Why Alliances Fail

Our industry can succeed only by collaborations," a biotech CEO recently told Pharm Exec, "because no company has the whole of the jigsaw complete-only a piece." Clearly many of his pharma counterparts agree. The web of alliances formed by the top two dozen biotech and pharma companies from 1973 to 2001-at least the 12,500 contracts made public by these firms-is as tightly knit as a linen shirt.

Meanwhile the strategic significance, complexity, and cost of biotech–pharma alliances are, if anything, expected to rise. Everyone knows why: the "hunger for product candidates," in the phrase of editor Jennifer Van Brunt of biotech deal specialists Recombinant Capital (ReCap). It seems that "virtually all drug developers are raiding each other's shelves," she says. Jim Hall, president of life sciences at advisory firm Wood Mackenzie, says licensing, which accounted for about $65 billion of Big Pharma's revenues in 2002–2003, will probably top $100 billion in three years. As Van Brunt says, "Companies are in-licensing and out-licensing like crazy."

Collaboration Continuum

The pace of alliance formation generally should decelerate, Hall says, while the average deal size increases. "The days of the small, testing-the-waters deal worth tens of millions of dollars is going away and will be replaced by very large equity investments," he says. According to consultants at McKinsey & Co., up-front payments in therapeutic alliances increased more than sixfold from 1988 to 2002, and average milestone payments soared, from $6 million in the years 1988–1990 to $85 million in the 2000–2002 period. They also found that 65 percent of pharma executives anticipated that alliances would become more complex. As Hall says, the easy deals have been done: "Large Biotech and Big Pharma are looking for ones in real therapeutic areas rather than discovery and early technology."

Another trend is less frequent and more dedicated partnering. Fintan Walton, PhD, chief executive of PharmaVentures, a UK-based company that helps biotechs make deals, says some pharma companies are effectively saying, "We prefer to work with a few biotech companies we know well over the years rather than reinvent new alliances and new relationships."

These developments put a premium on alliance know-how for pharma companies and executives alike. Patricia Martin, executive director of alliance management at Eli Lilly, says: "I've heard the head of our US affiliate tell his leadership team that hands-on experience managing a collaboration will be necessary to be successful because that's the way the environment is going."

This article, the second of two parts (see "Dangerous Liaisons," Pharm Exec, May 2004), discusses the alliance phenomenon, examines failure rates and what they're based on, and suggests ways pharma can improve its results.

On Their Toes

"A dance between an elephant and a flea" is how Ed Saltzman, president of biopharmaceutical consultancy Defined Health, describes the typical biotech–pharma alliance. Size discrepancies aside, why would different species care to tango? Simple. Each has something the other wants. But instead of buying, they borrow. And rather than take control, they share.

Lilly Promotes Alliance Spirit

Companies ally when certain conditions apply, says Professor Ben Gomes-Casseres of Brandeis University: (1) They have a capability their proposed partner values but does not have (and could not develop), and the partner has a capability they want but do not have (and could not develop). (2) It would cost more to buy these capabilities than to ally them. (3) It would cost more to merge the companies than create a series of alliances. (4) Combined, the capabilities are more valuable than apart.

Alliances can serve almost any purpose, from marketing to production and geographic expansion to R&D. However, the needs that prompt companies to seek partners are generally ones that can't be fully specified. The product or capability may not even exist. A complete contract can't be written.

Alliances manage this problem by operating under an an open-ended contract (also described as incomplete, relational, or evolving) that refrains from spelling out every aspect of the final objective and how it will be achieved. Rather than pin down each partner's behavior beforehand, it obligates them to coordinate and decide matters jointly as events unfold.

An incomplete contract need not be vague, Gomes-Casseres says. "You set down as much as you can about the various rights and responsibilities in the relationship. But you realize it will stand or fall on the partners' ability to come up with answers to questions they haven't thought about before." The heart of a relational contract is a commitment closer in spirit to what binds friends, family members, and spouses than typical business partners. The prevailing attitude is, as the song goes, "we can work it out."

