data, data science, data storytelling

The Skills Needed for Data Storytelling

By: Marc Fanelli

by Marc Fanelli, SVP Strategic Partnerships and Global Data Supply, Eyeota


It is important to recognise that the term “Data Storytelling” is a double-edged sword. Businesses increasingly require complex data analysis that drives business decision-making into narratives that are consumable, understandable and able to be democratised across organisations – to both technical and non-technical people alike.

However, one of the fundamental challenges facing organisations is that many still hold on to an assembly-line mindset, where the same technical personnel who perform data ingestion and cleansing functions (preparing data for analysis) are also expected to perform analyses and subsequently summarise them in a way that can be understood by a non-technical audience. This can go all the way to the C-suite, where decisions about where to invest, divest, add or cut resources get made. As such, would-be data storytellers may not have a seat at the top table, where their work is often “dumbed down”, misinterpreted to fit a narrative (intentionally or unintentionally) that can lead to erroneous conclusions and decisions. Perhaps part of this legacy mentality about the “schism” between left and right brain classification is partly to blame for this dichotomy. It has led too many organisations to adopt an unconscious bias in their thinking: data scientists crunch numbers, others do the storytelling. The results can often be misleading.


A case in point would be the COVID-19 pandemic, and how information is presented and subsequently interpreted based upon flawed data storytelling in the majority of the reports released to the public. Examples of incorrect conclusions arrived at by millions of people – based upon the storytelling narrative – include the rate of infection spreading, countries or population centers per capita infection rates, comparisons of country data with others, and so on. Those crunching the data are in the best position to explain the very important nuances, such as the reason infections in an area may have spiked dramatically is the due to the simple fact that more tests were administered over the same time period, or that certain areas that were seemingly unaffected are simply the result of having little to no testing. The amount of data storytelling that is flawed in relation to the COVID-19 pandemic only underscores the point: for data scientists to be part of the storytelling process – be they private sector or government – a new paradigm needs to be adopted under which those that analyze data stay with their findings and work as part of the storytelling process.

Perhaps one of the main reasons why the assembly line, linear approach from analysis to the presentation of results is more susceptible to flaws is that very few individuals possess the skill set required to do all the tasks needed to create credible, defensible data stories which can be acted upon with confidence4. As specialisations in expertise continue – bringing us deeper, more innovation in very specific areas – it creates an opportunity to alter the paradigm out of necessity. This may well lead to experts in different disciplines being harnessed in such a way that we see the diverse skill sets come together to provide the proper balance of business context, data preparation, data analysis and finally, the ability to “speak for the results”. As Nate Silver observed in his book The Signal & the Noise: “The numbers have no way of speaking for themselves. We speak for them. We imbue them with meaning”. Charts, graphs, and data points left to their own devices can be easily misunderstood. It is all too easy to mislead – be it intentionally or unintentionally, depending upon the way information is presented. This classic example from Darrell Huff’s famous book How to Lie With Statistics – published in 1954, but it still holds good today – exemplifies the point:

Screenshot 2020-07-10 at 15.50.51

Figure 2. From Darrell Huff’s How To Lie With Statistics (1954)

The same data presented on two different scales will elicit a response from most non-technical people (and C-suite executives) that creates an incorrect narrative. And once digested and acted upon, these misinterpretations can have detrimental consequences to an organisation that acts upon the incorrect telling of a data-driven story.


So, how does one build an organisation that possesses the right skills needed for data storytelling – the right philosophy – the right bench to foster an enduring culture of impactful data storytelling? It begins with, if necessary, a pivot in the way that data science (and data scientists) are viewed and structured within an organisation. As noted previously, there are still too many organisations keeping their data scientists buried within the bowels of their technology functions. The problem with this model is the data science function is cut off from any meaningful understanding of the business’ strategy and objectives. What’s more, they operate within a technological constraint- based framework. Instead, they should be putting pressure on the technology function to evolve to the market and to the business, instead of being told that they must accept and work within their current environments. Here’s a five-point action plan to help change the paradigm.

