Partner Selection in Market-Driven Strategic Alliances

How should one select a strategic alliance partner? An answer to this question is provided by extending the literature on symbiotic marketing and focussing [sic] attention on market-driven strategic alliances. Such alliances are defined as long-term inter-firm co-operative relationships that add value for the customer. Value is created by providing the advantages of multiple choice purchase options coupled with the convenience of seamless, one-stop-shopping. This means paying attention to customers and competitors in selecting alliance partners. Market-driven strategic alliances are posited to be more successful when usage and firm complementarily levels are correctly matched with the alliance strategy being pursued.

M uch attention has been devoted in recent years to the con cept o f strategic alliances and the need for businesses to co operate in order to be com petitive (Perlm utter & Heenan, 1986; O h m a e ,1 9 8 9 ; Lew is, 1990;Badaracco, 1991;Byrne, Brandt & Port, 1993;R angan & Yoshino, 1996). T he im port ance o f the topic is underscored by the fact that over 20 000 new alliances w ere form ed betw een 1987 and 1992 (Harbison & Pekar, 1994), and that the num ber o f dom estic and crossborder alliances grew m ore than 25% annually from 1989 to 1994 (B leeke & E rnst, 1995). G lobal com petition, inform ation technology, and increasingly know ledgeable and, sophisticated cu stom ers all call for new form s o f organiza tion. As evidenced by the rash o f press reports, in South A frica businesses are turning to alliances in an attem pt to overcom e the legacy o f isolationism and enhance their overall com petitiveness.
• P ressures to grow have led firm s to explore the adoption o f m id-range options such as alliances, netw orks, and other hy brid organization arrangem ents (Oliver, 1990) as alternatives to m ore traditional m a r k e t versus hierarchy options, w hich have proved problem atic. In the ill-fated A llegis experim ent in the U nited States in the m id-1980s, for exam ple, U nited A irlines attem pted to create an integrated travel supply organ ization w ith H ertz C ar Rental and H ilton International and W estin H otels, all und er com m on ow nership (W illard, Schoenecker & K reuger, 1990). T heir efforts to achieve synergies and offer g reater benefits at low er costs failed. T he intended benefits proved d ifficult to attain and costly to achieve. Alli ances represent an alternative structure to capture these elu sive benefits at low er cost.
Strategic alliances m ay be seen as a value-adding effort on the part o f the firm , th at affords fresh opportunities for build ing and m aintaining com petitive advantage (B ucklin & Sengupta, 1993). S trategic alliances have been viewed from a variety o f perspectives, including firm internationalization (B eam ish & Banks, 1987), transaction cost econom ics (Parkhe, 1993), netw orks (M iles & Snow, 1986;Jarillo, 1988), or ganizational learning (H am el, 1992), gam e theory (Parkhe, 1993), developm ental processes (Ring & Van de Yen, 1994), and ethics (Gundlach & Murphy, 1993). M uch o f this re search has focussed on risk, fixed costs, econom ies o f scale and access to distribution as motivations for alliance form a tion. W hat is lacking is an appreciation for the centrality o f the custom er in the conceptualization, design, and m anage m ent o f strategic alliances. A s Peter F rancese (1994), Presi dent o f Am erican D em ographics, observed, w hile an increasing num ber o f com panies are form ing alliances, they are doing so for the w rong reasons. 'T he custom er', he la m e n te d ,'is being ignored in all this'.
B ased on a m arket-driven approach to strategic alliances, this article presents an alternative perspective. Such a per spective encom passes both a custom er orientation and a com petitor focus (com pare Day, 1992). Varadarajan & R ajaratnam 's (1986) updated notion o f sym biosis, and its fea sibility for exploiting strategic grow th opportunities, is ap plied. First, an overview o f strategic alliances w ithin the context o f sym biotic m arketing is presented. Second, a m ar ket-driven model o f the alliance form ation process is devel oped, focussing on the partner selection question. A contingency m odel for partner selection is developed, propo sitions for future research are advanced, and the implications for structuring and m anaging strategic alliances are discussed. T hroughout, the analysis is illustrated with exam ples drawn from the travel industry, w hich has w itnessed a virtual explo sion o f alliance activity in the past decade (Business Week, 1992;Dev & Klein, 1993). .