One way to see how unusual alliances are is to compare them to acquisitions. According to Professor Mike Peng of Ohio State University,"Alliances work well when the ratio of soft to hard assets is relatively high, while acquisitions are preferable when such a ratio is low. Alliances create value primarily by combining complementary resources. Acquisitions derive most value by eliminating redundant resources." And, "consistent with real options thinking, alliances are more suitable under conditions of uncertainty, and acquisitions are more preferred when the level of uncertainty is low."

Alliance Success Factors

The various kinds of alliances-co-marketing programs, one-way licensing agreements, R&D contracts, strategic supplier relationships, cross-licensing, minority investments, cross-shareholding, joint ventures-can be arrayed along a continuum with nonalliance activities-with spot market transactions at one end and corporate mergers and acquisitions (M&A) at the other. (See "Collaboration Continuum.") The collaborative structures, moving from left to right, are characterized by increases in organizational interdependence, control, up-front costs, and time, and a decrease in flexibility.

Alliances fall into two classes, contract and equity-based, occupying opposite ends of the continuum. Contract-based alliances, like the one between Abbott Laboratories and Cambridge Antibody Technology (discussed in part one) are generally formed for a specific purpose for a limited time and entail a relatively low level of inter-firm involvement. Equity-based alliances (for example, the recently announced arrangement between GlaxoSmithKline and Theravance, also discussed in part one) tend to be more open-ended and usually inspire higher levels of interaction. (A contract-based alliance, despite the name, is still governed by an incomplete contract.)

The chief benefit of alliances is risk management. Firms deal with uncertainty in three ways, says Gomes-Casseres. They can postpone action, curb the cost of negative outcomes, or "influence the uncertainty itself." Alliances, he claims, accomplish all three: (1) They allow firms to hold off making premature bets by making "incremental commitments to an unfolding strategy," comparable to traversing a stream via stepping stones (the logic behind real option reasoning, exemplified by GSK–Theravance). (2) They let firms diversify risk by sharing it. (3) They create new responses to underlying uncertainties-technological, market-related, or competitive-by joining the collaborators' strengths.

Causes for Concern

But alliances also entail a tradeoff. You reduce performance risk in exchange for taking on a new kind: relationship risk. "You now have to manage something with an outside party," Gomes-Casseres explains. "You don't control them. You have some rights, some sharing of decision making. But clearly you don't have the full latitude you have when you control all the pieces."

The Meanings of Failure

For all their purported value and importance, strategic alliances pose two persistent puzzles: (1) Most fail. Numerous surveys say 50–70 percent in every industry don't meet expectations, raising the question: Why bother? (2) Most seem to fail for controllable reasons. Even when undertaken as a hedge against technological uncertainty, the reasons for failure most frequently cited by participants are nontechnological-culture, communication, leadership, strategy. (See "Causes for Concern.") This raises another question: If managers could do better in these areas, wouldn't alliances succeed more often? If so, then aren't resources-time, money, personnel-being squandered?

Side by Side or Toe to Toe?

Yes, says Stuart Kliman, co-founder of relationship consultants Vantage Partners: "A huge amount of money and time are wasted." Hall recalls that the cost of poor alliances was in the tens of billions of dollars several years ago.

Maybe so. But there may be less to reported failure rates than meets the eye:

What counts? It's hard to say what a failed alliance is. Are technical failures included? Can technical and nontechnical failures be distinguished? Kliman suggests that breakdowns in the working relationship can cause technical failure. How do you weigh relative and absolute failure, the difference between disappointment and disaster? Are a partnership's stated objectives to be taken at face value? What if they're unrealistic? What if an alliance meets its goal but is terminated-on schedule? What if it doesn't meet its goal but achieves some other benefit? Is it a failure?

Who says? The verdict on an alliance can be asymmetrical-one partner gives it a thumbs up, the other is less kind.

