Why outsourcing projects over-run

Being on the vendor sales side for number of years, I had always felt it was the client who caused outsourcing projects to over-run. Having spent the last year or so on the client side, I have miraculously come to the conclusion that vendors are out to fleece naive clients, thereby causing project over-runs.

What has been your experience?

If I have to combine my vendor side and client side experience in a dispassionate way, the conclusions on why outsourcing projects over-run are suprisingly simple and easy to deduce - however, it always amazes me why neither the clients nor the vendors learn. Probably because we use the terms 'clients' and 'vendors' as if they are two individuals - in real life, they are two organizations made up of different individuals with separate egos, metrics, aspirations and businjess understanding.

Consider the client side scenario - The CFO needs the outsourcing budget to be finalized even before the users and the CIOs have figured out what needs to be outsourced. Of course, being a CFO, he would like the costs to be as low as possible, and has no time for or understanding of mundane concepts like business processes, user experience, requirements analysis, functional specifications or estimation. The CIO / business user pulls a number out of thin air to get the budget approved - further quibbbling with the CFO ensues a budget number approved which will bear no resemblence to the final spend. The CFO knows it too well - however, he can later blame the over spend on the CIO. The CIO knows it too well and can blame the over spend later to the technology miracle called 'Change request'. The business user knows it well too, but can blame it on changed customer needs.

Consider the vendor side scenario - the sales person pushes the delivery arm to create an estimate. The delivery guru pushes the process down the hierarchy. At some stage, the estimation process lands on the table of a techie who is probably fresh out of a technology school. Using established estimation methodologies such as LoC or FP estimation, he creates what he considers a 'scientific' estimate. However, since about 70% of all costs are resource costs in an outsourcing project, our freshly minted techie forgets to apply a human judgement factor to the estimation. He makes up for it by applying what he calls 'risk contingency' and pushes the estimate up in the food chain. Throughout every stage of the food chain, the approver covers up for his lack of understanding of the project by buffering up the 'risk contingency'. By the time the estimate reaches the client, the 'risk contingency' is the major cost of the project.

Does the above scenario sound familiar to you? Do you believe there is a way out of above quandry or are we all resigned to a world where the final spend on over 80% of outsourcing projects bear no resemblence to the initial budget?

your views welcome!