Global networks

The need to access skills and capabilities means corporates must learn to draw value from R&D sites across the world.

by Professor Yves Doz and Keeley Wilson, INSEAD
Last Updated: 23 Jul 2013

The corporate world is now on the cusp of a new era: the globalisation of innovation. This can release huge value for a firm's R&D output in terms of its cost, speed and relevance to the market. But many companies over-depend on well-established sites in their home countries to develop new products and concepts that are then rolled out to regional centres to be adapted to local markets.

Others rely on semi-autonomous entities across the world, which focus on innovations for their own markets with little communication and collaboration across the whole network. Too many companies fail to draw value from their R&D network in terms of efficiency, opportunity and competitive advantage.

The extent of the globalisation of innovation was revealed by INSEAD's 2006 survey, Innovation: is global the way forward?, which looked at 186 global companies with a combined R&D spend in 2004 of $76.4 billion. It showed that in 1972 less than half of all R&D sites were located outside companies' home countries.

Almost 30 years later, this figure had jumped to more than two thirds. More importantly, over time, the reasons why companies establish innovation sites have changed. It used to be important for R&D to be located close to manufacturing sites, but as production has become more codified and digitalised, the need for proximity has declined.

Now a much more important driver of location is the need to access large, fast growing and demanding consumer markets, such as Brazil or China. Another factor today is the need to access skills and capabilities: first, falling numbers of scientific and technical students in tertiary education in the West mean there aren't the brains to meet the requirements of many industries. One has only to witness the growth of the software cluster in Bangalore, India, to see this. But nowhere is the skills gap starker than in the US aerospace industry, from which 40% of current engineering staff will retire within the next five years, according to The Economist.

Second, technology convergence has opened up new possibilities for collaboration. Increasingly, companies need to look further afield for core knowledge, often in adjacent industries. Take the example of Essilor, one of the world's leading manufacturers of spectacle lenses. Its R&D and production footprint has to draw on specific competencies from three continents: optical design from Germany, polymer technology from a joint venture in the US, and finishing and thin film coating from a joint venture with Nikon in Japan. So what seems a relatively simple product to the wearer is the result of specialist knowledge from across the world.

As pioneers in the field of global innovation are discovering, it is no easy task to co-ordinate and manage such diverse networks that traverse cultural as well as geographical boundaries. A key challenge for business is to build an integrated innovation network in which all of the constituent parts work well together, with each making a valuable contribution to global product development.

Strategically, an R&D network has to be configured to deliver value. This is done by ensuring that each centre is in a location that allows it to contribute unique skills or knowledge to the network as a whole. Integration is achieved by harmonising tools and processes across the network, with 'cosmopolitan' managers who are able to bridge the cultural divide between distant centres and global projects.

In comparison with the global innovation challenge facing many companies today, the globalisation of manufacturing was relatively easy to execute because it involved moving codified knowledge embedded in entire processes and systems. This is much easier to parcel out to different places.

As long as companies had the right IT and logistics controls, they could locate almost anywhere. The globalisation of R&D is much more difficult because the innovation process relies on tacit knowledge that is often context-specific. This type of knowledge can't be written down or stored, and can be learned only through experience and exposure.

So where are the best places to be? This is a more complex question than it sounds. Much depends on the industry, but a general rule of thumb is that the locations should bring the best differentiated knowledge or skills to the network. Healthcare firm Novartis takes a strategic approach to its R&D network based on access to unique knowledge.

Its R&D site in Cambridge, Massachusetts, is co-located with MIT and close to the medical science developed and practised at Massachusetts General Hospital. Its other R&D hub is at Europe's leading centre for pharmaceutical research in Basle, Switzerland.

These large R&D centres are complemented by specialist hubs, including a site in Singapore that focuses on the type of tropical diseases prevalent in the region, a new site in Shanghai focusing on infectious diseases common in China, such as hepatitis, and one in La Jolla, California, that draws on local knowledge to specialise in experimental technologies and biotechnology.

