Stephen Few’s arguments about dashboard gauges and dials can be summed up nicely in his own words: they "say little and do so poorly". Agreed. But there’s one specific reason that gauges and dials totally suck when it comes to KPIs.
While you might have more than just performance measures on your dashboard, you need to display your performance measures in a very specific way or they will fall far short of their purpose.
Just because you set a target doesn’t mean it’s sensible to compare every week or every month to that target. You have a target because you want performance to improve. Performance doesn’t improve immediately – you need to allow time to change your processes so they become capable of operating at that targeted level.
So performance measurement involves monitoring change over time, and looking for signals about whether it’s moving close enough and fast enough toward the target.
Dials and gauges only show a snapshot of performance. They show this month compared to target, for example. They don’t show if performance is increasing toward target, or if it’s moving away from target, or if it’s refusing to change at all.
You need this context in your performance measures to help you prioritise how much attention you need to give to the initiatives that are supposed to be closing the gap between your starting level of performance and your targeted level of performance.
And because Dials and gauges don’t use this context, they are also incapable of showing you true signals in your measures.
The size of the gap between this month’s performance and the target is NOT a measure of how far away from target your performance is. This method of assessing performance completely ignores the fact of natural, routine variation.
Dials and gauges have you knee-jerk reacting to routine variation in your measures, and fail to tell you when there is a true signal of change toward or away from your target.
The best way of presenting your performance measures in dashboards, which will both provide historical context and valid rules for signals of change, isn’t currently available. It’s what I call a smartline. But these are not available yet in any dashboard application I know of.
If only our dashboard vendors knew more about what dashboards are used for.
Do you agree that discrete measures are as useful as continuous measures? What’s your reasoning? Share your suggestions on the blog.
Very good points!
Understanding TIME in project work – and in work overall – is ultimately
I fully agree about the trends. I don’ t agree about the gauges, they give an immediate snapshot of the situation and sometime it is what you need. Therefore: graphs and gauges, please.
Probably – gauges are best in the situation, where you REALLY have “one-to-one” type of information behind. E.g. in automobile panel a speed meter is a very good one, because you immediately see ONE thing, where the calculation come – namely stye speed. You do not need any trend information in order to ease off the pedal.
But in many business situation, the information which is aggregated from several sources (e.g. in BSC-models) the gauge does not say anything if you do not go into details. Also as a graphical object, gauge can be very “busy” to end-user – if there are many gauges, people miss the message and do not really get want they want. We once made a collaboration survey to our customer with very fancy gauge graphics and the response was: “Very nice, but we did not get the picture”. :(
One gauge, which has deviation type of information (so that only and only when there are some relevant deviations comparing to) and then e.g. trends etc.
I agree with Andrea snapshots are needed – just look at the dashboard of your car it gives you most of the information you need now. But they don’t give the whole picture which is why we add satnav = where we are heading, how far away we are, how long to get there, warnings, etc. As usual there is no single answer, no simple right and wrong.
My previous comment was driven by the drastic assertion that gauges are useless for kpi. As it looks, everyone here (up to now) agrees that the proper tool has to be used depending on the aim and deepness of the figures’ analysis. I will never use a pneumatic hammer to hang a picture on the wall, in the same way I will never use complicate dashboards only to know if our sales are over/below the targets (and it can be considered a kpi). They are useful (dashboards) to investigate details. Fortunately the world is not in black and white.
For those of you arguing for gauges and dials, I encourage you to do some research on XmR charts and the concept of natural versus abnormal variability in a performance measure. Any visual display tool that does not take the context of this variability into account is too risky to use as it leads to improper interpretation. In addition, research bullet graphs and smartlines, which provide a much more space-efficient snapshot of what’s really happening in a collection of measures.
I fully agree with non-gauges and non-dials fraction. Reporting is 99% not for looking on a snapshot. You very often want to know more like trends, comparisons to other hierarchy elements or scenarios.
But the real weakness of those charts is about their scaling. Every element has it’s own scaling like the United States Patent and Trademark Office visualization shows:
Take a look at the 4th row starting with “Final Disposition Compliance Rate”. Is it good? Is it bad? What was it last month? No chance to answer one of those questions and 92% on the right look like 94% on the left. Nice visual effect.
That’s why we at DENSiO http://dens.io decided to build a mobile dashboarding tool which is based on some easy to remember visualizaitons rules (International business communication standard http://www.ibcs-a.org). Additional we think comments help to understand your data.
Andrea doesn’t want to use complicate dashboards. I don’t neither. That’s why layouting is done by the machine. It’s the analyst’s job to put in the figures into well structures DENSiO templates.
http://is.gd/densio shows you how it looks like.
Stacey, sorry for the links – but you asked for dashboard vendors. ;-)
Joerg, DENSiO still seems to focus on point to point comparisons or percentage differences. I’d suggest XmR charts or smartlines give a much more accurate and statistically valid assessment of performance. See: http://staceybarr.com/measure-up/are-sparklines-smart-lines/