It’s probably appropriate for me to be writing this post as I am (as anyone who knows me will attest) at heart a cynic. Or a realist. Or just old and grumpy. Take your pick. But I have seen a lot of dashboards come and go and there are several themes that I do think are worth calling out before we all run off and start building one.

1) The “We must have one” approach.

Ever popular, this one. And has really endured over the years since dashboards first started getting popular. “The board needs one. A view, you know in real time, for DECISION MAKING. What should be in it? I don’t know, THINGS. You’re the data man. You’ll think of something”.

Moral: Who is using it and for what? Give me an example of when the MD will actually use it. What is that situation? Go and talk to them.

2) The “Oooooo” factor

Also known as the Fisher Price Activity Centre approach. Contrary to popular advertising lore, you don’t only get an “Ooo” with Typhoo tea, you get them from a dashboard to0.

MetaMetrics Blog - Cynics View of Dashboards

“Look, if you press this, that thing goes up and down AND it turns GREEN. Isn’t it GREAT?” You may or may not believe me, but I have been in presentations with quite senior executives who have been willing to sign a cheque there and then on the basis of this. When someone (usually me) says, “Yes but what is the data behind it, what does it tell us?” The response is often “…and if you slide that thing that dial rotates! Sorry, what did you say?”

This really started to come in during the 90’s with tools like Xcelsius. Literally any muppet (including me) could build one of these things with interactivity, gauges, dials, colours. And they really worked.

MetaMetrics Blog - Cynics View of Dashboards

Except they kept breaking. And after an hour of playing with them, everyone got bored.

Moral: Hack it in Powerpoint first. And if you are buying something, be that annoying person who doesn’t let style triumph over substance. Or just be happy pressing buttons.

  • Details, details

So, we want a dashboard. We bought some software. Excellent. Let’s put a dial gauge thing on it for sales. What sales are we going to put on? Volume, Value? OK both. UK or GB? We report GB to the Investors, but we don’t get the same granularity of data for GB. Hmm… Let’s leave that decision for now and come back to it. Oh, and we need to include orders. Committed orders? Or Pipeline? The thing is that we don’t get the orders data until 2 weeks after the sales data because it comes off that other system. So, to be consistent we have to always be 2 weeks late with the sales data and everyone will read the normal sales Excel report anyway and not bother with the dashboard.

Now, we need to set the red, amber and green zones to show The Board when to panic or not. Where do we set those? Is below last year Red? Or below target? Well “really bad” is when we are below both. So, if we are just below one, is that Orange? What if we are above target but below last year, what’s that? And green is..? Up. Right, how much? Well, quite a bit definitely.

One dashboard I actually used when I worked in a large company incorporated traffic lights. Red, Amber and Green. Except then they added Blue. And a bit later Black. I’m not making this up.

Moral: Is the data all aligned enough to make this work? Will anyone actually use it in anger, or will it be out of date before it’s populated? Do a data audit first.

Another Moral: Several people have to make lots of very, very, tedious and detailed decisions. Do they know this?

  • It’s a THING, data visualisation.

Then there came books. Big expensive books. I have one propping up my PC screen as we speak. There were Gurus. Weird graphs started emerging. And rules. NEVER use Ariel font over 12 on a pie chart UNLESS you are using exploded segments.

Moral: Make sure the book is the right thickness or you will strain your neck looking at your screen.

  • Dashboards are SO 2000’s, It’s all Infographics now.

Go to Step 1.

Where does this leave us? Even I, in my cynical manner, can’t say there are no useful dashboards. We’ve just built something very similar to help our clients explore their modelling results. And it was painful. This really is one of those cases where if it’s not painful you definitely aren’t doing it right.

The steps we went though were much as I’ve outlined:

  • Who is going to use it? No, I mean actually use it, really. And what for?
  • What data do we have? How aligned is it?
  • Hack it in Excel or PowerPoint
  • Go easy on the knobs and sliders, it’s a world of pain
  • “It’s a journey Bob, I’m just happy to be here”. You won’t get it right first time, that’s OK, we haven’t. Using it will be the best form of testing and debugging. Get a small pilot group to use it first and tell them to try and break it. People do things you wouldn’t possibly imagine given the opportunity.
  • Hire a good developer,  and be nice to them.

As to whether our dashboard is one of those useful ones, we’ll let you decide. Get in touch for a demo to see it in action.