Best Practice for developing dashboards

[barrysaab]

I  have tried to find out best practices to be followed while developing dashboards with both relational and dimensional data.Any ideas would be highly appreciated.Thanks

[Dineshkumar]

Hi,

Best practice would be reusing quries (query reference), layout component reference, less javascripting. Show summary data would be a better approach.

-Dinesh

[charon]

Also, reconsider the requierements/ requests and evaluete them in order to use the best possible fitting tool.
You can build dashboard within a cognos connection portal page, in Report Studio, Workspace Advanced or Workspace or even with Active Reports.
Which tool and technique to use depends on the set on requierements.
Good luck :)

[barrysaab]

Thank you all.

[CognosPaul]

Terminology can be a difficult thing. I tend to try to stick with Stephen Few's definition:

A dashboard is a visual display of the most important information needed to achieve one or more objectives, consolidated and arranged on a single screen so the information can be monitored at a glance.

A dashboard has no prompts, presents only the information needed in an easily digested and instantly understandable layout. Many visualization experts talk about the ink to information ratio. Chart borders, background images, harsh colors, and extraneous highlighting and animation are all things that serve to distract and hide information from the user. Instead of highlighting, small indicators can be placed to draw the users' eyes to information that needs their attention. Section separators are not as important as people tend to believe - we are very good at seeing invisible lines that separate groups (the gestalt principal). See the Kanizsa triangle as an example of seeing invisible lines (read more here: http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0040106)

Workspace and pages with prompts are interactive analytical applications. Their design should follow the rules of dashboards, but should take into account where the user prefers to look in order to place the prompts. The data should always be the focus. I tend to put the prompts on the right side of the screen, with a slider to let the user hide or show the prompts as needed.

There are thousands of books, articles, and research papers on how to build effective dashboards. I hope you like reading. Here are just a few:

• Common Pitfalls in Dashboard Design, Stephen Few
• Visual and Interactive Analytics, Stephen Few
• Designing with the Mind in Mind, Jeff Johnson
• Envisioning Information, Edward Tufte
• Information Dashboard Design: Displaying Data for At-a-Glance Monitoring, Stephen Few

That is just the short list. I strongly recommend going over Stephen Few's and Edward Tufte's websites for more.
You can tell - I'm a lot of fun at parties

That's just the design phase, not related to Cognos.  On the implementation side you have a many options. To narrow down how to proceed, you should determine exactly what the users' needs are.

• How long is the user willing to wait for his data after clicking on the link?
• Will the user need to view the data in any other format?
• Will the user need to view the data in any other applications?
• Will the user need to "play" with the data, or be able to filter by a list of attributes?
• How much wood can a woodchuck chuck if a wood chuck could chuck wood?
• How many people will actually be using the dashboard, how important and how much time should be devoted to it?
• Where is the data coming from? Is this a single data source or many?
• Do you need to use JS to give them certain functionality? How many developers know how to write JS? What browsers are being used?
• Does the user want an actual dashboard, or is he looking for something else? *Be diplomatic with this question. Users' may take offense if they catch on to the fact that you think they don't know what they want
• Are they asking for this because they don't know that other possibilities are available?

If the user wants a 20 page document in his email every morning, it will involve a different design methodology than if he's asking for a dashboard combining data from disparate parts of the organization. If he's looking for prompts, then using multiple cubes in a report will be difficult (though not impossible). If he wants instant results, then look into caching and scheduling.

When interviewing the user, make sure to keep a pad and a pen. Take notes, and draw sketches of what you think the user is asking for. The last question is the most important, users can only ask for what they know. If something is difficult, inefficient, or goes against best practicies offer alternatives. "Instead of a 3d animated pie chart, how about a static column chart?", "instead of showing a graph showing a constant 100%, if you need to see outliers how about 100 minus actual?"

[barrysaab]

Thanks a million,PaulM.Very informative indeed!

[charon]

Awesome answer, thank you Paul. Love learning new stuff in here  ;D