How to Close the Data Literacy Gap

The data-literacy gap is widening, and value is draining down it.

The skills gap between those who can and can’t work with data is not only a major barrier to enterprises becoming data-driven; it’s been called an “inhibitor to growth”. It’s clear why. Despite the exponential growth in data volumes, when useful data goes untapped, missed opportunities proliferate.

In contrast, company-wide data literacy has a positive snowball effect. When individual business experts are empowered to make data-driven decisions, it improves the gross margin, return on assets, return on equity, and return on sales, and will be a “crucial driver of business value by 2023.”

Closing the gap and creating a data-driven enterprise means upskilling workers. One way enterprises can efficiently do this is using low-code, no-code data science platforms. Through intuitive visual interfaces, these platforms enable non-data-literate users to build data workflows and collaborate with data-literate users. What’s more, these platforms keep the business running, and don’t delay work while employees take courses.

This article considers how enterprises can roll out data upskilling through low-code, no-code platforms, and become data-driven quickly.

Gartner defines data literacy as “the ability to read, write and communicate data in context, including an understanding of data sources and constructs, analytical methods and techniques applied — and the ability to describe the use case, application, and resulting value.”

In more pragmatic terms, a data-literate employee can:

• Interpret and understand data, deriving useful and meaningful insights.

• Ask relevant questions of the data, limited by the scope of data type and source.

• Think critically about and analyze data, independently of data experts in the enterprise.

• Use data to make evidence-based decisions, and even combine this evidence with gut feelings and experience when necessary. They understand how to test or double-check its validity and have a good sense of when to doubt the data.

• Use data to communicate job-relevant ideas and opportunities, such as helping to create new products and strategies. In other words, they can use data to tell persuasive stories.

So a data-literate employee can understand where their data comes from, what kind of data it is, how reliable it is, and to some extent know what can be done with it. By contrast, although a non-data-literate employee may be a fundamentally competent Excel user, they will lack enough understanding of the data to make robust job-relevant decisions.

Many enterprises roll out courses to build employees’ data knowledge from the ground up. Sometimes these programs contain several layers of complexity. First there’s learning about data and statistics, then learning about job-specific data challenges, and finally learning how to apply that knowledge using new technologies and platforms.

However, it can be more efficient to invert the formula: using low-code, no-code data science platforms in real-life work contexts for data-literacy upskilling. Implementing these platforms is a far more pragmatic way of developing enterprise data literacy than interrupting employees’ normal schedules for training. Instead, low-code, no-code platforms create everyday learning opportunities while also enhancing operational effectiveness. They enable experts and non-experts to work with data quickly, and facilitate collaboration.

Through visual programming interfaces, these platforms lower barriers to entry, enabling users to carry out simple data operations or build complex analytic models in the same environment. They let non-technical users do data processing and analytics themselves without any coding. Not only does this make them more comfortable with data, but it also allows them to collaborate with experts on more complex problems.

Users of low-code, no-code platforms are thus freed of IT delays or resource constraints. In particular, beginners aren’t hindered by the additional need to learn complex programming languages or advanced data science techniques. As Microsoft comments: “82% of low- or no-code users agree that the technology helps provide an opportunity for software users to improve their development knowledge and technical skills." Even further upskilling is facilitated by platforms and technologies with vibrant user communities, where information and skills can be shared in open, free exchanges.

Guided analytics in low-code, no-code platforms offer another pragmatic way of developing enterprise data literacy. By developing job-relevant apps for workers, enterprises can provide them with access to data resources and form the basis for further learning. This method is called Guided Analytics. The process could look like this:

1. The data team accesses and blends data in a low-code, no-code environment.

2. Using visual programming (as well as Python scripting, if they choose), they build a workflow.

3. The data team deploys this workflow for the use of the non-data team via an app.

4. The non-data team uses it and gives feedback to the data team to improve it.

In a real-life example, Continental showed that upskilling their workers for data literacy this way can result in tangible changes. They improved their month-end tasks by cutting down lead time from two days to 30 minutes.

Another example is how Exsyn, an Aviation Services Provider that performs analysis of MRO (Maintenance, Repairs, Operations) data to analyze aircraft airworthiness, created an app enabling customers to perform analysis themselves without any coding. Exsyn reduced aircraft phase-in time from 2-3 weeks to 1-2 days — a win for both customers and the company. Customers were in the air quicker. as time on the ground was reduced, and Exsyn’s engineers could spend time on more important tasks, and not repetitive ones, while customer service workers could learn more about what customers needed to make phase-in time smoother.

A final theoretical example could help shine a light on how exactly this works. A large retailer’s marketing team would like to understand their customer base in more detail. The data team develops an app which uses ML-based customer segmentation to provide highly detailed information about customer behavior.

The marketing team uses the app to derive customer insights, such as time spent browsing in-store, which would have been difficult to identify without machine learning for their marketing activities.

For this scenario, the marketing team does not need to know how the app was built inside the low-code, no-code platform. They don’t need to know anything about the k-means clustering method which the app uses. Nor do they need to understand anything about the principles of machine learning.

Instead, with the app they cut customer data in ways which are useful to them through an intuitive interface. They can see hidden customer behavior insights regarding in-store and online purchasing behavior, or the relationship between age and the types of goods purchased.

The marketing team then works with the data team to improve the app. They learn which customer data is available for model retraining, and discover more about the possibilities in cutting the resulting data.

This method may seem gradual and piecemeal, but it is a non-trivial way of getting users to begin to understand enterprise data. Since one challenge data experts face is explaining data science to their technical and business counterparts, simply creating and deploying the tool for job-relevant work creates a basis for understanding.

A low-code, no-code approach to data upskilling works if data teams can build transparent, flexible apps without any HTML, CSS, Javascript, or staging environments. They can then guide the non-data experts through what happens to the data step by step, without distracting them with opaque scripting language and symbols.

It can be a mutually beneficial process. The marketing team can provide feedback to help the data team understand what the model can do better, and thereby improve their line-of-business knowledge, and the data team can help stimulate learning by helping the marketing team understand how to make decisions with the data.

As data becomes ever more influential in daily life, it’s imperative that enterprises are able to work with it. By localizing data literacy in one specialist department, enterprises risk shooting themselves in the foot by creating bottlenecks and missing opportunities.

Furthermore, upskilling workers for data literacy and empowering them to make data-driven, job-relevant decisions has a positive impact on employee retention. More than 80% of workers are likely to remain in jobs which enable upskilling.

While data literacy can be its own reward, acknowledging workers’ success and encouraging them to make decisions can have further benefits. Some workers may initially find terms like “statistics” and “lookups” forbidding, but by encouraging them to see the results of those techniques, including better decision-making, enterprises can increase their confidence.

Since a data-literate workforce is more productive and happier with their employer, the ability to understand and work with data shows is a key value driver for enterprises. Whether they opt for a pragmatic, thoroughgoing, or mixed approach is down to their goals and resources. What’s important is getting there.