Fourthly, alliances can gain scale - much like traditional M&A. Think about how British Telecom and AT&T consolidated their international telecoms into a joint venture with more than $5 billion in assets and 35,000 employees.

Fifthly, we have alliances which improve supplier effectiveness. Mercedes has done this in the US by shifting much of the risk, capital investment and effort to long-term suppliers.

Sixthly, and perhaps most significantly, we have alliances to create advantaged networks.

Look at how Microsoft kept its options open with OS/2 joint ventures and alliances with Apple and Unix vendors whilst it was investing in Windows. Microsoft used alliances to hedge its bets and ensure that it would remain a contender.
 
The book suggests that striking an alliance fast is crucial. It quotes a seasoned alliance manager who claims that: "Do it in 3-4 months and you have an 80pc hit rate.  If it takes 6 months that falls to 25pc. Over a year and it has been a nice experience."  Alliances also fail if they don't have a sponsor, probably on the board.  But it is dangerous if the sponsor is the CEO, as this means there are limited checks and balances.

The first 30-180 days are critical.  Millions can be lost as time drifts by.  Consider having a team of launch managers to blitz the key issues.

Why do alliances fail?  Ernst lists seven big failures - unclear objectives, lack of a detailed business plan, decision gridlock, aligning with a weak or competitive partner, unmanaged cultural clash, the failure to learn or to protect core capabilities, and failure to plan for alliance evolution.

Many fail because partners do not set up formal processes to ensure that things are in place.  Valuable information about the partner held by negotiators at the start is often not passed on to the implementers.

The role of the alliance manager is becoming ever more crucial, and the book quotes a chief executive who described the goal of  an alliance manager as "not to create harmony, but to create a sense of dynamic tension.  To be a flying buttress in a cathedral!"

Interestingly, well-managed companies do not tend to do well in alliances, according to Charles Roussel.  He quotes research showing that companies with a high market-to-book value lost $355 billion through their joint ventures and created a negligible $1 billion through their licensing agreements. Low market-to-book value companies created $102 billion through joint ventures and $438 billion through licensing.

This suggests that well-managed firms are often centralised, hierarchical and bureaucratic, characteristics that can hinder alliance success. Second, poorly managed firms benefit disproportionately from association with better run companies.  The real winners are those companies which are systematically putting together alliance capability, Cisco, IBM, BP Amoco, Corning, Millenium and Eli Lilly.

The next stage is the growth of constellations where rival alliances fight each other. This is the norm in the airline industry, where the number of alliances rose to 120 by the late 1990s.  The book argues that companies need to be able to manage not just one, but a constellation of alliances.