Compared to what? What does a single number mean? Most business startups fail. Most mergers and acquisitions fail. And most drug leads don't pan out. So what if most alliances fail? When someone told science fiction writer Theodore Sturgeon that 90 percent of science fiction is "crud," his response-since enshrined as Sturgeon's Law-was "90 percent of everything is crud."

What do we really know? It's "difficult to establish a true failure or termination rate," says Michael McCully, director of consulting at ReCap, "because companies tend to 'bury their dead at night.'" Even his company's database of 18,000 deals does not have "good numbers on alliance termination rates because we don't have enough good information."

Deciding to form an alliance may not necessarily be daft. Still, there's plenty of room for improvement in the doing.

Murphy Was an Optimist

Alliances fail for a multitude of reasons. Whatever can go wrong, will. Companies get stuck with a bad match, a partner that is a poor strategic, organizational, or cultural fit. They drag out negotiations or saddle themselves with onerous terms. Hall recalls a $60–$70 million, five-year deal between a large pharma and medium-sized biotech to develop a therapeutic franchise to plug a gap in the former's portfolio. "It fell apart," he says, "when the biotech found that the more 'successful' the alliance became, the more money it would lose. The terms were beneficial when they were a small startup. They just never thought they'd grow up." Companies can suffer from partner opportunism or poor governance. They can also lose the "learning race," the contest to unlock their partner's bag of tricks (firm-specific resources and capabilities) before the partner figures out theirs.

Patricia Martin, executive director of alliance management at Eli Lilly, believes alliances fail because: (1) The science doesn't work. (2) The scientific decision isn't good because the communication isn't good. (3) There's no hope of a scientific decision, good or bad.

The last two are just the sort of factors one would expect partners could minimize. But if the research is to be believed, the situation has actually worsened. In 1999, three of the top eight reasons for alliance failure (see "Causes for Concern") were external. By 2003, just one is. Only "drastic changes in business environment," doesn't indict the executives involved.

But perhaps these figures deserve skepticism. Gomes-Casseres wonders, "When managers report that 'our cultures didn't mesh,' what does that mean? It could mean we had a conflict of interest and didn't see eye to eye. It could mean we spoke different languages. It could mean we had different decision-making protocols. To use a large term like culture or trust as an excuse for failure lumps it into a large category that doesn't tell you what's going on."

Shallow analysis is not the only problem. Though not obviously self-serving, these self-incriminating lists of failure factors make a habit of omitting some even less flattering possibilities:

Overoptimism. Economists explain business failure as the expected outcome of a rational decision regarding a risky scheme. But psychologist Daniel Kahneman, winner of the 2002 Nobel Memorial Prize in Economic Sciences, and Dan Lovallo, a senior lecturer at the Australian Graduate School of Management at the University of New South Wales and a former strategy specialist at McKinsey & Co., wrote in the Harvard Business Review (July 2003): "When forecasting the outcomes of risky projects, executives all too easily fall victim to what psychologists call the planning fallacy"-the product of cognitive biases and organizational pressures-"which leads them to make decisions based on delusional optimism rather than on a rational weighting of gains, losses, and probabilities." Eric Bolesh, senior analyst with Cutting Edge Information, would agree: "At the beginning of an alliance, there's a lot of promise, good feelings, handshaking, backslapping." Folks don't think about what can go wrong, Bolesh says. It's not that they ignore the worst-case scenarios so much as they never consider the "not quite worst case scenarios."

The winner's curse. "A lot of salesmanship" is required to get a deal done, Bolesh says. Kahneman and Lovallo believe the struggle for project approval raises the odds that the one chosen for investment will be the most overoptimistic-with the highest probability of disappointment.

Ego. Hall calls this "the number one reason alliances fail." Whose ego? "Typically the smaller partner," he says. "The inferiority complex that drove them to the table gets in the way. They're doing a deal because they need the cash, as opposed to because it fulfills their vision of the company."