Alliances for competitive advantage
U ntil recently, inter-firm relations have been given sur prisingly short shrift in m arketing (Fisk, 1987), industrial organization econom ics and organization theory (Thorelli, 1986). F o r a thorough review o f the conceptual foundations o f strategic alliances and the role o f m arketing, see Varadarajan & C unningham (1995). In a general sense, strategic alliances are defined ,in term s o f synergistic or sym biotic relationships betw een tw o or m ore independent entities (Adler, 1966); that is; w hen certain outcom es result from the com bining o f tw o firms that w ould not occur if either o f the firm s w ere to attem pt to achieve the outcom e independently. T h e inability o f m arketing theory to address such issues o f co-operative behaviour was highlighted by Arndt (1979), w ho argued that theory rem ains based on a notion o f m arkets w here transactions are; anonym ous and transient. 'Go it alo ne' and 'W inner take all' are the funda m ental,u n d erly in g tenets o f such a paradigm .
There are tw o g enerally accepted m otivations for alliance form ation: efficiency and effectiveness (Jarillo, 1988). Effi ciency objectives o f an organization, to 'do things rig h t', in volve exploiting cost-reduction opportunities by form ing alliances with o ther com panies that have com plem entary ex pertise. Effectiveness objectives o f an organization, to 'do the right th in g ', involve m arket-based m otivations to increase m arket share and size, or to create barriers to com petitive en try. By taking a m arket-driven approach to strategic alliances, efficiency and effectiveness can be addressed sim ultaneously.
The m ost im portant aspect o f a strategic alliance, and per haps the m ost difficult to define, is the output m easure, that is, perform ance. Success m ay be m easured in term s o f longevity (relative to intended duration) or enhanced perform ance, al though m easuring p erform ance is notoriously difficult (An derson, 1990). Suggested perform ance m easures range from qualitative process m easures, such as satisfaction w ith the re lationship (B ucklin & Sengupta, 1993), to quantitative out com e m easures, such as num ber of: new products developed or m arket share gains (A nderson, 1990), to indirect perform ance indicators, such as spillover effects (Parkhe, 1993). A l liance perform ance, and consequently the success o f the alliance, should be tied to the goals o f the alliance at its incepr tion. I f an alliance is m arket-driven, its success should be measured in terms o f m arket effects, such as m arket share or custom er loyalty.

Alliance failure
W hile there has been grow ing in terest in the subject o f co operative relationships, there has also been grow ing dis enchantm ent w ith them . Som e studies suggest that 70% of joint ventures fall short o f expectations or are disbanded (Levine & B yrne, 1986), w hile B leeke & E rnst (1995) report that nearly 80% o f jo in t ventures ultim ately end in a sale by o n e 'o f the partners. W hile som e o f the disenchantm ent may be due to unfam iliarity w ith a new form o f business activity, and consequent unrealistic expectations, o ther factors may also explain alliance failure. M any o f these co-operative relationships are costly to m aintain and encounter problem s in m anagem ent. F ranko (1971), for exam ple, has identified problem s w ith jo in t ventures and subsequent divorces in multinational com panies.
Failure often stem s from the types o f relationships under taken, and questions w hether the relationships w ere appropri ate from the outset. A re strategic alliances m erely the latest m anagem ent fad or an enduring organizational form that will benefit the m odern corporation (Byrne, B randt & Port, 1993)? A re alliances form ed because they are appropriate, or sim ply because they are possible? Certainly, the m anagem ent o f such relationships p oses unique problem s o f conceptuali zation, design and p erform ance evaluation.
A substantial body o f literature exists on the factors that contribute to the success o f inter-organizational relations. W hat is m issing is a sound, m arket-driven basis for partner selection. A lthough som e studies recognize organizational com patibility as an im portant determ inant o f alliance success (Spekm an & Sawhney,, 1990), m ost do not focus on the part ner selection issue. V aradarajan & R ajaratnam (1986) argue that custom ers' buying behaviour, and w hat such behaviour im plies, should be the fundam ental issues in partner selection.

