The What and the Why of Data
Published in Nov-Dec 2022
Some time ago I attended a marketing session on the best practices of data analytics. The conversation veered towards data literacy and data democratisation. A VP stood up to say: “It is great to remove silos, create open data platforms and have access to dashboards, but I need someone to explain to me what the data means – like a journalist would.” That comment really resonated with me, because I feel that a lot of the conversation around data analysis focuses on the techniques rather than on the output.
A recent report by Exasol surveyed 500 leaders and data professionals from enterprises with more than 1,000 employees in the US. Eighty-two percent of them said they use dashboards for their daily communication with colleagues, although at the same time, 53% agreed that dashboards are being disregarded because of the time it takes to interpret them. A dashboard is great at highlighting that sales have dropped by 10%, but fails to sufficiently explain the reasons for this. Combining the ‘what’ with the ‘why’ is key to data storytelling and generating insights.
Here are a few tips on how to create a compelling narrative:
Inculcate A Hypothesis Answering Approach: When asked a business question, begin with a few hypotheses that you can explore in your analysis. This helps to develop a framework to answer the ‘why’ in the analysis. These hypotheses can be formulated by asking stakeholders what they think (by doing high level exploratory analysis or even through anecdotes). This exploration will often force you to consider multiple sources of data, which in turn will make your analysis more comprehensive. However, be prepared to pivot your hypothesis if it is not reflected in the numbers or if you come across a data point that refutes it. The evolved hypothesis will then act as the skeleton of your report. For instance, you may begin by laying the context: “Sales were down this quarter.” The next slide would answer a hypothesis, such as: “Were sales down in a certain region/demographics?”, followed by a drill down into something more specific: “Was it a specific type of product? Was the drop driven by value or frequency?”, and so on. At this point you may want to consider overlaying a third party data point to address potential blind spots in your internal numbers; consumer confidence indices or economic metrics can help explain if something like inflation has played a role as well.
Less Is More When It Comes to Visualisations: A big challenge of data storytelling is to hit the optimum threshold of data absorption. You need to manage conveying bucket loads of information without overwhelming the audience. As a practice, I try to follow a rule in my reports: one slide with a maximum of two charts to answer the hypothesis. One way to adhere to this rule is to use callouts (in the form of comments or colour coded symbols) instead of another chart/table, to highlight significant variances in behaviours within subsets. Similarly, you need to keep the key takeaways specific and concise – ideally a key takeaway should be no longer than a single sentence. It is also important to ensure that your charts are linked to each other (it helps connect the slides to each other). Generally, I try to ensure that the charts on the same slide reinforce each other by making a similar point to validate a hypothesis. Next, I try to ensure that each slide builds on a point made in the previous one. When presenting, I try to end every slide on a data point that will be explored in the next one. This allows for a natural progression of the report’s narrative – which is the ‘story’.
Make Numbers Relatable: “We need to particularise so that we are not anesthetised by the sheer volume of the numbers,” said Ken Burns, a documentary filmmaker, when asked how he made the scale of the Holocaust more relatable to the average person. Leverage qualitative data points to bring proportions and numbers to life. I add snapshots of tweets and social media comments to explain a behaviour, as well as snatches of something I may have heard in focus groups or in interviews. I have even played back actual responses given on telephone surveys to contextualise behaviours. Framing a hypothesis in terms of behaviour enables audiences to draw a line between action and numbers, paving the way for follow-ups and further exploration. For example, in your first party data you may observe that customers in a certain market buy your product only once in their lifetime, while the market research findings may show that you are considered a premium brand. You may ask your audience if it is safe to assume that people think your product is a once in a lifetime purchase and get their opinion on it. Keep in mind, using qualitative data is an attempt to explain the numbers – it is not an all-encompassing answer.
The Number of Pages/Slides Will Never Reflect the True Effort: Data analysis is a never ending process, and one has to look at data from multiple angles to derive insights. The challenge is stopping yourself before entering the void of analysis paralysis. As your hypotheses evolve and you start to pick up trends, you have to decide which data points and visualisations best reflect the point you are trying to make without numbing the audience. I like to split my reports into a main section that contains the most important insights, and then have an appendix that gives more details or other information. I once heard a story of a stakeholder telling an analytics firm something along the lines of: “We paid you $100,000 for just 10 slides? Each slide better be worth $10,000.” I think the blame lies on all sides. For some reason, the number of pages in a report reflect value for a stakeholder. In their defence, a lot of reports tend to be meaningless tables and clumsily put together charts. Having said this, effective data storytelling negates most of these concerns. Your stakeholders already know ‘what’ is happening. They want to know ‘why’.
Ans Khurram is an analytics and insights professional. He was included in the Insights250 List 2022 and the 30 Under 30: Rising Stars of Business Analytics List 2021. firstname.lastname@example.org
Comments (0) Closed