There are many milestones through the lifecycle of a dashboards. The initial stages are the most important, in preventing headaches or failure. This list will hopefully give some guidance to start the ball rolling on a correct path to success.
1) Avoid waiting for technology waiting for traditional BI projects to begin could take months or even years with delays. Find what technology you have available and start working on some implementations. Once you have started some ground work, the design can easily be applied to other BI technologies with little adjustments needed.
2) Avoid underestimating time and resources – Though a dashboard is designed to be typically compact, it does not mean it will be quick and easy to create and maintain. Expect it to be an on-going, frequently changing process.
3) Avoid static/stale data Having data automatically and frequently updateable is very important to the user and designer. Static or stale data will lead to a false sense of confidence.
4) Avoid limited access Making dashboards easily accessible is ideal. Making it easy for the user to access the dashboard will allow for more frequent uses and feedback; helping keep the dashboard current and useful. For example, distributing dashboards on the web can help keep data current and still adhere to IT protocols and security standards.
5) Avoid choosing the wrong metrics Find the metrics that will end up answering the underlying questions.
- How does each metric contribute to the objectives?
- Is there data that can shed light on the objective?
- Is there a meaningful metric that measures those contributions?
- Are the metrics, chosen, necessary to contribute to the objective?
- Is the metric a continuous organized measurement?
6) Avoid not knowing the goals and objectives of the dashboard. Some examples of the most common goals found in companies can be generalize as:
- Make better-informed business decisions.
- Improve customer interaction and satisfaction.
- Gain an overall competitive advantage.
The goal(s) usually can be seen having these characteristics at its core:
- Accessible to its audience
7) Avoid meaningless variety When starting, keep it simple and aim at 3-5 metrics. Never go above 10 metrics. Too many metrics will likely cause added charts and graphs that add no value to the dashboard.
8) Avoid ignoring the target audience give the audience what they need (e.g. timely summaries, links to supporting details, what actions are applicable, etc..) The usual types of audiences can basically be broken down into 3 categories.
- Strategic (Executives) these users want a summary of the overall health, long term progress and goals, historical data.
- Operational (Department Heads) these users need a snapshot of the current operation, clear meaning, real time data, interaction and link to information.
- Analytic (Engineers, Programmers, Researchers, Lawyers, Paralegals, etc¦) these users need ways to do comparisons, review historical data, evaluate performance, drill into causes, needs feedback, established goals to course-correct.
9) Avoid using metrics no one understands it is wise to avoid any unnecessary teaching of some new metric. Stick to the common and familiar metrics that the audience will know of and understand.
10) Avoid using compound metrics Compound metrics are usually developed by using a formula comprised of a variety of measurements to come up with a single value. (Example: FICO, Klout, etc¦) The problem is that compound metrics masks insight and understanding making it hard to compare and contrast the real underlying numbers.
11) Avoid blindly following easy measurements Easy measurements are great, but can cause users to lose sight of the bigger, more important picture. For example, SAT scores may be a great measurement of what a student’s potential will be at a college, it can also cause a blind spot on the whole life of the student.
12) Avoid complexity Keep it simple. Focus on trends rather than point-in-time comparisons. Keep focus on the handful of metrics and design principles throughout the dashboard(s). Reduce, minimize, or eliminate non-data elements. Try reducing emphasis by changing the size, color, and borders; meanwhile, removing data that does not add insight. Draw attention and maximize the important data element.
Some helpful resources:
- Stacia Misner http://blog.datainspirations.com/ and Dashboard Design Dos and Don’ts www.24hoursofpass.com
- Performance Ideas Best Practices for Business Analytics Professionals. http://www.performance-ideas.com/
- Occam’s Razor by Avinash Kaushik http://www.kaushik.net/avinash
- Tableau whitepaper on effective dashboards http://www.tableausoftware.com/sites/default/files/whitepapers/effective_dashboards.pdf
- All Analytics 5 Ways to Get Effective Data Visualizations http://www.allanalytics.com/author.asp?section_id=1411&doc_id=264280&f_src=allanalytics_sitedefault
- Stubborn Mule: http://www.stubbornmule.net/
- Six revisions http://sixrevisions.com/usability/data-visualization-gestalt-laws/?utm_source=buffer&utm_campaign=Buffer&utm_content=bufferc5b7d&utm_medium=twitter