Contingency approaches
In past studies o f strategic alliances, contingency theory has been the central theoretical perspective (Varadarajan & Raja ratnam , 1986;Spekm an & Sawhney, 1990;B ucklin & Sen gupta, 1993;Burgers, H ill & Kim, 1993). This approach suggests that organizations seeking to im prove perform ance do so within the context o f internal and environm ental situations that influence their organization. This is the reason why alliances succeed for som e businesses, but fail for others (Spekm an et al., 1996). Efforts to develop a contingency theory o f strategic alliances fall into tw o main groups: synergy and sym biosis. T he synergy-based theorists focus on the antecedents o f (or bases for) alliance form ation and the consequences (or outcom es) o f alliance form ation (Spekm an & Sawhney, 1990;B urgers, Hill & Kim, 1993). T he sym biosis-based theorists focus on the alliance itself, delineating issues related to the conditions under which it form s, w hile also discussing the m anagem ent o f these relationships (Varadarajan & Rajaratnam , 1986;Parkhe, 1993). A marketdriven approach allows us to m erge these two theoretical streams.

Market-driven strategic alliances
A m arket-driven strategic alliance is defined as one that adds value for the custom er in a com petitive context. For exam ple, route and code-sharing agreem ents betw een international, air lines provide overseas travelers with the advantages o f m ulti ple purchase options coupled with the convenience of seam less one-stop-shopping. A dded value is imputed from custom ers' reactions to a com bination o f g oods and services, com pared to their reactions to the products o f the various firms taken individually. A lliances are successful w hen cus tomers perceive added value in their relationships with the firm as a result o f the alliance, since alliances are fundam ent ally tools for gaining com petitive advantage (Day, 1995). O ur focus is on w hat types o f alliances are appropriate for the individual firm under different c irc u m stan ces.,

Stages in alliance formation
Three stages describe strategic alliance form ation w ithin a Strategy-Structure-Perform ance paradigm (see F igure 1), and failure m ay occur at any stage. In the first stage, business goals and strategy dictate the type o f organizational structure to be used. A t this stage, internal and external environm ental considerations: are param ount; for exam ple dem and un certainty, com petitive uncertainty (Burgers, Hill, & Kim, 1993), resource dependency, transaction costs* life cycle, or strategic orientation (Spekm an & Sawhney, 1990). Failure at this stage results from a poor fit betw een an organization's strategy and its environm ent (M iles & Snow, 1978;Porter, 1980

