Translating Data into Better Decision-Making: People & Culture

For an increasing number of organizations, the demand for data skills is growing – not just for the data science department, but also for employees involved in day-to-day operations. It requires a shift. Employees will have to learn new skills, for example by means of an educational training program. To provide a large number of employees with a relevant and useful experience, such a program will have to be tailored to the organization and audience in question. Robert Monné, manager of The Analytics Academy, recommends a mix of online and offline learning interventions. “The training content has to be closely adapted to your employees and their daily jobs.”

The employees of an organization can often be categorized into different groups. After all, IT experts need a different set of skills than customer service employees. “Offering a tailored education program to each individual employee group maximizes adoption”, says Monné. “You can offer theory and exercises in a learning environment, before encouraging employees to apply their newfound knowledge in their daily jobs. The results can then be discussed with peers or internal experts. In addition, employees will have to be coached and the organization will have to communicate positively about new technologies and applications within the organization. That is how you offer your employees the right toolset, skillset ánd mindset.”

_________________________________________________________________________________________________

“Offering a tailored education program to each individual employee group maximizes adoption”

_________________________________________________________________________________________________

As an example, Monné brings up training sessions developed especially for data scientists. “These sessions revolve around a certain theory, which is then immediately applied to data. It is best to apply the theory to real-life situations right away, combined with on-the-job training in which participants work on their own data projects with or without supervision. Experienced specialists can teach them the ins and outs during this process, as building a model or analyzing a sample dataset during a training session is considerably different from a real-life application. After all, in a training session it’s difficult to simulate conflicting stakeholder interests, unclear business questions, poor data quality or unavailable data, programming challenges and poorly performing models, to name a few factors.

Organizations can also set up community sessions. Those sessions help to even involve employees in the process who do not participate in the training sessions. It always leads to great impact.”

_________________________________________________________________________________________________

“Community sessions help to even involve employees in the process who do not participate in the training sessions to increase adoption”

_________________________________________________________________________________________________

Monné is a firm supporter of this approach. “If you really want to work with data, you will have to take a range of measures to ensure that people apply it in their day-to-day jobs. The first step is creating the right mindset, followed by learning how it works – how to apply the skills and use the tools. Finally, you’ll have to apply the theory in practice and in use cases. This step is essential for organizations aiming to take a more data-driven approach.”

Not just for specialists

In practice, Monné has noticed there is a great appetite for knowledge about data-driven working, and not just among specialists. “Since a few years, The Analytics Academy delivers trainings sessions designed for data scientists employed by a major Dutch service provider. The target group of the training seemed however to be bigger than was originally expected. The training appealed to a much wider target group, despite being too advanced. As it turns out, even people without the necessary background are interested in applying new techniques. Although they might not possess advanced technical skills, they do have an interest in how to apply data science to enhance their financial reports, for example. To support this wider audience, we have developed a foundational course, which teaches the basic data skills. I firmly believe that this is a sound approach to developing skills: changes are more sustainable if they are adopted organically and people aren’t forced to take a mandatory course. Having more employees with basic knowledge and skills has clear advantages, says Monné. “Data scientists, for instance, do no longer have to be burdened with relatively ‘simple’ tasks, such as integrating two datasets or to gain a relatively ‘simple’ insight. Creating basic insight and reporting issues, which used to be the purview of specialists, can then as well be solved by the business itself. This ‘self-sufficiency’ leads consequently to an increased velocity of decision-making. In addition to the basic and advanced courses, we have developed a third course aimed at creating awareness, Monné explains. “Its purpose is to inspire and to show people the options available. After all, boards and management tasked with using the models developed by data teams to support their decision-making, must have a general understanding of these models, and support their use. Multiple major organizations are currently going through this development.”

_________________________________________________________________________________________________

“Changes are more sustainable if they are adopted organically and people aren’t forced to take a mandatory course.”

