The inmates and the asylum.
Listening to your customers (or end-users) is a must for BI systems, or any system for that matter, right? If the users tell you they want a feature you should give it to them - and quickly if possible. Well, no. We often talk in our lectures that the role of an analyst in a decision support setting should be - what we call - an active one. The analyst shouldn't just be a passive interpreter of the user's requirements, translating everything the user's ask for into system designs. They need to take into account what the user community wants, but every now and again they need to intervene and change something. They might need to explain a better way of doing things, or explain that what the users are asking for is silly is somehow or another. All this takes great skill and care not to seem like a typical know it all IT person.
Here is an unusual example of taking what your user community wants too far - and sadly this is going to have a major impact on just about everybody working in BI. There is a feature in the upcoming version of Excel that is just plain wrong.
Microsoft has gone out of its way with its next release of office to take usability seriously. They have tried to embark on a brave remake of the product set. One of the potentially great new features they are including in Excel is the ability to quickly create bar charts that sit (ala Tufte's sparklines) in the spreadsheet cells, next to the numbers and text, not as a separate chart object. Great idea. However, the feedback they got from focus groups is that users don't like the look of these charts when one of the cells is 0 or very low compared to the others, so they have adjusted the bars displayed so there is a minimum that shows even when the quantity is 0. Oh dear. For some example charts and a discussion visit the (excellent) blog at Juice Analytics.
1 comment:
Thanks for the Juice call out.
I couldn't agree more with your point that many requirements are counter-productive. Sometimes business folks will ask for something purely out of curiosity, even if the results aren't the least bit actionable. Worse, people will feel that they should provide a laundry list of data asks simply to appear data-driven.
Post a Comment