Typology
In listing m arket-based goals for sym biotic m arketing, Varadarajan & R ajaratnam (1986) expand A n s o ffs (1957) typo logy o f grow th strategies. T he grow th opportunities can be distilled to tw o main m arket-driven perform ance goals: m arket penetration or share, and m arket developm ent or size. In other w ords, a firm can increase its sales by doing m ore business w ith its existing custom ers or by creating new custom ers, or both. T hese goals can be seen to deliver both greater effectiveness and efficiency.
Alliances, betw een present o r potential com petitors, may be intended to 'brid g e' gaps in product-m arket coverage, or to 'bundle' com plem entary products and services. T he concepts o f bundling and b ridging w ere suggested by Farquhar et al. (1992), w ithin the context o f branding strategies. T heir defi nitions, however, are no t identical to the usage here. Bridging alliances provide value to consum ers by expanding the cover age, either g eographically or dem ographically defined; o f ex isting firms. T hey also reduce costs through scale effects. B undling alliances provide value to consum ers b y : creating one-stop shopping, opportunities, reduced search and transac tion costs, and seam less delivery o f benefits. A lliances be tween m ajor airlines and feeder carriers are exam ples o f bridging relationships, w hile those betw een airline, hotel and car rental com panies are exam ples o f bundling.
Com bining the m arket developm ent versus m arket penetra tion goal o f the organization w ith the bundling versus bridg ing alliance approach, four types o f strategic m arketing alliances exist (see T able 1). :W hen the goal o f the firm is m ar ket penetration, and a bridging alliance is chosen, w e label this 'Bridge for S hare'. In an exam ple, C om A ir feeds its short haul point-to-point travelers and B ritish A irw ays feeds its long-haul travelers into each other's system s. Similarly, in an overseas exam ple, R adisson and SAS H otels cam e together in 1994. SA S ' main aim was to find a partner w ith strong m ar keting, skills, and thereby increase its m arket share. Since the alliance w as form ed, SAS reservations through R adisson's global reservation system have doubled {Hotels, 1996). O ut side the airline industry, bridge for share alliances are also re flected in h ospitals' building relationships with feeder clinics.
'Bridge for S ize' alliances are based on m arket develop ment goals, and reflected in the scram ble by m ajor interna tional airlines to find partners. SAA has form ed alliances with Lufthansa and Am erican A irlines in pursuit o f such objec tives. D esires to increase the size o f their m arket have also motivated alliances betw een Delta, Sw issair and Singapore Airlines, between U nited A irlines and Lufthansa, and be tween N orthw est and K LM Royal Dutch A irlines. Since their alliance was form ed in late 1989, Singapore A irlines reports that revenue derived from Sw issair and D elta grew 10 to 15% in the 1990/91 financial year over the previous year. Sw issair flights from Zurich to Singapore have increased from three to five flights a w eek due to the alliance w ith Singapore A irlines (Jennings, 1991).
'B undle for S hare1 alliances strive to penetrate existing markets. A m erican Airlines, H ertz G ar R ental and H ilton H o tels attem pted, but ultim ately failed, to get together in the CO N FIRM project, targeted at com m on business travelers. The. aim was to cem ent cross-usage by providing one-stop shopping for com plem entary products, w ith ease o f use and seam less integration o f services. T he failure was due to lack o f operational fit, in the difficulty in integrating their inform a tion technology system s, despite spending several million dollars, and not due to the underlying logic o f the venture. A successful exam ple is that betw een M cD onalds and Coca Finally, 'Bundle for S ize' alliances are intended to develop new markets. They m ake sense w here brands that are not cur rently used com plem entarily can be com bined to encourage greater jo in t usage. F inding that pizza was a favorite room service m enu item, M arriott Hotels and Pizza H ut Restaurants have form ed a partnership, providing a valued m enu item for M arriott guests, encouraging them to order room service, and giving access to the in-room dining m arket for Pizza Hut. In Europe, SAS Hotels and M cD onalds co-operate for sim ilar reasons, increasing the size o f each firm 's m arket by linking product-m arkets that w ere previously distinct.

