• Control.  When you've got Excel spreadsheets being used by traders, there's absolutely no control.  Tools like ClusterSeven can help here and are a neccessity if you're going to stick with Excel.  Know that traders can and will screw up their formulas and this can have massive repercussions.
• Maintenance.  You've got to get DLLs/XLLs onto a client's machine if they're going to run the complex calculations that they need.  Then you need to maintain the different versions of these libraries.  Yuck.
• Performance.  If you're doing work in Excel, it's likely that all of your calculations, etc. are running on the client side.  Yuck.

Even with these points, it's hard to get the traders onside for a move away from Excel.  A couple of ways that we're getting around this are:

1. Build a flexible framework that allows for rapid deployment of new features and products (often with an XML based solution).
2. Change the game by creating a system which does not look and act like Excel but one which does what the trader needs to do in the most user-friendly way possible.  It helps if this is visually pleasing (we've been working with WPF recently and it provides capabilities to change the game).

Number 2 is hard and takes considerably more time than just replicating Excel functionality.  It requires deep knowledge of what the traders are trying to do and a time investment from the traders and technology.  Ultimately though, it's the only way that a system which is flexible, usable and better than an Excel spreadsheet will get produced.

However, there are still people who believe that Excel is best for trade capture (don't count me in that crowd).