Size Doesn’t Matter… as Much as We Thought

This post is part of our Big Ideas series, a column highlighting the innovative thinking and thought leadership at IIeX events around the world. Isabelle Marchand will be speaking at IIeX Europe 2019 in Amsterdam. If you liked this article, you’ll LOVE IIeX Europe. Click here to learn more.

---

A bigger dataset doesn’t necessarily ensure more insights however the way a data journalist interprets it always does. 

Jerry Friedman, a Stanford statistician, recalls a time where the size of the dataset mattered more than its content. He noted during the 1977 conference ‘‘Large and Complex Datasets’ that ‘the thing that you were supposed to be impressed with was how large speakers’ dataset is.”[1] With the rise of Big data, what we considered a strength 50 years ago has become a blight nowadays, as the colossal amount of information available has made it difficult to extract any meaning from it. Unless you approach the challenge in an investigative manner.

When ‘correlation doesn’t imply causation’, how are we to discern meaningful insights and avoid picking up on spurious patterns? Ex trader and essayist Nassim Taleb puts it this way: “The problem – the central issue (with Big Data) – is that the needle comes in an increasingly larger haystack.”[2]

Finding the proverbial needle or in this case those actionable insights that will make a business thrive requires proper investigative work. It requires to understand the why of the dataset: why is that pattern a relevant one, and not just a fake correlation?

This is where data journalism, a still relatively unfamiliar subject in business, comes in.  Data journalism, as the Bureau of Investigative Journalism defines it, is a “Journalism that uses technology to access, analyse and find stories in data and then tell the stories of the people the data represents.”[3]

The data journalist is the data equivalent of an investigative journalist. A journalist of investigation will interview various people and piece out every bit of information in order to uncover the truth of a story. He will then communicate his findings through an article or a TV report, using his talent to resume a complex inquest into a simple story that his readers will seek to read / watch.

Similarly, a data journalist will investigate data by crossing the numbers from an initial dataset to a set of internal and external variables, in order to uncover meaningful patterns of behaviours. He will then share those insights in a format that audiences from all sides – both data and non-data literate alike – can understand and act upon. A data journalist’s basic methodology can be resumed in four main steps:

1. Choose the right dataset.

This might sound obvious but it’s a crucial and often overlooked step. Finding the right dataset is like selecting the right source to get much-needed information.

For instance, the reason why retail sales have dropped in August may not actually be found in a sales spreadsheet but in a weather forecast data set instead as warm weather can also hurt sales.[4]

2. Contextualise and compare the initial data.

This point links to a fundamental question: what is an insight? An insight is a piece of information, which, once put in relation to a wider intelligence base provides a valuable indication on a pattern. A piece of information on its own is not an insight. Therefore contextualizing and comparing internal data to internal and/or external reports is an important step when searching for insights.

For instance, comparing summer sales data to previous years summer sales, to competitor’s sales, or to sales in countries with warm weather, could be a start.

Understanding which particular product(s) have undermined the sales figures, and which ones have contributed to their benefit is also key – for instance a flourishing of ice cube bag sales would not be surprising during the summer. However a plummeting one would raise questions on the product itself, its supply chain, its reviews, its competition, etc.

3. Visualise the findings in order to share them with a target audience.

This step is too often overlooked, but an insight cannot be actioned if its intended decision makers do not understand it. Data visualization, with its extensive range of charts, captions and titles is of great help to get the findings across.

 4. Storytelling.

Once the three previous stages are completed, the last step is to articulate the findings in a coherent story. Analysts may have worked months on the data, but the final recipient of the analysis will probably not share a similar understanding.  Therefore, in order to convince the decision maker of the legitimacy and validity of the conclusions drawn, it is necessary to create a story where each point flows logically from the premises.

Now that data is increasing at an exponential rate the need to create and explain insights that are easily understood and acted upon is a trend that has never been more relevant. Data journalism not only bridges the gap between analysts and decision makers, but it also helps harness the power of Big Data.

 

More about data journalism? Check the What is Data Journalism presentation.

Or you need help with storytelling? Check the Uplift your Presentations with a Storytelling Infographic.