Determinants
In the second stage o f the process, within the context o f the broader strategic decision, specifying m arket penetration verms' m arket developm ent goals o f the firm, and bridging versus bundling m otivations, structures are designed and specific partners are selected. It is at this stage that the bene fits of a m arket-driven approach are the greatest. The nature and -extent of custom er usage com plem entarity offers the basis for defining a broad 'consideration set' of potential partners (Varadarajan & R ajaratn am ,1 9 8 6 ). C ustom er usage com plem entarity refers to the extent to which a firm shares com m on custom ers with prospective alliance partners. Cus tom er usage com plem entarity is high when there is little over lap between the firm 's custom er base and those o f potential partners. T he nature and extent o f firm com plem entarity then defines a narrow er 'choice set' from which to select actual partners. Firm com plem entarity refers to the fit betw een part ners', resources and skills, and is based on pow er and cultural compatibility;: Firm com plem entarity is high w hen there is little: overlap betw een the: firm 's skills and resources and those o f potential partners. Partners may have different objectives, lack; required skills or resources, or differ in cultural orientation (G eringer, 1988a; 1988b). Partner match offers i th e : greatest oppo rtu n ity , for alliance effectiveness (Bucklin & Sengupta, 1993).
Firm com plem entarity is reflected in cultural com patibility and bargaining pow er parity. M uch has been written about cultural differences in international markets, and it is readily apparent that conflict may arise from differences in notions o f accepted conduct or use o f inconsistent symbols betw een so cieties. It m ust also be recognized, however, that organiza tions have their own notions o f accepted behaviours, and that clashes betw een corporate cultures may have m ore im m edi ate, and less recognized, im pacts on alliance success (Barnes & Stafford, 1993). A m ain them e in the writing on organiza tional culture and its im pact on inter-organizational relations is the degree o f sim ilarity or dissim ilarity betw een organiza tions (Hatch, 1993).
By bringing together different firms, alliances invariably create tension. M anagers m ust be aw are o f the source o f this tension, its potential negative consequences, and possible coping strategies for dealing with the unavoidable by-prod ucts o f alliances (Parkhe, 1991). Socio-cultural forces can create differences in the perception and interpretation o f phe nomena, based on different cognitive styles (A bram son e ta l., 1993). D ifferences in hom e country environments, as reflected in governm ent policies towards co-operation, indus try structures, and institutional support system s, create differ ences in expectations and experiences. Different corporate cultures, with unique ideologies and guiding values, may re quire alliance partners to restructure their norm s and belief systems to deal with conflicts between pow er and parity, or between m arket share and profitability.
T he relationship between cultural sim ilarity and alliance success has two opposing effects. On the one hand, alliances between organizations w ith high cultural sim ilarity may suc ceed due to a better fit and less friction. On the other hand, or ganizations with low cultural sim ilarity may benefit from an 'invigorating' effect that cultural diversity brings. The issue has more to do with cultural compatibility than with sim ilar ity (Bucklin & Sengupta, 1993).
In the literature on bargaining power, there are also con flicting arguments as to its effect on alliances. One point of view holds that a firm with: high bargaining power, whether on account o f superior products, internal resources, m arket position or technological skills, can use that position to its ad vantage. A stronger firm is able to position itself at the hub of a w eb o f parallel relationships (Harrigan, 1988), each clearly targeted and focussed. W ithout bargaining power, strategic re lationships are very risky and difficult to m anage effectively. Power allows the firm to raise the costs to its partners o f infi delity, either through up-front dem ands or through dissolution penalties. As a consequence, firms are m ore likely to use eq uity jo in t ventures rather than contractual arrangem ents when conducting jo in t R&D (Osborn & Baughn, 1990). The rea sons are that deeper relationships facilitate inform ation flow, align the interests of the parties, and provide for day-to-day co-ordination. \ The opposing point o f view is that firms have been found to develop their production bases through alliances with part ners o f different size, and strengthen their knowledge bases through alliances with partners in the same subject areas (Hakansson, Kjellberg & Lundgren, 1993). Empirical evi dence on this issue suggests that 'im balances in power and in managerial resources that each partner provides are significant drawbacks to alli-; ance operations and, as organizational theorists pre dicted, have an im portant role in lim iting alliance success' (Bucklin & Sengupta, 1993: 43).
As such, bargaining power parity can be said to have an effect on partner selection and alliance success.
Success at the second stage, leads to enhanced perform ance in the third stage, when the desired alliance m ust be im ple m ented and m onitored. M any scholars have exam ined the question o f managem ent o f on g o in g : relationships (Killing, 1982;1983). M arketing writers in particular have been con cerned with conflict and co-ordination in channels o f distribu tion (Stern & Reve, 1980;Dwyer, Schurr & Oh, 1987). Others have identified the need for appropriate safeguards against partner opportunism to be put in place (D ay & Klein, 1987;H eide & John, 1988). M uch o f the difficulty that arises in the third stage m ay be attenuated by m ore appropriate part ner selection in the preceding stage. The m ore appropriate a relationship -the better the fit w ith its environm ent -the greater will be its success. The conditions determ ining appropriateness reflect environm ental contingencies, both in ternal and external to the organization.

Contingency model
In the previous section, alliance strategies were classified into four types. Now, w e show how different partners are appro priate for each type, em bodying the Strategy-Structure linkage. Custom er usage com plem entarity and firm com ple m entarity are the two variables defining appropriateness. W hile choosing the m ost appropriate alliance does not guarantee success, it is a necessary prerequisite. The pro positions are sum m arized.in Table 2 which lists the partner selection criteria for each alliance type. Developing skills in m anaging alliances, required for successful im plementation, is addressed by Spekm an et al. (1996).

Customer usage complementarity
A nalysis o f custom er usage com plem entarity is similar for bridging and bundling alliances. The conclusion, however, depends on the goal o f the firm; whether m arket penetration or m arket development. As suggested by Varadarajan & Rajaratnam (1986), an opportunity clearly exists for firms con templating alliances to expand their m arket share and/or size by analyzing usage com plem entarity am ong related services. There is also evidence that a com bination o f brands with com plementary attribute levels receives better evaluations than a brand extension (Park, Sung & Shocker, 1996).
Gains in m arket share are achieved from increased pur chases from existing custom ers in the market. Alliances re ward custom ers for choosing to use the alliance products or services jointly, offering them better selection or greater con venience, coupled w ith lower transaction costs. In so doing, partners cem ent their relationships with existing customers and reduce brand switching; To achieve these goals, there m ust .be existing com m onality o f custom ers between a firm and its prospective partners. ;

Proposition 1
To achieve m arket penetration goals, firms should select al liance partners with whom there is low custom er usage com plem entarity.
M arket size is increased by reaching new customers and at tracting them into the market. Low levels o f cross-usage make it possible for firms to identify growth opportunities in new markets and focus attention on the challenges o f under standing how the new m arkets operate. If usage com plem en tarity were low (high overlap), there would be little incremental benefit as few additional custom ers would, be gained and new markets would not be developed.

Proposition 2
To achieve m arket development goals, firms should select alliance partners with whom there is high custom er usage complementarity.

Firm complementarity
W hile custom er usage com plem entarity determ ines the 'con sideration set' for prospective alliance partners, firm com ple m entarity defines the 'choice set'. Usage is a necessary but not sufficient condition for partner selection. In dom esticated, or administered markets the identity o f the buyer and seller m atters to the transaction. Anonymity is given up in exchange for reciprocity. Such markets are characterized by relational contracting (M acneil, 1978),. where negotiations are inter nalized in the relationship, rather than settled anonym ously in the market. Dom esticated markets are not devoid, o f conflict or com petition; these are inevitable when the goals o f differ e n t: parties are not fully congruent. However, the locus o f conflict resolution resides, in the interdependencies between the parties, and not in a neutral equilibrating m arket system. Hence, the managem ent o f coalitions is a key function of m anagement, and the heart o f strategic decision-m aking. In a strategic alliance, firms co-operate based on mutual need and share the risks to reach a com m on goal (Lewis, 1990). A lliances depend on relationships, but m utual reliance means mutual vulnerability. For m ost firms, alliances repre sent significant problem s o f m anagement. They bring to gether different parties with different cultures and bargaining power. A lliances raise the possibility o f conflict between the parties, and risks of dependence on one another. They also bring about new problems in perform ance evaluation. It is of ten difficult to establish whose perform ance should be meas ured, to agree on the appropriate; tim e-fram e for evaluation, and to make trade-offs between partners' divergent interests. , The appropriate level o f firm com plem entarity varies ac cording to w hether the alliance is bridging or bundling. An understanding o f the appropriate level o f firm com plem enta rity can be found in transaction cost reasoning (W illiamson, . 1985). The logic behind bridging alliances is to close gaps in product-m arket coverage.: By definition, appropriate partners have different skills and resources. Cultures and bargaining power should thus be com plem entary rather than overlapping factors in bridging alliances., U nder such circumstances, which involve the creation o f transaction-specific assets, we would expect high transaction costs to be a risk. The risks o f opportunism, however, are m itigated by the product-m arket knowledge that the firm already has developed, based on an understanding o f its own resources and skills. It is thus well able to m onitor the activities o f its partners, and avoids the control losses that would otherwise accrue;

