Wednesday, October 26, 2005

Customer Data Integration

I tend to describe the work I do variously as Enterprise Application Integration, Business Process Management or Business Integration.

I will shortly be getting involved in a project where the concept of Customer Data Integration will be highly relevant.

As with all things in the IT industry, this appears at first to be a new name for an old concept - in this case, the construction of a single view of a customer. However, Gartner regard it as a distinct segment and I can see why.

The core sales pitch seems to go something like this: "You have multiple systems containing information about your customers and they aren't integrated. Your customers get justifiably enraged when they phone up and this means you can't answer their question or solve their problem. Wouldn't it be great if you could link all your customer data so a change in one system is synchronised to all the others and the problem of managing customer information just goes away?" How could you say no to that?!

Now the interesting thing is that this is a problem that some EAI products also do particularly well (witness the "CustomerSync" pattern in WebSphere Interchange Server).

So what's the difference?

I did some reading on DWL's website and listened to an interview with a Gartner analyst.

The answer seems to be that whilst it's possible to "roll your own" on top of a pure EAI product, there are lots of subtleties and extra opportunities to add value that are most usefully provided by the vendor. Examples would be inferencing techniques to make an educated guess if two customers in a system are the same, pre-packaged content (industry focussed where appropriate) and common services that make sense only if the data being synchronised is customer data.

Interestingly, IBM has recently acquired DWL (hence my choice of them for my link above) so it's obviously something that we also see as a promising market.

I see this as yet another example of vendors taking generic integration technology (in this case a J2EE runtime - the WebSphere Application Server) and adding value by making it specific to a particular problem. The challenge for me as a consultant is knowing when it's best to recommend the use of a generic solution ("build") or the use of a targetted solution ("buy"). My forthcoming project should give me lots of opportunities to experience both first hand and begin to form some opinions.

One other tidbit that caught my eye on the DWL website was their reference to themselves as "DWL, an IBM company". I am, of course, privy to no internal information on why this may be the case. Perhaps it's there for legal reasons to suggest the acquisition isn't complete yet (not sure if that's the case). Or perhaps it means that DWL has become a wholly-owned subsidiary, rather than having been subsumed into our Software Group. Tiny things like that always intrigue me.

No comments: