When you design a dashboard, your users are naturally going to be excited about all the things they can do with the new information. However, as soon as they answer the questions that motivated the creation of the system, they will think of new questions that they want to answer, often by going even deeper in the data. It's like they're following the yarn that you've given them to play with, and are tugging more and more out of the tangled ball of data that is your warehouse. When they get stuck, they will pull harder and harder, trying to pull further insights out of the data you have given them. They'll ask for just a little more, like filtering by a new dimension, or going back further in time.
This is a natural part of the evolution of your data dashboard, and is hardly unique to this aspect of software development. However, unlike natural evolution, in which a vast majority of species die off, you'll never have a user come to you and say that a feature, dimension, data source, or graph is no longer necessary and can be depreciated. Imagine the forest where early humans are foraging for food alongside Neanderthals, apes, Dodos, Triceratops, and little legged fish. That would be anarchy! But it's what your data dashboard will end up looking like as each new iteration is simply added to a growing pile.
The solution for this is that you must have taste, exactly as if you were running an art museum. The chief curator at the Museum of modern Art can't spend their entire day saying "This looks great, let's squeeze it in!". The decision is always "Which existing piece goes back into storage so that fresh new material can be visible to our visitors?" It makes for tough decisions, particularly when some of their visitors, and your users, are going to be disappointed by their choices. You can't please everyone, but it's your responsibility to determine which user decisions are most important, which data sources are most reliable, and in which user's needs to be met today. It's important to remember that all these decisions are fluid. Putting something up doesn't mean that it will be up forever, and the inverse is true for taking something down. For this reason, reversing your decision down the road doesn't mean that you are wrong in the first place! A view that is only up for a short time may have simply fulfilled its purpose quickly, like a special exhibit by an innovative artist.
Curation may not have been in your job description (which should be a point in your favor when you ask for a raise at the end of the year!) but you will find that it makes every other aspect of your job tremendously easier. Less complex data integrations will be easier to debug, and entire issues can be sidestepped if a low quality data source is retired before it finally breaks down. Even the users that complain about some specific view being missing will often be able to find their other information that much faster. It's quite possible, in a parallel universe, that they would be complaining about how cluttered the dashboard is! You should certainly seek user input, because they know their decisions better than you do, but at the core of the issue remains a basic principle: You are responsible for the data and supporting users decisions to the best of your ability.
Read the original version on my blog here