Each Novartis R&D site contributes a unique piece to the overall jigsaw and thus allows the company to access and meld knowledge. For example, the Singapore R&D centre collaborates with a wide range of local partners, which all contribute a fresh perspective to the drug development and delivery process. A compound developed in Singapore to treat tuberculosis could also be effective for treating other respiratory diseases elsewhere in the Novartis network.

The second challenge is to ensure that the R&D networks are properly integrated. On average, the companies we surveyed believe that closer integration could improve the speed of the innovation process by 33%, increase the quantity of innovations in their pipeline by 29% and cut the cost by 22%.

But we have found that while companies are aware of the potential value from closer integration, many of them struggle to make it a reality. Clearly, they need to have the right tools, processes and systems to support global communication and collaboration, but integration also means building a set of capabilities to support multi-site teams and global projects.

It may sound obvious to say that companies' systems, processes and tools should be harmonised across an R&D network, but companies need to be aware that the extent of harmonisation is dependent on the nature of the task being undertaken.

For example, dispersed development activities will benefit enormously from standard tools and processes without which the testing, verification and integration of modules will be impossible. A less clearly defined and more creative innovation activity, on the other hand, will suffer from the imposition of too many tools and processes.

Within the development processes, the extent to which the harmonisation takes place is a matter of choice. At one end of the spectrum, companies give managers the autonomy to make some decisions based on their assessment of local conditions.

They install the same systems and introduce common practices, such as project lifecycle management and standard verification procedures, at every site, but leave some flexibility for local managers to make their own decisions. At the other end, companies adopt processes such as Six Sigma, which create stringent project management structures with clearly defined roles, common tools and uniformity of context across sites.

Whichever approach is adopted, harmonising systems and processes will cost money and create disruption, as staff need to be fully trained in the use of these common elements. But failing to grasp the nettle merely reinforces a significant barrier to a more integrated R&D and in turn shackles a firm's ability to innovate. We found that companies that followed an aggressive innovation strategy were significantly more likely to have standardised processes and systems across their networks than other companies.

Global innovation means creating a climate of openness in which staff can engage in exchanges throughout the network and not just in their immediate environment. In an integrated network, knowledge should be shared and re-used. Everyone knows that reinventing the wheel is costly in terms of resources and time to market.

Giving sites access to existing knowledge also reduces the risk of internal patent infringements. Xerox is an example of where the sharing model worked. In order to leverage its global knowledge assets, it developed CodeX, a system based on open-source principles to allow R&D staff and its partners to share and re-use code.

Before embarking on a new piece of work, developers can check whether any software has already been written within Xerox that performs the same function. If they find existing code, they can use and adapt it. Xerox found that, on average, each piece of re-used code was done so two and half times. Across an organisation the size of Xerox, this translates into significant cost savings, as well as building bridges across sites.

Common processes and systems need to be supported by reliable and wide-ranging communication tools. The CodeX platform was successful in Xerox because it supported the transfer of codified knowledge, and therefore accessing it via ICTs worked well. But not all knowledge can be expressed in databases, intranets or emails. Most R&D involves huge amounts of tacit knowledge, which can be understood only through experience and action.

When GSK acquired hi-tech start-up Affymax in the 1990s, it had to understand a new philosophy of drug discovery in which robotics engineers, computer scientists, biologists and chemists together build a machine that performs combinatorial chemistry (molecular synthesis). The only way GSK could grasp the new technologies was to send an R&D team to work with Affymax in designing, building and testing machines.

Different communication channels are appropriate for different types of knowledge. Managers will have to know what tools they need to create effective communication between dispersed sites. But too often companies overlook, or underestimate the cultural barriers to communication. Without shared values and rules of behaviour, communication lines can quickly become dysfunctional. As a general rule, it is far better to invest time and money into site visits and exchanges to build trust before relying on email or video-conferencing.