_________________________________________________________________________________________________

A different mindset

As soon as organizations have trained ‘enough’ data scientists, they often signal a subsequent need, mentions Monné. “Organizations become aware of the fact that innovative projects and experiments in a center of excellence or data science department are not enough to turn the wheels. They get to experience a growing need for new skills in day-to-day operations. As the journey to a data-driven organization continues, non-specialists that are able to ask the right questions, that can act as sound stakeholders in data projects, understand how models work at a conceptual level and know at a strategic level what investments are necessary, become crucial for ‘making it work’. Developing these skills throughout the entire organization, rather than focusing solely on a centralized data team or center of excellence, is therefore a valuable next step to build solid ground for a data-driven business. Many companies start with experimenting, often with standalone teams. Change is initiated, experiments have proven their validity on a small scale, and the models that have been developed produce usable forecasts and insights. The next step is to upscale these experiments and projects, both with a higher frequency and in a broader range of application areas. Unfortunately, there will, however, be situations in which the company’s mindset has not yet gone through the required change, blocking innovation and growth. It is therefore sometimes first needed to take away people’s concerns with regards to losing their jobs to AI. On the contrary, we strongly believe that AI creates more enjoyable and fulfilling jobs and having this attitude in an organization helps to realize change. In addition to dispelling specific fears, you will have to grow a different mindset across the organization. This is where communication and basic training programs are vital. People will have to learn to recognize, understand and apply data to their daily work. The most important thing is to get everyone on board. The best way of firing up a broad group of people is to showcase inspirational, company-specific use cases. This requires more than “just opening up LinkedIn Learning” to a broad audience: it demands a tailor-made story based on input from the company itself. Examples include videos from people at key leadership positions explaining why the transition is so important, as well as inspiring stories about successful, and a clear explanation of the roles everyone can play. Ultimately, the goal is for the entire company to recognize the value and their role in increasing data quality and data driven decision-making.”

_________________________________________________________________________________________________

“Developing skills throughout the entire organization, rather than focusing on a centralized data team or center of excellence, is a valuable next step to build solid ground for a data-driven business”

_________________________________________________________________________________________________

Translation to business through broad interventions

The translation of new technologies to daily operations often lags. Even if an organization has the tools at its disposal, it is not a given they are used. “People need to be encouraged”, says Monné. “Suppose that you have made PowerBI available to everyone in the company and have given all employees the necessary training for using the software. If people still go back to their old Excel-based ways of, you will only have performed a rather useless intervention on the toolset. The organization’s mindset and skillset remain unchanged. In other cases, only a small group of people such as the data team uses the tool. Others are completely unaware they can create a dashboard to make data-driven decisions, or they lack the necessary mindset, skill set and/or tool set to do so. Ideally, you want to teach the entire organization the same frameworks and techniques, though at different levels. Data scientists should learn expert-level machine learning skills, whilst also growing an understanding of business processes. Correspondingly, management should have an in-depth understanding of how AI can impact strategy, whilst also having a basic idea of how machine learning models work under the hood. The best would be for both groups to have some knowledge of all the elements involved, each on the group’s own level. Using the same concepts and terminology across all groups will lead to cross-pollination, because people speak each other’s language.”

Intuition versus data

At top level, people need to understand that data-driven decisions will sometimes be better than decisions based on intuition, Monné explains. “Understanding when to follow your intuition and when to follow by what data is telling you is a valuable skill for top management. They must understand the application areas, and what changes are required to turn the wheel. For example, having data lead the way makes more sense when making recurring decisions, and the same applies to launching updated products. On the other hand, you could have never predicted the success of the very first iPhone using data and machine learning. Steve Jobs simply understood the market before the market knew what it wanted. Of course, not all product launches are as innovative as the iPhone. Data can provide excellent insights into what customers like, especially when introducing more incremental innovations. Radical innovations, however, require more intuition, so in the end, I think we need a bit of both. Data may tell you that customers are ready for a new product, but you need human intuition to start the conversation. Basing your most important decisions on a mix of intuition and data will most likely lead to much better results.”

“Basing your most important decisions on a mix of intuition and data will most likely lead to much better results.”