Ice sculptures and production lines

Before you start analysing data - ask yourself if this feels like the usual way you go about this work? If the answer is yes, then I'd suggest you stop and start again with more intention, with a clear question that you've refined and worked on for more than just a few minutes.

There's rarely any commercial value in data that was easy to collect and didn't take very long to analyse. If you're not looking for commercial value, are you just looking to fill in your time?

I describe this process as having more in common with an ice sculpture than a production line.

If you looked at how your organisation uses data and saw that most of the effort was automation, you're running a production line. Taking analytical tools from manufacturing and applying these to other industries has failed more than it has succeeded.

If I saw an enormous amount of effort to answer one question with beauty and grace, you're producing an ice sculpture. The sculpture takes planning, and there's risk. You need to know what you're working towards from the first cut with the chisel. You don't have the luxury of 'seeing what's interesting' after you're removed half the block.

You want your data analytics to be like an ice sculpture so that people will engage with the work, not glance over it. Few casually walk past an ice sculpture, and even fewer marvel at a production line.

To deliver on the promise of data analytics, spend more time thinking about ice sculptures and less time about production lines (yes, even if you're in manufacturing).

Why is this hard to implement?

The general risk-averse nature of business means that while we minimise downsides, we miss out on incredible opportunities (yeah, that's the concept of a normal distribution). There are no 'quick wins', or first steps when considering this idea.

You need to be all in to see the potential benefit whilst accepting that what you choose to do may not result in something meaningful. Most of the 'insights' are just a form of summation. Summation is literally just adding items up. Summation is a one-line statement that sounds interesting but means nothing. One example would be - you drove 15,000km last year. So what?

Going to a data storytelling course isn't going to fix a lack of imagination in the work in the first place. Data storytelling is like adding filters to photos. If your photo needs a filter you should spend more time working out how to take better photos than working out the right filter. Conversely, you don't need a filter if you take a great photo.

Now you know the problem - what do you do?

Leave the dashboards, and datasets alone - go and find something worth doing.

Organisations can start by running a workshop where staff explore interesting ideas with someone trained in statistics guiding them through the difficult parts. Most people think the difficult part is the analysis; the difficult part is forming a hypothesis.

For individuals, the starting point might be measuring something they'd never consider because it's too hard, or doesn't seem relevant.

I hear many people talking about wanting to make an impact, but I've never heard someone describe how they know this is happening through data. Leaving a legacy is another example of a vague sentiment that is not measured.

Organisations like to talk about 'providing value', being 'customer-centric' or 'improving productivity'. Show me your hypotheses and we'll go find the data.

All of these questions could be answered if you were willing to dedicate the time and effort to making an ice sculpture instead of trying to make the production line more efficient.