Why Gauges and Dials Don’t Work for KPIs, Take Two

May 6, 2014 by Stacey Barr

Measure Up reader, Rick, opened up this debate again on my first post about whether gauges and dials can work for KPIs with this claim: “When it comes to KPI’s that are not time related gauges still can give fast insights to performance.” And a few other readers agreed. But I still don’t.

What KPI isn’t time related? Well, a few examples given by more Measure Up readers suggest:

  • “You can measure your current percentage and compare it with a target, which is not time-related.”
  • “In our line of business we also offer solutions for Trading Companies, which are highly interested in their (long-term) risk position. They want to know if it’s good any given moment in time, and not compared to previous or upcoming periods.”
  • “There is a difference between snap shot performance indicators and those that show trends and they are to be used for two completely different things. When driving, I want to quickly see how much gasoline I have left, and only want to see the trends when analyzing the performance of my car.”
  • “A good dial, representing the degree of timely progress on the objective stated in a given month (period) gives a shocking first visual input…”

Even Bernard Marr’s story about how jet fighter performance makes use of gauges for real-time immediate decision making seems to support these views.

But is immediate decision making the same as performance measurement?

I want to focus on this assumption that there is such a thing as a KPI or performance measure that is not time-related or where historic information is not relevant.

When you want to make an immediate decision about something as operational as whether or not you need to put fuel in your car, take advantage of stock market fluctuations, or keep your fighter jet in the air, you are working IN your process.

You’re doing the work itself. And often, immediate information is necessary for you to do the work well. But that information is not performance measurement.

Performance measurement is about working ON your process. It’s about taking a broader view to understand your process’s capability and how far that capability is from your targets. You can’t do that with snapshot information.

You can’t do it, because to get a measure of process capability, you need to understand the pattern of variability in performance over time. You also need to understand a measure’s pattern of variability before you can know how far from target it truly is.

Knowing what a particular statistic or measure is at a single point in time does not tell you this. In fact, those who rely on point-in-time snapshots as indicators of performance end up tampering with their process and making performance worse.

And on whether gauges and dials are suitable for displaying anything, let alone performance measures, you really should be taking Stephen Few’s advice. He’s given more time and focus and intellect to understand these things than anyone else I know of.

TAKE ACTION:

What are your thoughts about whether there is any such thing as a snapshot KPI or performance measure, where variability and change over time don’t matter?

Facebooktwittergoogle_pluslinkedinmailby feather

Speak Your Mind

Your email address will not be published. Required fields are marked *

  1. Larry says:

    Very helpful differentiation between working “IN” the process and working “ON” the process. Of course we are all aware of the processes within processes or sub-processes, but many of the things of concern are out of the “immediate” time window and we cannot know from snapshot measures what the consequences may be for the total process outcomes. Snapshot measures are important for IN process control! Performance Measures are another thing! Yes!

  2. John Hunter says:

    All performance measurement involves time, because we don’t exist in a series of discrete instant events – life consists of an on-going process. I mainly work in the stevedoring industry and a primary measure of performance is container crane handling rate – containers handled per hour but what does “handling” mean? The import container doesn’t instantly materialise on the wharf – we don’t have Star Trek “beam me up Scottie” technology. There is obviously a process involved which takes time, so it is impossible to measure performance without considering time e.g. in this case, the time taken for one lift cycle.

    A dial guage could do this e.g. show the time taken for the last recorded process e.g. the last lift cycle was 60 seconds so the performance is 60 lifts per hour. However, that is meaningless as the next lift cycle could take 2 minutes, then 90 seconds and so on. The end result would be a performance that is fluctuating. Anyone acting upon the flickering of a dial risks damaging the whole process by constantly tweaking the operation based upon insufficient data.

    In the case of the ship, instead of measuring what has gone on before, why not measure what we have to do to achieve our target? For example, if we have planned to handled 30 containers per hour over 8 hours (=240 containers), don’t measure how many we’ve handled, but how many we still have to handle and the rate we will have to achieve from now to reach our target. A simple bar chart could tell us that, showing us how far we are away from the target, in much the same way that TV screens show a line across the pool where the swimmer has to be to break the record.

    The point of a dashboard should be to convey performance information that can be easily interpreted and used by the observer to drive performance. It can be black and white text for all I care, as long as it helps me.

    • Stacey Barr says:

      John your examples are SPOT ON! Thanks for sharing them. You have made the case that natural variability exists in any and every measure of performance, and we ignore that at our peril (and sanity).

  3. Greg John says:

    Thanks, Stacey, for nudging me out of my comfort zone. ‘Very thought provoking.

  4. george says:

    We use traffic-light indicators to show if our radar stations are working properly or not – these are useful for pin-pointing a problem in near real-time. This is, as you pointed out quite clearly, Stacey, IN-PROCESS. We can log this information and use it to create a time series which can be used to build availability information (which is a KPI we have for meteorological radar), but it is only part of the story, because the radar might be off the air for scheduled maintenance or other reasons…
    You nailed it… while they may be useful for real-time and near real-time display of in-process data, dials can not provide good trend indications.

    • Stacey Barr says:

      George, that’s another good example of how dials ignore natural variability and it’s the natural variability we need to understand BEFORE we can see signals. Thanks.

  5. Marcelo says:

    “And on whether gauges and dials are suitable for displaying anything, let alone performance measures, you really should be taking Stephen Few’s advice. He’s given more time and focus and intellect to understand these things than anyone else I know of.”

    I could not agree more, just because Stephen’s “Information Dashboard Design” is the BEST technical/management book I’ve ever read.

    The only thing Stephen condemns that I cannot get rid off is the traffic light system. Take a look at the stock exchanges dashboards all over the world – I don’t know a single one which does work with the red/green colors. Do you?

    • Stacey Barr says:

      I’m okay with the concept of traffic lights but I like Stephen’s idea of moving away from red/amber/green. Colour blind folks aside, they just look too busy. Mono-colour scales work more nicely.

Connect with Stacey


Haven’t found what you’re looking for? Want more information? Fill out the form below and I’ll get in touch with you as soon as possible.



*We respect your email privacy.
PO Box 422
Samford, Qld, 4520
Australia
Stacey Barr Pty Ltd
ACN: 129953635
Director: Stacey Barr
Simple Share Buttons