How firms can make informed, project-based choices between single- and multi-sourcing.
Let’s assume you’re a company scouting for vendors for a variety of business-critical IT projects. Every candidate you approach wants to scoop up as much of your business as possible. Should you try to find the best partner for each task, or look for a one-stop shop?
The theoretical advantages of relying on one company, rather than several, are clear: less time lost in juggling multiple contacts; greater leverage resulting from being a more important client; lowered risk, assuming your partner is trustworthy, etc. Conversely, throwing your entire lot in with a single vendor seems inherently risky as well.
For many years now, multi-sourcing has been a rising trend, representing US$7.2 billion in global IT and IT-enabled services contracts in 2007 alone. Firms are increasingly choosing the flexibility and in-built competition between vendors that are part and parcel of multi-sourcing.
My recent paper for Management Information Systems Quarterly (co-authored by Shantanu Bhattacharya of Lee Kong Chian School of Business and Alok Gupta of Carlson School of Management) explores how firms can make informed, project-based choices between single- and multi-sourcing. Our research is the first on this topic to take into account how incentive structures commonly used in information-services outsourcing contracts are often out of step with outcomes (i.e. revenues). Moreover, we shed light on how the relationship between project tasks provides clues to the optimal sourcing strategy.
Performance metrics vs. revenue
Conventional linear contracts, which combine variable rates tied to performance metrics with fixed fees, often may not generate perfectly aligned incentives. To illustrate this point, let’s take the example of a firm that employed the same vendor both to manage its call centre and to redesign its website. The vendor’s pay was partly pegged to the volume of phone calls fielded at the call centre, a relatively easy-to-gauge metric that no doubt drove revenue. By contrast, the website redesign, though also significant from a revenue perspective, was not assigned a performance metric, because the user experience of a website is largely intangible. So we see that for this project, not all the levers affecting revenue could be manipulated through the contract. In that case, we can assume that in order to maximise revenues for a project whose outcomes were already difficult to measure, the company should have contracted with multiple vendors – so that there would be more levers of control.
Our theoretical model bears out the hypothesis that when the performance metric and project revenue are imperfectly aligned on the vendor side (as in the above example), multi-sourcing indeed should be the default – provided the tasks concerned are modular, i.e. fairly discrete and separable from one another.
When the metric and revenue are well-aligned, however, single-sourcing performs just as well as multi-sourcing – again, for modular tasks only.
For modular tasks, the client’s objective is simply to extract the maximum possible effort from the vendor. The client exerts little to no effort at this stage of project execution, because modular outputs snap together easily. One plus one equals two. So multi-sourcing and single-sourcing diverge only when the aforementioned misaligned incentives jeopardise the outcome. In that case, clients can address the problem by drawing up separate contracts for each vendor. Obviously, that solution is unavailable when there is only one vendor on the project.
When the various tasks involved cannot be separated quite so easily, i.e. are integrated rather than modular, things get a bit more complicated.
Here, single-sourcing is advantageous when performance metrics and revenues are well-aligned. Integrated tasks are not wholly dependent on vendor effort – the client has to be involved as well, because getting to two is not as simple as adding one and one. When performance and thus compensation hinges on joint work, there is increased temptation for any one vendor to put in less effort and let partners (the client and other vendors) take up the slack. It follows that under conditions of perfect alignment, working with one vendor entails less risk of this nature than does employing several. Therefore, single-sourcing is advisable.
Imperfect alignment, however, requires a trade-off between the above-mentioned risk and the potentially negative effect of partial alignment. Companies should consider which is the greater issue for them: impaired visibility of outcomes, or the free riding that can occur when multiple vendors tackle integrated tasks. The answer will depend on each company’s individual circumstances. For example, a massive Fortune 500 company will have a very hard time tracking performance on any one project. In such cases, it may be wise to use multi-sourcing as a default option.
Multi-sourcing’s biggest fans
There is anecdotal evidence that the banking and manufacturing sectors are the leading devotees of multi-sourcing for IT projects. Our findings suggest why that might be so. For single-sourcing to match or outperform multi-sourcing, vendor performance and revenue must be tracked together. This fundamental alignment cannot occur when the nature of the project – information services – is unrelated to the core business. Tech companies such as Google and Microsoft, on the other hand, would have stronger reasons to consider single-sourcing as a hedge against the risk of vendor free riding.
However, we believe our conclusions apply beyond IT. Companies of all sizes and industries are increasingly finding they cannot do it all and are investing heavily in partnerships. In addition to whatever internal contingencies may be shaping their sourcing strategies, they may also want to weigh the factors covered in our research – task modularity and output verifiability.
Sameer Hasija is an Associate Professor of Technology and Operations Management as well as the Shell Fellow in Business and the Environment at INSEAD.