The dashboard development process is more difficult than it appears. Dashboards must be informative but not distracting,...
quick to read but allow for drill down into details, and with the appropriate information. A good rule of thumb is to have a single screen communicate the information you want to convey, and make it so the user can understand the message in 30 seconds.
Who is your target audience?
First define your target audience. A CEO will be interested in overall KPIs for the business including sales trends, expenses, net income and stock price. A sales VP would be more interested in sales trends, top five sales people, bottom five sales people, pipeline measures, etc. If you display information irrelevant to the audience, you'll not get their attention.
Ideally, there should be no text on the dashboard aside from graph labels or headings. If something needs explanatory text, it is too complicated and needs to be revisited. If readers are interested in more information, they can drill down to more detail.
Allow drill down
Because the information presented should be just an overview, interested parties will want to drill down to details. Where appropriate, two levels of detail may be added: a superficial level made available by hovering the mouse over a key graph or figure, and a deeper level accessible by clicking.
Obviously, this implies there is a system behind the dashboard that can provide the type of additional data your users will be looking for. Make sure to include these screens in the original project plan and budget because they can add significantly to the cost of the dashboard development process.
Don't waste space
Many dashboards put the company logo or something else uninformative in the top left corner of the screen. This is the first thing the user will look at, so it is the most important location. Assuming users remember what company they work for, there is no need for a logo. Put the most important item here -- the one thing users should see above all else.
Real time or close enough
Keeping in mind that the dashboard is meant to be a quick overview of the status of things, consider whether some of the information can be less than real time. For example, do you need to show sales up to the second, or are sales statistics from the previous night good enough? If your dashboard allows click through for details, then users can always pursue real-time data that way.
Asking these questions opens the possibility of collecting data in a nightly job and storing it in a database. This can improve system responsiveness and decrease the load on other systems that are probably more critical during the day than at night.
Learn some key requirements for building business intelligence dashboards
Is the "single pane of glass" dashboard a myth?