loader image

Helping People Understand Data

Helping People Understand Data

Introduction

When most companies set off on their data journey, they worry about technology, they worry about costs…but they don’t often worry about the fundamentals. They don’t consider the fact that they might put in place the right data, serve it to the right people, but they may simply misunderstand and misinterpret it.  

The Idea

Charts, graphics and statistics are hardly new phenomenon, and it’s generally accepted that people know how to read them, and what actions they should take from that information…but this isn’t always the case.

Data is easily swayed with the wrong sample size, the wrong sources and from conflicting truths from different sources. This makes it all too easy to draw the wrong conclusion with potentially massive implications. Here’s a famous example:

During WW2, the allied forces were losing a number of planes to German anti-aircraft fire. Their solution was to reinforce the planes with armour, but with materials being limited they needed to identify which areas would be best to plate. An analysis was needed, and so every plane that returned was assessed and had each bullet hole logged to see which areas were hit most often. 

It was mathematician Abraham Wald that pointed out that the sample size did not include planes which had been shot down and that if they saw a large cluster of bullets on the wings of planes that returned, it actually meant that the bullet fire to the wings wasn’t causing planes to fail. They could have easily reinforced the wrong areas of the planes simply due to an incorrect sample size and who knows what the historical consequences would have been!

The same is very much true today. People get so lost in digging into the data they don’t think about how they’ll use the information, or even if they have the right information to begin with. Data literacy and accuracy are huge challenges facing most organisations today.

In terms of data literacy, there may be a bigger problem with basic numeracy skills then you might imagine. Every few years the OECD runs a large study of 245k people across 38 countries, examining the basic numeracy skills of adults. A 2018 survey found that England was ranked 21st and America 28th out of 38 countries.

A shocking 6 in 10 Americans struggle to “recognise and work with mathematical relationships, patterns, and proportions expressed in verbal or numerical form; and can’t interpret and perform basic analyses of data and statistics in texts, tables and graphs.”

For those who can interpret graphs and charts, there is a huge issue around trust in the data itself. An Experian study polled C-suite managers and found that whilst 98% said that having high-quality data was either “extremely important” or “important” in achieving their business objectives, 40% said they didn’t trust the insights they get from their data and 28% described their current customer/prospect data as inaccurate.

So How Do We Tackle These Issues?

Robustness – Take time to make your data pipeline robust. Make sure you have relevant data for the issue you are trying to solve. We often find that people try and skip ahead or don’t give enough consideration to Quality Assurance at the modelling stage and this has downstream problems with dodgy dashboards later. You need to build time into your planning to allow for maintenance and review of your current models, tables etc, to ensure consistency of numbers across your pipeline. Always peer review your code and be sure to have answers around any disparities in the numbers which may simply be time to timings/when information is updated etc. Without robust data, your end-users will easily get confused and annoyed, lose trust in the numbers and this will make it so much harder for you to help drive data-driven thinking.

Data Literacy – Ensure that you are doing the simple things well. Are your charts are well labelled? Do you have sensible colour options? Do you offer insights as to what the numbers are showing? These small enhancements can all play a big part in helping people get their heads around data and interpret it correctly. Training programmes and certifications are a good place to start but think about fun ways to engage your team, challenges you can set, you don’t want data to become another chore for people.

Data Basics – The first step in getting everyone in your organisation singing from the same hymn sheet is to create a Data Dictionary. You can think of this as a glossary of terms, where you are defining your key business metrics, what they mean and how they are calculated and also where that data point sits. This lays out exactly what your one source of truth is for each of those metrics. You should also consider a central repository for all your data training, how-to’s etc.

Data Stories – Try and bring to life what charts are showing, how it might apply to real customers for example. Don’t just tell people what the data shows, but try and give reasons why as well as advice on possible courses of actions. Data teams still struggle sometimes to go beyond the role of reactive reporting to guiding business strategy.

Conclusion

Never take for granted that people will readily know and understand data. Analytics is a niche field in which few people have a depth of knowledge, and understanding is often made harder by a lot of industry terms and changing technologies. If you want to get the most out of data though, you have to find ways to connect with business users and translate.

173tech can work with your organisation to produce reports and dashboards that are fully automated, bespoke to your needs and easy for anyone to understand. Why not get in touch with our team today?

More Articles

Our eBooks

Our Services

Contact Us

150 150 173tech