Which Data Skills Does Your Future Workforce Need to Learn?
“Companies must understand that as their use of data continues to transform, so too must the ability of their workforces to deliver on these new opportunities.” – Sanjeev Vohra, Accenture Technology
We’re all about data skills at QuantHub. That’s why recently we’ve been talking a lot about the pressing need for companies to begin assessing and upskilling their workforce for data skills.
Why is upskilling for data skills so important?
It’s now widely recognized by experts at the World Economic Forum, McKinsey, Accenture and elsewhere, that over the next few years, the need for data-related skills in future jobs will explode due to automation, AI, and other rapid technological advancements.
As a result, a large part of the workforce will see their current job skills either made redundant or needing to be enhanced by more technical, data-related skills.
The problem this presents to companies is two-fold:
- There aren’t enough university graduates and job candidates with requisite data skills. (Plus it’s seriously expensive to hire those who have enough data skills)
- Many current, valued middle managers and other valued employees don’t have the basic data skills to do their future jobs well.
It’s critical that companies recognize these trends and begin upskilling and reskilling their valued employees to operate in a data-driven world.
In two previous articles, we looked at:
- WHY organizations should upskill the workforce for data and analytics
- WHO organizations should upskill as a priority
In this article, we look at WHAT data skills employees of the future workforce will need to upskill in to be effective in their roles and to enable the “data-driven organization”.
What Data Skills do Your Employees Already Have?
Many talent managers might assume that most employees have some degree of data skills. This may come as a surprise, but evidence shows that in the United States, relatively few adults have data-related skills, especially when compared to the rest of the world.
For instance, a study conducted a few years ago by The National Center for Education Statistics found that out of 23 countries, adults in the United States ranked 21st for data interpretation and problem-solving skills, well below the international average.
More recently, Coursera’s 2019 Global Skills Index found that the United States ranked #16 in the world for data science skills, just above Portugal, and well behind most other European countries and China. For tech skills, the US ranked #23.
Employees themselves are expressing unease with data concepts in the workplace.
A recent study from Accenture found that US employees feel unprepared to work with data and are stressed about it.
A whopping 1/3 of employees admitted to taking at least one sick day due to stress associated with having to work with data. Three-fourths of employees said that they feel overwhelmed and unhappy when working with data at all.
To be sure, this lack of data skills and unease working with data partly contributes to the fact that most data science initiatives are met with resistance and a significant number of big data initiatives fail due to people issues.
The moral of the story: Assume your workforce knows very little about how to use data.
First, Assess Your Workforce for Data Skills Gaps
Given the apparent lack of data skills in general, it follows that you’ll need to assess your workforce (which does not have to be done all at once) as a first step to deciding what data skills they need to develop.
Ask yourself, do you know how many people in your workforce can interpret common statistical outputs such as correlation or moving averages? How many employees truly understand, appreciate, and use data visualizations and insights in their daily work?
The fact is, many people in your organization aren’t familiar enough with data concepts to be able to understand and engage these kinds of skills.
There’s a basic level of knowledge and competency with data and analytics that you will need to assess before putting in place any data upskilling program.
Ideally, data skills assessments are administered at the start of employment or during the recruitment process for new hires. That way, when a candidate is hired, HR will already know what kind of data literacy learning and development opportunities should be offered to the new hire over time.
Apart from the hiring process, assessing existing employees for baseline data skills allows HR and business leaders to engage in an initial “skill mapping” exercise.
Management can map requisite data skills to particular roles and teams. Through assessment, they can then identify which employees lack the requisite data skills needed in their roles, essentially identifying individual skill gaps.
Assessment results can also identify employees who might benefit from a transfer into a role for which they already have the appropriate level of data skills. We call this situation a “positive skills gap”.
Once you have a picture of data skills requirements according to roles and have identified individual employee skills gaps, you can begin putting employees on individualized learning and development paths to help them upskill in data concepts.
Understand Different Data Skill Levels
It’s difficult to know where to start when trying to build data fluency among employees because data fluency encapsulates a spectrum of skills ranging from a basic understanding of what data is to understanding high-level analytical concepts like predictive analytics.
As an initial target, you want the workforce to be able to read, manipulate, analyze, and argue using data. This is called data literacy.
Data literacy is akin to learning basic language concepts and being able to communicate in that language in a competent way.
Data literacy prevents employees from taking sick days and being unhappy working with data.
As they move on to higher-level thinking, data-based inquiry, and proactive use of data analytics and modeling, employee skills move closer to data fluency.
Data fluency is just what it sounds like. It means being fluent in the language of data and using and interpreting data language in more complex ways.
In a data fluent organization, everyone speaks and understands the language of data. Being fluent in the language of data allows employees in diverse areas of the organization to communicate and connect over data concepts, tools, and processes. It’s what fosters a data-driven culture and data-driven decision making.
10 Essential Data Skills
Because there’s a spectrum of data skills, data upskilling should start with basic data literacy skills for all, and build up to higher-level data fluency skills as identified through skill mapping.
We propose 10 key skill areas for what we call “data essentials” skills below and give a high-level description of each. These skill areas are listed generally in order of complexity and applicability to the entire organization. The degree and number of requisite data skill types will vary at the corporate, business unit, team, and individual levels.
1. Data concepts
At a minimum, all employees need a general understanding of what data exists today, how it can be used, and the value it brings to an organization. They should have a foundational understanding of the uses and applications of data.
Employees should also have an understanding of the high-level issues and challenges associated with data. They should understand the importance of data and be familiar with concepts such as data ethics and data security.
2. Data governance and stewardship
Business leaders, managers, and employees alike will all need to be familiar with data governance issues such as security, privacy, trust, ethics, and bias if they are to fully engage with data systems on the job and minimize associated risks.
