You can outsource responsibility but you can't outsource accountability

A couple of times in the last decade, the (medium-sized) organisations I worked for have outsourced IT software development projects to large consultancies. In both cases the results have been disappointing, to say the least, and I would like to use this post to explain why.
The problem with a small organisation (in this context, anything under £100m turnover is small) outsourcing to a large consultancy is the asymmetry of risk. Put simply, the consequences of failure for the customer are much greater than the consequences of failure for the supplier. The last company I worked for outsourced a development project with a contract value of £2.3m. This represented approximately 25% of the total IT budget to us but was no more than a rounding error for the supplier which typically dealt in contracts worth several hundred million. The project was absolutely critical for the customer, whereas the supplier literally didn't care whether it succeeded or not.
To understand why this is the case, we need to understand how large consultancies work: In any large consultancy, at any one time there will be a number of people who are not employed on any paying project. These people are 'on the bench' in consultancy parlance. Indeed, one of the selling points of big consultancies is their supposed ability to scale up quickly. Human nature being what it is, the majority of the people who are on bench will not be the best staff in the business, as these will get snapped up to work on the important (i.e. high-value) projects.
So if you have, say £5m to spend, you will end up with a team that is mostly inexperienced with below average skills, leavened with a few good people who were unlucky enough to be in the wrong place at the wrong time. Of course, you will still be paying £800-£1200 per day (£1500 if they see you coming) for staff who are little better than graduates, and untalented ones at that. Worse still you may be actually paying them to learn skills (programming in new languages, project management) that they should already have.
The end result is a late delivery of a sub-standard product with all the tricky stuff left out. The in-house team will have to work like trojans to turn it into something that the business can actually use, all the while seething at the proliferation of Audis and BMWs that turn up in the car-park, driven by the consultants.
The solution?
There are two parts to this: 1: Outsource to a company that has as much to lose as you do. So, if you have £3m to spend, try to spend it with a company that has a turnover in the region £30-100m. That way, if they fail it will hurt, but it won't kill them. 2: Make sure that the person who is responsible on your side really understands how to manage outsourced contracts.
If you can't honsetly say you can do both of these (this is known as capability to perform in process improvement circles), then don't outsource; build up your skills in-house, or buy off the shelf and make it work for you.