambridge, MA.

Sultan, Fareena, John U. Farley, Donald R. Lehmann. 1990. A meta-analysis of applications of diffusion models. J. Marketing Res.27(1) 70–77.

Documents

AMultimethodApproachforCreatingNew BusinessModels