Managers will need to set and employees will need to follow guidelines about user interfaces such as the consistency of reports across departments.
Employees will also need to understand their role in collecting and protecting data and in maintaining a single source of truth for the organization.
3. Data-driven decision making
The ultimate goal of data fluency for all is to achieve that buzzword of a term, “data-driven decisions”. Investments in data analytics can be useless unless business people can incorporate data insights into their complex decision making.
Most organizations are not quite there. A recent Forrester Research study, “Data Literacy Matters, The Writing’s On the Wall”, showed that only 21% of workers know when to question the results of an automated tool. In addition, only 48% of employee decisions are based on quantitative analysis. This, Forrester says, is due to a lack of data interpretation and analytics skills at all levels in the organization.
Good data-driven decision making based on insights generated by data is founded on a variety of learned skills. For example, to make good decisions employees need the technical and soft skills that give them the confidence and ability to question the output of an algorithm or model. They also need to recognize that not all numbers are reliable and that some data is better than others.
Data-driven decision making is easier when employees understand the factors and calculations behind the numbers they are using to make decisions and learn to think critically about the accuracy, sample sizes, biases, and quality of their data. In this sense, most employees should learn basic statistics.
Finally, to make data-driven decisions, employees need to be able to construct a business case based on accurate and relevant data outputs.
4. Data discovery
Data discovery is the process of collecting data from various databases and consolidating it into a single source that can be easily and readily analyzed. Once the raw data is consolidated and converted for use, an employee can explore an idea by drilling down into the data.
Being able to engage in data discovery is a key building block on the road to data fluency. It gives the individual employee the ability to organize, explore, and make sense of multiple types of data given the context of the employee’s role. Employees should thus become familiar with concepts and techniques for pooling different data and results.
As part of data discovery, employees also need to be trained on how to look for trends and patterns in datasets they create, and how to go about analyzing contributing factors. To do this, they need to be able to explore data by different characteristics such as region, different employees, product type, and more.
Part of data discovery also involves knowing how best to visualize data trends and patterns. More on that later.
5. Data applications
Data applications enable the use and operationalization of data in an organization. They are the tools that allow any end-user to interact with data. Examples of data applications include Excel spreadsheets, business intelligence applications, CRMs, and self-service data analytics platforms.
Employees should have knowledge of common data analysis tools and techniques and should be able to select and apply the appropriate ones for their job or the problem they are trying to solve.
That said, as a first principle, employees should only be assessed and educated on tools that are actually available to them for use in their individual roles.
These days, more sophisticated and user-friendly data analytics and data discovery platforms are allowing users to work with data without supervision.
What this means is that not only does a Marketing Manager need to know how to pull numbers from the data analytics platform, but he/she also needs to understand what questions they should be asking of the data and understand how those questions are being answered in the context of the data that they are pulling and manipulating within an application.
6. Data visualization and communication
Employees need to be able to create easy-to-understand representations of data analytics results so that others in the organization can understand the results and recommendations. They should feel comfortable creating communications about data insights for a variety of stakeholders.
Following on the previous point about unsupervised self-service analytics, employees need to be able to choose visualizations that best convey the characteristics of their data, rather than the chart that looks the most visually appealing to them. They need to know when to use a scatter chart instead of a bar chart and so forth. This takes practice.
7. Data interpretation
Likewise, those receiving the presented data analytics results, especially key decision-makers, need to be able to interpret data outputs and visualizations well, so that the results and decisions made based on the data have the expected impact.
Data interpretation involves the processes through which data is reviewed for the purpose of arriving at an informed conclusion. The interpretation of data assigns a meaning to the information analyzed and determines its signification and implications.
8. Data project planning
As we move further up the data literacy curve, project management skills and the ability to envision and collaborate on data initiatives to solve business problems come into play.
A key goal of moving employees along the spectrum of data literacy closer to data fluency is the ability to engage in data project planning and execution alongside the technical teams, thereby reducing resistance and enhancing chances of success for data initiatives.
Skills to master include planning the project, identifying data requirements, data security and privacy requirements, model requirements, and more.
9. Exploratory data analysis
Higher-level data skills include knowing how to conduct exploratory data analysis, or “EDA”. EDA is an approach to data analysis that employs mostly visual/graphical techniques – hence the need for employees to develop data visualization skills before engaging in EDA.
An employee conducting exploratory data analysis will be able to suspend their assumptions about what they think their data will reveal and instead allow the data to reveal its underlying structure and model.
The EDA process might reveal missing data and data collection mistakes. It may also include identifying data anomalies, outliers, and identifying influential variables.
EDD is the first step to more complicated data analysis. Therefore, EDA is something that any employee seeking new insights from their departmental data can learn to do as a first step to developing more complicated modeling skills.
Skills involved in EDA include being able to plot data, for example by creating a histogram or block plots, as well as plotting simple statistics on say, a box plot.
The employee then should be able to position and examine these graphics and use their intuition and underlying knowledge to recognize patterns in the data.
10. Data-driven inquiry
With many of the aforementioned skills in hand, employees should eventually be able to engage in an ongoing process of identifying business problems and forming questions or hypotheses about these. This is essentially basic data analysis.
They should have a sense of how to collect and analyze information and data or be able to seek out someone who can help them do this.
This is the point where the bridge between the data-fluent workforce and the data science teams can be formed internally. Data Scientists can help a data-savvy employee translate their information inquiry into a valuable action plan.
Just like any type of “literacy”, the degree to which people are “literate” or “fluent” in these data skills can and should vary according to their roles and responsibilities and the degree to which they will need to use the language of data to communicate across the organization.
The bottom line, however, is that upskilling for data literacy means leaving no employee behind. Pretty much everyone will have to work with data and analytics information at some point in the future. Working towards these 10 skillsets will enable your workforce to do so.