1. Data scientists should have an equal seat at the table as the CMO, CTO, and others. Consider employing a Chief Analytics Officer (CAO) and building a center of excellence under them. Many organisations have done this with great success, as the business-impacting capabilities that data scientists possess will be elevated, as they move from “order takers” to “collaborators”.

2. Determine the talent you need in your organisation to achieve your business objectives. Remember, it is unreasonable to think the same person can perform all of the necessary functions that lead up to and make for great data storytelling. To build a high-performance team, you’ll likely need people adept at data manipulation and data structures for accessing, cleansing, standardising data for analysis. You’ll also need data analysts, well-trained in designing proper experiments, such that hypothesis testing is structured according to fundamentally sound mathematical and statistical principles. Once you have identified the skills you need, perform an inventory of your current staff’s skill sets to identify who has experience of what. Often, you will find that there are personnel with existing experience and skills that are required but simply not being leveraged.

3. Expose your data science team to your business strategy and objectives. They are one of your secret weapons in developing creative solutions to help you realise these objectives in our increasingly data-driven world. Again, an equal seat at the table as the other core traditional functions will accelerate their contribution. Advocacy is key here and a huge part of the CAO’s responsibilities. They should always be looking for opportunities to engage their team in high- profile, high impact initiatives and then providing a stage for the team to present its findings – yes, to tell its story – instead of having someone else do it for them. Exposure builds trust and excitement within the data science function, one that typically lives in this paradoxical world of hearing how their skills are in huge demand, but little sunlight is cast upon them in terms of exposure within their own enterprises. This is changing but we still have a long way to go.

4. The team will likely need a Project or Program Manager who can straddle the line between business needs and the work product being produced. They should be excellent communicators, navigators that can be counted upon to keep balance and order in the universe. If there is a disturbance in the force, they feel it first and work with constituents to rectify it. This could be projects running late or how to handle the impact of a sudden left turn. These team members need to understand the big picture and never lose sight of it, while at the same time being deep in the day-to-day realities associated with the work being produced. They help to foster the collaborative, agile environment within the center of excellence, connecting many dots and managing expectations.

5. Engage your marketing organisation – or augment it with external resources – in the data storytelling process. Some disciplines are simply more gifted than others at communicating. I have often observed that, if you engage talented writers and storytellers early in your process so that they have a good idea of what you are trying to do, this simple act of inclusion pays dividends later, when it comes to striking the right balance between too much / too little information, making concise statements / assertions, and overall helping frame the storyboard in a way that reads naturally.


The guidelines provided here are just that – guidelines. Every organisation needs to find the model that works best for them. That said, the world of business – and particularly the world of marketing – continues to spew out more data exhaust every day. This means that the ability to access, manipulate, analyze, and articulate clear and concise data-driven stories is only increasing.

To affect meaningful culture change and create more effective data storytelling, organisations should seek to build teams that comprise all the complimentary skills required – just flick back to Figure 1., above. It has become increasingly clear that world-class data storytelling is a community activity and isn’t just a singular function. Identify what you need, assess your gaps, and recruit to fill those gaps. Organisations that choose to embrace this challenge – and move to establish a more impactful data storytelling culture – will reap the benefits and obtain competitive advantages over their more “traditional” peers who see data science as a purely technical, assembly line function.

I-COM+Data+Storytelling+Council+WhitepaperExtracted from the I-COM Data Storytelling Council whitepaper: Five Areas Marketing Needs to Address for Better Data Storytelling.

Get your copy now to read insights from:

David Lloyd - Head of Data & Insight, Wunderman Thompson Data, UK

Emma Whitehead - Creative Director, Kantar, UK

Jared Rodecker - VP Advanced Analytic Solutions, RAPP, USA

Stefano Vegnaduzzo - SVP Data Science, Integral Ad Science, USA

Dr Sam Knowles - Founder & Managing Director, Insight Agents, UK


Download Now