It was Craig Confoy’s presentation on Johnson & Johnson Pharma where I really started to get interested in the issue of where EA sits in the enterprise. Although the “E” in EA stands for “Enterprise”, it seems that most organizations, and J&J is no exception, start out with EA in the IT infrastructure group somewhere. Like many large conglomerates, they had a bit of a mess with five pharmaceutical R&D companies (out of J&J’s 200-odd companies), each with its own IT department supporting 14 different functional units per company, and little alignment between the company functions. Since EA was in IT infrastructure, anything in the business layers of EA, such as business modelling, was done on a project-by-project basis and not shared between business units or companies.
Sound familiar? Almost every large company that I deal with has the same issues: some real architecture going on at the lower infrastructure levels, but practically none at the business levels.
About 5 years ago, J&J Pharma decided to do something about it, and created a business architecture group. There were a few stumbles along the way, such as the use of a (seemingly inappropriate) CASE tool that resulted in business process documentation that stretched over 42 feet at 8pt font — unusable and unsustainable — before they started using Proforma.
One of their models that I really liked was an enterprise data model that could be overlaid with departmental ownership, so that you can easily see how changing any part of the model would impact which departments. I think that this is one of the basics required by any large organization, but often not used; instead, companies tend to replicate data on a per-department basis since they don’t have any enterprise data models that would tell them who is using what data.
This was one customer presentation that showed some clear ROI of using the Proforma tools: they found that systems could be implemented 30% faster (a huge advantage in pharmaceuticals), that the modelling process identifies system integration points and allows them to create standard EAI models for reuse, and that the data models helped meet their regulatory requirements more easily.