Proposition 3
In bridging alliances, firms should select alliance partners with whom there is high com plementarity, based on cultural differences and asymmetric pow er bases.
/ The essence o f a bundling strategy is learning about new products and delivering com plem entary benefits to one's cus tomers. In contrast to the bridging case, however, with a bun dling approach the firm cannot easily avoid or detect opportunistic conduct and, despite the learning benefits fore gone, m ust select partners w ith w hom there is high overlap o f skills and resources. T his low com plem entarity, w hose pur pose is largely defensive, could arise from either sim ilar cul tural perspectives o r equal bargaining strengths. T he form er sim ilarity provides reassurance o f com m on values, w hile the latter ensures higher penalties for inappropriate conduct.

Proposition 4
In.bun d lin g alliances, firm s should select alliance partners with w hom there is high com plem entarity, based on cultural sim ilarity and bargaining pow er parity. W hile lack o f data on custom er usage com plem entarity and firm com plem entarity, both before and after these alliances w ere form ed; prevent us from testing the specific proposi tions, w e find tentative support from anecdotal evidence that alliance form ation is consistent w ith the propositions ad vanced above. For exam ple; in investigating the K LM -Northwest A irlin es alliance, Business Week (1992) reported that KLM badly needed the U nited States feed that N orthw est could provide and struggling N orthw est needed help as well; that is, a bridging alliance m ade sense. Similarly, it was re ported that N orthw est set out to change its im age for lousy service and top flight service was K L M 's forte; indicating high firm com plem entarity. T his alliance, though strategically sound, however, is encountering serious problem s in im ple m entation. T he tw o airlines have d ragged their feet on linking flights, which w ould help them feed transatlantic passengers to one another instead o f to com petitors. O ver time, one would expect this alliance to com e under increasing strain as N orthw est im proves its service im age.

Illustration
In the D elta, Singapore and Sw issair alliance, the stated goal w as 'a seam less relationship for the consum er' . Despite som e successes, the alliance is show ing signs o f strain. D elta has been reluctant to bring its partners into its term inal at Los Angeles International A irport, and is seen by its partners as dragging its feet (Foley & Yoffie, 1994). T he strain in the alli ance m ay be explained in term s o f the m odel developed above. In 1989, w hen the alliance was form ed, D elta had little international presence, serving only a few locations in Europe from secondary m arkets in the US, and was looking desper ately to develop new m arkets. In 'B ridge for S ize' alliances, we proposed partners w ith high custom er usage com plem en tarity (P2) and high firm com plem entarity (P3). From D elta's perspective, both Sw issair and Singapore A irlines fit these re quirem ents well. Both o f the latter carriers had high quality im ages, strong service reputations, extensive international route structures, and brought European and A sian custom ers, all very different from D elta in the U nited States. Learning would be enhanced from Singapore A irline's low -cost struc ture and Sw issair's cultural understanding.
By the early 1990s, however, D elta's strategy had changed, and so had the characteristics o f desirable partners. In 1992 D elta acquired PanA m 's E uropean routes, and began to com pete m ore directly with Swissair. D elta's.o b jectiv e becam e one o f m arket penetration to restore profitability to P anA m 's ailing route structure. A ppropriate partner characteristics in 'Bridge for S hare' alliances include low custom er usage com plem entarity (P I), m aking Singapore A irlines a less desirable partner. B ecause the focus is still on bridging alliances, high firm com plem entarity rem ains desirable. A ccording to our m odel, w e w ould expect to see D elta form new relationships characterized by less custom er usage com plem entarity, namely m ore custom er overlap. Recently, in fact, D elta has announced a new link-up with Sw issair, Sabena and A ustrian Airlines to im prove its m arket bridging capabilities. Due to PanA m 's previous presence in Europe, these carriers share m ore custom ers with D elta than do Asian carriers. The logic behind the new alliance is to be able to carry passengers from m ost A m erican cities to Europe, A sia and A frica on a single ticket.

