Power in Data

I’ve spent a lot of time over the past year upskilling myself in data. And I’ve really enjoyed being able to apply new skills. And it’s given me new perspectives too.

Slow Down

In terms of a biggest single learning point, I’d say it has been to slow down. So, when looking at data that means: Understand the data, what it contains and what it doesn’t.

Do you understand all that is in there? Have you decided what you are doing with missing or erroneous data? Have you got a way to clean, transform or scale the data?

What do you need to report on and how do you need to prepare the data for what you want the outputs to be? How are you going to check the quality of your processed data?

All of this takes time and thought, whereas my instinct used to be to get stuck in and start getting results out ASAP. This kind of approach tends to mean more rework, and every bit of rework introduces more potential error.

Slowing down gives you different viewpoints, allows time for different approaches to be considered. Ultimately it saves time. It’s an attitude I’ve brought to lots of other areas of my work

All Aspects of Data

I’ve learned about visualisation and storytelling with data. The tools, processes and programs I need to use to analyse data. The ethics and bias I need to be conscious of with data. I’ve learned a huge number of statistical techniques including the ones you’re possibly familiar with like correlation and regression, as well as some machine learning techniques that you maybe aren’t. And, at a very basic level, I can now do a lot more with Excel!

I thought I was good with data, and I wasn’t bad by any means, but there was a lot more I could unlock.

Data in The Real World

What I’d like to share is a very simple but powerful example of what I’m now able to do.

I’ve been analysing an engagement survey, and I did all the things one would normally do. I visualised the results as charts, and made them into an interactive dashboard, that protects the anonymity of small groups of people.

I showed how people had responded to each section of questions. I showed the top and bottom questions, how questions had changed since the last survey, and I made comparisons by the different demographics we held about the participants.

As well, I analysed all of the free text responses. These were in the order of hundreds, so I was able to manually categorise them, and use AI to accurately describe the categories back to me. And then I could use that to show how different sections of the workforce responded too.

So far, so interesting. But all of this just shows you what’s more or less then ethers, or what’s different or changed. You can make good assumptions about what to do, but it’s harder to prove why should prioritise spending time and money in response.

The Next Level

It’s the next stage that made the difference.

I had grouped questions into sections: Leadership, Enablement, Communication etc> but the aggregate scores of these were pretty tightly bunched, when there was a wider spread of individual question scores.

That told me that our sections don’t describe how people think about the experience of their work.

So, the task was to uncover how they DID feel about work. Or in, other words, what questions do they group together, to give a better description of the working experience and what’s important to them?

It’s a statistical exercise to find “components” of the survey. And being a statistical exercise, you get a number of possible answers, with some statistical clues to which might be most useful. What is needed is that human eye to review and understand the right number of comments to make sense of the data.

In this case it was four, which broadly amounted to, in order of strength:

1) The practical experience of work: manageable workload, support, safety, appreciation and basics to perform

2) A theme about confidence in growth and having a voice

3) The social and cultural experience of work: inclusion, trust, respect and personal connection

4) A theme about belief: trust in leaders, strategic clarity, values alignment and pride

And you can see they “ladder up”, kind of like a hierarchy of needs, you can’t go to the next step without having the one before it.

Understanding Personas

Having done this, which simplifies the data from many themes to a handful, it then becomes far easier to look at groups of people or personas – or in statistical terms clusters. People that respond to those themes in the same way.

It’s another statistical process, so again, that only takes you so far. You get good hint of how to form the clusters, but the ultimate decision is a human one.

Now in this case the clusters came out really cleanly, and broke into three chunks: those that are positive, negative or neutral. Very often, it’s a lot more nuanced, Cluster A might be people that are positive about Themes 1 and 2, Negative about 3, but neutral about 4. And so on.

But the cleanliness of that result allowed me to dig much further, more easily.

Since people in each cluster (pretty much) “vote” the same way on (pretty much) all topics, we can find out which topics makes the real difference?

Practical Conclusions

And it’s that final analysis that meant I could show that it was theme about the practical experience of work that had by far the greatest effect on scores. If you scored low here – and on a specific handful of questions – you’d score low on everything.

Or, in other words, changing your experience on that handful of topics, could transform your whole experience.

From a fug of possible responses, we now have a very tight set of clear actions, grounded in provable truth.

Many to Few and the Reason Why

And that is what my work is about: many to few.

Employers know there are many positive things about them that they can promote as part of their employer brand. I show the few they need to showcase.

Employers also know there are many things that they can improve about the experience of working for them, and why they should be doing some of them. I show them the ones that will make the greatest difference to the most people.

(As well as: how you can best improve communication, how to identify elements of your culture that most inhibit, or are further away from where you desire to be.)

But it wouldn’t be so valuable, if it was just about simplifying. It’s got to be about effectiveness. It has to be about choosing those fee things to focus on for the best outcomes.

I have to demonstrate the Reason Why you should focus on this handful of ideas.

And - alongside a critical sense of judgement for how humans respond and behave - it’s as data-led, insight driven approach that I believe will best get you there.