Though tyou might find this interesting…you have avoided all those that you can I think…don’t agree with the last one…

Sen bunlar? okuyor musun?

Thanks for an interesting approach to this problem. As CCF in my Co (industry of plastic products for automobile industry) I am a responsible head of BI projecet. on of the main part is how to produce several dashboards for a several levels of mgmt.
one detail is - drill down technics, because the users needs very often detail informations, specially about “red light” indicators and figures.

I would agree with what Cathy says…should not be sharing all possible detailed information..apart from the mandatory ones. The crystallizing of KPIs for dashboard I feel many a times gets layered with (static extracts of data -wishes/wants). However feel the wants/needs for KPIs should be seggregated based on descriptive and predictive statistical results. Typically descriptive KPIs should be for process owners whilst predictive KPIs should be decision making bodies only with drill downs for detailed information . Exceptions could be made for some of the process centric ‘what ifs’ for process owners. Regards, Shri

Hello all,

First I would like to thank you for your feedback.

Kathy and Shri,
I concur with you in a sense that sensitive information can not be shared with all users, but I personally see a change with giving only descriptive KPIs for the process owners. Predictive KPIs could let process owners to actively collaborate and be part of the decision process, of course, there are confidentiality issues to solve. But at the end, some organizations are taking chances by empowering people with information and the means to play/research with by themselves.

True that initial requirements for a dashboard are many times generated by static sets of data, but in my experience, quite often BI original solutions evolve and transform to offer information users more dynamic ways to explore data. Why don’t we do this from the start?

Please let me know if I’ve address your comment or let me know you valuable thoughts.

Best regards.
Jorge García

Dear Reader,
That’s my point-of-view about the pitfalls:

About Pitfall 1:
Dashboards are designed in several ways not only itself but how they are related to each one else. It’s used to have more than 1 dashboard, and dashboards that are expansions (drilling down, second level, organizational level and so on).
The top Dahboard should be designed according the metodology of process management you are running one and according the user. Of course if you are designing a Corporative Dashboard, there will be a top level one. If you are using Balanced Scorecard, at least 4 KPI will be used, however if you regard for each kpi, actual position, last month and last year then your main top dashboard related to only KPI will have at least 12 visual areas.
Of course you should regard what is the hardware (graphics card, resolution amd monitor) you are using to display your dashboard.
Then, I think it’s a little bit more complex the definition to avoid pitfall 1. And a start point is the GRAPHICAL DESIGN and presentation hardware you will have (the user itself).

About pitfall 2:
Here you will have to count on two kind of technologies: pull and push technologies. They will be applied according the usage of the dashboard. If you are monitoring a process in near realtime or realtime really, then a instant position should be passed and a pull technology should be used as the most effetive way to enable the information to the consumer. Here the point is to deliver the information.
However, if the dashboard is based on a yearly view (e.g. sales performance at the last 12 months), than it doesn’t care to use a pull technology, because the point is not the delivery process but the regular usage into the month.
Then the timing of information will determine the most appropriated technology to be used (pull or push), the method to deliver the dashboard.
Another aspect is to measure the usage of it. If you are using push technologies (like web portals), the hit on page can be a measure. the hit from an IP can be a precise measure to determine if the user who should use really used it. I think it’s far more complex to avoid this pitfall.

About pitfall 3:
Companies must decide if they want their users playing with colors or working with the head, thinking about the business. If they are worried with productivity and higher rates of results due to the investments realized on BI solutions, then, I think they must provide the required dashboards a very attractive design, end point. I do not agree the user should waste his time in playing with customization. However I do agree that, not with dashboards, but with analytical interfaces, they must have freedom to choose and mount on their required analysis at the way they need. Dashboards comes from static interfaces. Analytical interfaces comes from dynamic interfaces. Static interfaces supports decision process, analytical interfaces supports analysis then decisions. I never have seen a C level meeting where the people desired to lost time with analytical discussions and interactions in meeting time. They come ready to discuss a point and oriented to make decisions. However, at the operational level, dashboards are used to show how the things are doing. An analytical interface is required to enable the supervisor to investigate a tendency or a non conformity in place. The usage are totally different, so is different the interface for showing up consolidations and details about the process. What kind of customization we are talking about ? Beauty or utility ?

About pitfall 4:
Dashboards are not graphical analytical interfaces. Dashboards are static interfaces. Dashboards must be connected to analytical interfaces to support simulations and investigations. Dashboards are designed to support process monitoring (Validation or verification). DATA MINING needs full set of information. The people who works with data mining have different role on business. There are three different things: Dashboards, Analytical Interfaces and Data Mining interfaces. Data mining has no commitment with a session of work, because he is trying to identify something very interesting. However, Analysis is oriented to identify the best or worst case or the wrong or right thing, it’s centered at a clear and well defined conclusion that will support a decision. Data mining is oriented to mine data and find the hidden opportunities in a huge amount of data …

Conclusion

Very good article and I think he/she have a donne a great job in openning a discussion about pitfalls in dashboard design, conception and usage.

Excuse me for the extended text …