Conclusion
Because o f their prevalence in globalizing m arkets, coalitions and alliances are topical m anagerial issues. A stronger: focus on non-or quasi-m arket behaviour is sorely needed at this time. The phenom enon o f co-operative activity carries im plications for broadened paradigm s of business behaviour. Sim ply regarding these alliances as evidence o f m onopoly pow er ignores their efficiency m otivations in com petitive markets. Such strategic alliances should be seen in the context of the pursuit o f sustainable com petitive advantage (D ay & Wensley, 1983). T he use o f non-m arket and quasi-m arket ar rangem ents to achieve such advantage is o f m ajor im portance. These arrangem ents reflect strong efficiency considerations and the changing scope o f the competitive arena. The form ation o f co-operative arrangem ents often en tails the shifting o f com petitive activity from one dim ension to another. C onceptualizing strategic alliances from a m arket-driven perspective is useful because it suggests specific guidelines for setting objectives and for partner selection, and identifies areas where research can be useful in assisting m anagerial de cision m aking. Two im portant points em erge from such a con ceptualization. First, m anagers should take a broad view o f strategic alliances and recognize the various effects they have on consum er purchase behaviour, as well as how interfirm re lationships affect m ore traditional m easures such as m arket share. Second, managers m ust realize that the long-term suc cess o f m arketing program s is greatly affected by the poten tial inherent in an organization's strategic alliances. Sim ply stated, because the structure o f strategic alliances will influence the effectiveness o f all future m arketing strategies, m anagers m ust u nderstand how their m arketing alliances af fect consum er purchase behaviour and thus m arket share.
M anagers need a m ore com plete understanding o f the pur chase behaviour o f their custom ers as a basis for m aking bet ter strategic decisions about target m arket definition and product positioning, as well as selection o f partners for strate gic alliances. Perhaps a firm 's m ost valuable asset for increas ing m arket share is know ledge regarding its cu stom ers' usage o f other brands in the sam e product class, as well as in com plem entary and substitute product classes (B essen, 1993). N ew products.developed or m arketed through product-based alliances have little chance o f success if no additional value is created for the custom er or i f managers do not know how to exploit the value o f their m arket potential by developing profitable alliances. W ith the elaborate w eb o f alliances de veloping in m any industries, the appropriateness o f partner selection will no dou b t have an im pact on alliance success.
A final issue deals with m arketing's contribution to the dia logue on strategic alliances. By conceptualizing strategic alli ances from a m arket-driven perspective and incorporating the central role o f custom ers and erstw hile com petitors, m arket ers are able to contribute m ore effectively to this discussion by surfacing issues relating to co-operative strategies, supe rior custom er value creation, and relationship m arketing (Day, 1992).

Future research
C om plem entarity in usage patterns and in firm capabilities offers a pow erful w ay to conceptualize and develop sym biotic relationships or strategic alliances. D esigning alliances in term s o f w hich partner to choose depends on w hether the objective is m arket penetration or m arket developm ent and w hether the alliance is intended to bridge gaps in productm arket coverage or to bundle com plem entary goods and services. U ltim ate effectiveness o f this method can be evaluated by com paring pre-and post-alliance penetration and share m easures. D espite the lack o f data to fully sub stantiate our m ethod, there is a com pelling theoretical ration ale for the argum ent presented. We hope to see future research providing em pirical support for the propositions that have been advanced.