Vanity Dashboards

November 25th 2014

Vanity Dashboards (or Pat-Yourself-On-The-Back reports) are insidious time-wasters. They are reports whose sole purpose is to prove that things are happening, with only a vague sense of why it matters that those things are happening or what should be done if they are not happening.

How can you tell a vanity dashboard from a regular dashboard? The key is actionability. If the end-user is not going to do or change anything based on the dashboard, then it is a vanity dashboard.

The trouble with vanity dashboards is that someone important is usually behind the request. If it was a regular Joe asking for a vanity dashboard, the answer would be no. Regular Joe's do not get vanity dashboards. Managers and executives get vanity dashboards. For that reason, they are hard to say no to: the best thing to do is recognize each of the below symptoms and address them one by one. By treating the symptoms, you can turn a vanity dashboard into a useful tool.

The first tell-tale sign of a vanity dashboard is ill-defined metrics. For example, measuring the number of cases or the length of each call without ever stopping to ask "Are more cases good or bad? Are long calls good or bad?". For every metric on a vanity dashboard, be sure there is a clear target (KPI) and a plan in place if the target is not being met. If there is no target or if there is no clear plan for when the target is not met (other than to excuse it away), then there is no place for that metric on a dashboard. Ask why five times.

The next sign is metrics with side-effects. If I am being measured on the number of cases opened, then I will open cases for every little thing. If I am being measured on call length, then I will finish up the call faster and let the person call back if they still have a problem. Watch out for these unintended consequences. Make sure every metric aligns with your company vision. If your mantra is "Customer Service is our Priority" then a metric that causes people to hang up the phone quicker is not what you want. Be prepared to ask the dashboard requester about any side-effects you can anticipate.

Another sign of a vanity dashboard is rogue master data. For example, perhaps you need a hand-maintained mapping table to map users in one system to users in another. Since the master data is only needed for this dashboard, it is likely you will end up maintaining that master data. Avoid this as much as possible - ongoing maintenance of master data takes up a lot of time. (Especially since the maintenance will need to be done before the report is run, meaning you will also have to run the report manually too). If the mapping is not easily automated, and if it is not vital to your companies core mission, then it should not be done. For example, instead of mapping users in two different systems together, have end-users select who they want to report on. It will take them much less time to find the 10 people they care about than it will for you to map two systems together.

Yet another sign is complex calculations. Whenever you have a dashboard that you are holding people to, the temptation is to account for every single exception or rare circumstance. Each individual situation can sound perfectly reasonable, but the result is often complex metrics that are not easily tied back to reports from the source system. And when people see numbers that do not match what they expect, they lose trust in the data. It is better to account for rare circumstances or exceptions in your KPI target (i.e. do not aim for 100%), than by writing pages of code in the underlying measure.

The next sign is a request for a very high level dashboard that has no drill-down capability. If there is no drill-down, the temptation will be for end-users to repeatedly challenge you on the accuracy of the numbers (especially if the underlying calculations are as complex as outlined in my prior paragraph). You do not have time to be constantly proving the dashboard is right, so it is in your best interest to build a way for end-users to understand and validate the data themselves - even if they do not ask for such functionality. (Bonus: If you have made a mistake, it will be spotted much sooner.)

The final sign of a vanity dashboard is presentation distraction. The dashboard must look good. It must have everything on one page. It must not require any effort to consume, regardless of the cost of the effort to make it effortless. The focus is on form over function. In my opinion, a good dashboard starts at the grass roots: a KPI to measure and a simple report that helps the people in the trenches improve that KPI. It is quick to build, low maintenance and quick to see a return on investment. If it proves useful, it gets rolled up and a dashboard evolves naturally.

Keys to a Good Dashboard

  1. Ensure every metric is clearly defined with a goal.
  2. Ensure metrics do not have unintended consequences.
  3. Watch out for master data you have to maintain yourself.
  4. Avoid complicated calculations when simple calculations will work.
  5. Ensure end-users have tools to drill-down into the data and validate it themselves.
  6. Focus on individual actionable metrics rather than the overall presentation.

Vanity dashboards are expensive cars - they look good, but they are expensive and don't help you get anywhere faster than a cheaper option. No executive wants you wasting weeks or months creating a dashboard with limited ROI, but it is your job to identify the dashboard as such and suggest a better way.

Go back...

Visitor Comments

Post Your Own Comment

Note that comments may not appear immediately on this site.

Commenting is not available in this channel entry.