We asked the companies in our survey about the 'people dimension' of global innovation and found that they were overwhelmingly in agreement that people with multicultural experience have a greater ability to absorb, interpret and use unfamiliar or new knowledge, and were much better equipped to work in virtual teams. 'Cosmopolitan' managers, who have experience of living and working in different national, organisational and functional cultures, are essential for global R&D.

Joint venture Schneider Toshiba Inverter provides a good illustration of the value of having cosmopolitan managers in a global R&D network. In 2000 the group, which makes the drives for a wide array of machines from elevators to escalators, began a new product development project across sites in France, Austria, Japan and New Zealand; to add a further impediment, the site in New Zealand was part of a new acquisition. Managing a project of this size and scale across different time zones, organisations and cultures would normally be courting disaster.

But the head of the New Zealand site had previously worked for the joint venture in France and Japan, the head in Vienna was a Bulgarian who had worked in France and Germany, and the heads in Japan and France had worked closely with each other on previous projects. They were able to spot and defuse potential conflicts or misunderstandings before they had a chance to derail the project or damage relations, and also had wide personal networks across the organisation they could call on when needed.

The challenges to getting value from global innovation are significant and require companies to look at both the strategic and day-to-day aspects of their R&D assets. This type of network needs to be configured to add value, after which it needs to be managed and integrated to deliver value. Few companies can afford not to face up to these challenges. Companies that fail to integrate their R&D assets will quickly find that the cost of running them far outweighs the value that they create.


The integration of global R&D networks should naturally compel companies to shift from co-located projects to global ones, which, if approached and managed correctly, can provide much greater resource flexibility, reduce development time and deliver more robust products. But for many companies the benefits never materialise and instead multi-site projects are marred by disruptions and overruns.

Multi-site projects are inherently difficult to manage. Many of the process-based approaches successfully applied to co-located projects fail to take account of the complexity and systemic nature imposed by dispersion, where everything is different - from the set of questions that need to be asked in the pre-project definition phase to the leadership structures that have to be put in place.

First, not all projects are suited to dispersal. When the architecture of a product can be easily modularised, then splitting work packages across sites is feasible. For integration to be successful, there needs to be a small overlap of skills at each site.

Although each site should have unique skills, if the sites don't understand the work and capabilities of each other, then designing and developing modules for integration won't work. When products are based on highly integrated, interdependent or locally rooted knowledge, projects should be co-located as it would be impossible to share knowledge across sites.

Second, how tasks are allocated can create problems. Too often, companies embark on global projects as a way of bringing together available resources that are scattered across their R&D network when no single site has the capacity to run the project. Global R&D networks configured to create value will have differentiated capabilities at each R&D centre, and it is these capabilities that should define task allocation in global projects. In other words, each site should contribute something that can't be found elsewhere.

Third, it is imperative that the strategic context and marketing perspective are shared with all project teams at the outset. With large, dispersed projects it is easy for gaps to appear not only between R&D staff in various locations but also between functions. People need to understand what they are doing and why they are doing it.

Each R&D site will have its own internal hierarchies, a self-perception of the worth of its contributions to the company and its own internal procedures for problem-solving and decision-making. There needs to be a clear locus of control, and well-defined roles for managers and R&D staff, to avoid infighting and empire-building.


1. Select project sites based on the unique capabilities they will bring to the project rather than resource availability.

2. Allow sites to work together on small projects to build trust and familiarity before launching a large global project.

3. Involve all sites in the pre-project phase to ensure everyone has a shared vision of what the project will deliver, when and by whom.

4. Project managers should be 'over-informed'. By knowing everything that is happening, they can detect and defuse potential problems before they become serious.

5. Project sites should share common tools, processes and reporting structures.

6. Provide project teams with a wide range of communication tools and channels, including reliable ICTs, as well as site visits and secondments.

Yves Doz is the Timken chaired professor of global technology and innovation at INSEAD; Keeley Wilson is associated senior researcher at INSEAD.

Find this article useful?

Get more great articles like this in your inbox every lunchtime