At Dativa, we believe that creating a data science culture is the single biggest thing that business owners and CEOs can do to ensure the success of their enterprises in 2018 and beyond. Now, when most businesses believe that they are GDPR compliant, is not the time to sit back and forget about data until the passing of a new data law. To the contrary, there will never been a better moment to drive the data science culture of your business. Data science culture is the essential element of a robust and successful data strategy. In our experience working with data teams of all shapes and sizes, we've seen companies getting data science culture both right and wrong. Here are some of the things the successful examples have in common:
1) Start at the top
The first and foremost component of a data science culture is that it starts at the top, with the CEO and other key decision-makers fully committing themselves to its cultivation. Their aim should be to create a data-driven culture across the business, using the data it has and could have to create leverage wherever possible. Another area where executives can create this culture is through recruitment, not only hiring people with the right experience and enthusiasm for data but also ensuring that the company has a data steward, to oversee data strategy and a data protection officer (DPO) for data security. Facilitating the other features, such as ensuring there is time and money available for training is essential, as is making sure all top-level executives are well-versed in data issues with at least some hands-on experience. Finally, leaders set the example and so the actions of your executives need to demonstrate how much they value a data science culture. If employees see that management can't be bothered to get the data right, it is likely they won't make an effort either.
2) Focus on the staff, not just the policies
Data science culture needs to be staff-focused. After all, culture is all about the human aspect. We find that the leading cause of the failure of a company's data strategy isn't data itself, but staff-related. This failure is why we stress the importance of creating a vibrant data science culture within any organization where there are interactions between human employees and data. We find that the executives in most companies underestimate the importance of fostering a data science culture, even though it is the staff and not themselves who are having to collect, cope with and ensure the smooth working and flowing of the data on a daily basis.
3) Regular training and education
Regular training and educational sessions for mid-level staff are crucial, and not just practical sessions on the tools that they are using. Enthusing the teams about the power and value of the company's data, and the data they work with as individuals, not merely as a one-off session but in an ongoing way, is the first step. This educational process begins with getting the staff interested in the data. Getting them to identify the data they are using in their working lives and how they can use to improve their working lives - from a purely selfish point of view - is a good starting point. It is essential to paint the bigger picture to help motivate staff to feel part of a larger team but also so that they get a strong sense of the value of their work with the data being part of a greater whole. There's a negative side to this positive too, which is how even small data errors can have a more significant impact Concrete examples help reinforce this message. And your teams need to be trained to ask the right questions and to see the grey areas that always exist in the data they are dealing with and in the human-data interactions.
4) Teach and encourage intuition
Staff should be taught to use their intuition when it comes to data, with concrete examples of how this might play out. If an employee convinced that some data passing through her desk seems incorrect should trust her senses and either explore further or tell her manager, and this spirit of questioning should be encouraged. In spite of the swiftness and intelligence of many algorithms and automated processes, humans can sometimes still spot specific errors in data more accurately than the most advanced computer program. Also, they are the ones who have similar datasets on their computer screens every day and will be the first to spot that something is different. When this happens, staff need to trust their intuition and be listened to by their managers.
5) Sit your business and analytics teams in the same place
Your business and analytics teams are both working towards the same goals, so a great deal of productivity can be encouraged by interfacing between analytic and business units. Getting these teams working together in a training environment is beneficial, but we would extend this by having the people physically located in the same place, where possible, to encourage these interactions. This proximity breeds the kind of tacit understanding between these teams in which analytics thrives.
6) Make the data as easy to access as possible
All KPIs need to be data-driven and data-justified. Most companies know this. But getting buy-in for these KPIs across an organization is where organizations tend to go wrong. Your staff should have constant and easy access to the data you have so that they are part of the process of driving and improving KPIs and other important goals. Data should not only be shared at all-hands meetings. Most companies have a hierarchy of data users and set of associated data rights and privileges for each. The level of data that the Board and Management Team needs is usually quite clear; so is the level of access that a BI team needs. We often find that the middle of this spectrum non-senior team members, who do not have specific "data roles" - can be overlooked. But it is precisely these teams that need to be engaged to drive a widespread data science culture throughout the organization. So making sure these teams also have the right access to the data is crucial.