Accepting that most alliances fail (whatever that means and for whatever reason), research shows that not all companies fail at the same rate. A 1998 study by consultants Booz Allen & Hamilton showed some firms had failure rates as high as 70 percent while others had rates as low as 10 percent. What makes the difference?

How to Succeed by Really Trying

To see why alliances fail, it helps to examine them from three angles. Each corresponds to a set of problems and a prescription for success.

An alliance-level perspective doesn't focus on the individuals or the environment, or one partner or another, but the relationship between them. "If folks aren't getting along," Kliman says, "it's because of structural issues." What about culture-whatever that is? "Good management is about figuring out how to handle cultural differences," he insists. "A good working relationship shouldn't depend on an absence of conflict." Alliances are founded on uniting different resources, different views, different capabilities. "The goal is not to eliminate differences," Kliman says, "but to create value out of them, turn them from a liability into an asset."

Companies that ignore an alliance-level analysis are asking for trouble. One partner's "mixed messages or unresponsiveness" can lead, Kliman says, to "individual behaviors that seem closed or disrespectful-that damage the working relationship and lead to mistrust and poor performance and thus slower decision making, missed milestones, lack of innovation, and, in time, a failed alliance."

Relationship management is the solution. Too often, Kliman says, management "thinks smart people will just be able to work it out." They rush into an alliance, hoping folks will figure it out on the fly, when instead, they first need to achieve "clarity about how decisions are made, what kind of information is to be shared, and so on."

Relationship management requires organization and preparation. He recommends an "internal launch" that clarifies the team's roles, responsibilities, and operating protocols and "gets internal boundaries effectively managed." The next step is a "joint launch" that brings the right people from both sides to the table to talk about how they'll handle problems.

Firm-level analysis brings a company's collaborative capabilities to the fore. The ad hoc approach sees alliances as a string of unique relationships. But there's a better approach, one that focuses on systematic alliance capabilities that transcend particular partnerships. That means creating tools, templates, and processes, and inculcating an alliance mindset throughout the organization-just what Eli Lilly has done. (See "Lilly Promotes Alliance Spirit.")

Successful companies understand that alliances are not peripheral. But neither are they replications or extensions of the company. They are strategically vital and organizationally distinct. Such companies are prepared to commit resources to alliances, to measure, evaluate, learn, specialize, train, and strategize. Many institutionalize their alliance capability by establishing a dedicated alliance management function. Professors Jeffrey Dyer of Brigham Young University, Prashant Kale of the University of Michigan, and Harbir Singh of the University of Pennsylvania compared 203 firms, including a few from Big Pharma, some with and some without alliance management functions. They found that those with alliance functions had a 25 percent higher long-term alliance success rate and generated almost four times the market wealth whenever they announced the formation of a new alliance. Bolesh confirms that almost everyone in pharma that Cutting Edge surveyed said it's important to have someone dedicated to alliance management, "not to do day-to-day project management and get caught up in the commercial or science side, and not to be a high-level white knight within the organization, trying to fix things without really understanding what the problem is, but just having someone there to shepherd the project along, to make sure that management on both sides are speaking, to make sure that senior management, if necessary, is involved, and basically bail the ship out if it starts to get a little leaky now and then."

Network-level analysis surveys industry structure and competition, forces that compel companies to create constellations of alliances. Companies need a map to navigate not just their own web of partnerships but their competitors' and potential collaborators' too. However, as Gomes-Casseres is fond of noting, many of the companies that form strategic alliances lack an alliance strategy. The tragedy is that strategy is what should dictate partner choice, governance structure, and every other aspect of alliances. Strategy even determines what counts as failure or success. It may not be a short-term result. It could even be something other than cash-knowledge, repositioning, or a new product or source of supply.

To place individual alliances above a coherent alliance strategy is a grave error, Gomes-Casseres believes. "When you start thinking about optimizing the deal instead of optimizing the strategy, you're looking at the wrong target." One symptom, he says, is misinterpreting "the tendency of alliances to change over time as a weakness." The goal of an alliance, after all, "is not its survival but the success of the alliance strategy."