I have just had an interesting discussion in a very large global enterprise around standardisation. These guys are good, very good in fact, and have really got control in a way most people in centralised IT would envy, but when it comes to trying to get ‘cross’ the enterprise business requirements to work then they are still finding it hard. The central IT department is surrounded by the operating divisions each of which has its own IT functions dedicated to managing IT that relates to the divisions individual specific activities and processes that are specialised. Central IT manages the generics of the business, i.e. financial reporting, and the cross services such as document management, as well as the basic infrastructures.

There are agreed, yep they really manage this, standardised applications and tools, established through a mixture of discussion to get opinions established clearly, followed by a relatively ruthless final decision which is not necessarily based on a consensus. They have a technique that requires projects to list all the ‘parts’ that they intend to use for approval, there is some debate about the level of detail, but it currently certainly identifies any recognisable applications. Deviations from the standard agreed application standards have to be approved, i.e. may be a particular database is required to work with a specialised divisional application, and in this case as this is a given, and will not affect any other activities, approval would be given.
Sounds great doesn’t it? And I was impressed, but it’s still not working. The why turns out to be that they, in common with just about everyone else is should think, don’t provide any guidance on the ‘how to implement’ side, with the result that a wide range of implementations occurs destroying the advantage that product standardisation was supposed to deliver, i.e. improved interoperability between the various instances. Now obviously this has not just come to light, but it hasn’t mattered too much until recently when collaboration tools and Service Oriented Architecture started to become important.
In both cases the method is what will make or break the whole benefit case of the product, yet how many large enterprises have got this on their agenda? It’s made doubly interesting by the concentration that the IT software vendors have had over the last two or three years to introduce their own tools, and the comparison to Open Source development methods. That remark should stir up a few comments!