How do you build a data-driven organisation?

It is not technology but soft skills that determine success…

Many organisations want to do more with data. To deploy data effectively, important steps must be taken. We walk through the transition to a data-driven organisation at pace. Harmen, senior data scientist at Datalab, puts Jeroen, project manager at Datalab, through his paces.

Read and watch what this delivers…

Harmen: 'First of all: what is data-driven work?'

Jeroen: 'It is widely used as a concept, but many do not know precisely what it means. The moment you start incorporating data from your business - or sometimes from outside your business - into decision-making, and you do it across the entire organisation, then in my view you are a data-driven enterprise.'

And when can you say you are truly working data-driven?

'In the classic model, you make decisions based on knowledge and/or experience. When you start working data-driven, you incorporate data into decision-making. Naturally, that data must be cleansed and organised, and sometimes already analysed.'

And what does that look like in practice for a business?

'You have a number of steps to take. First, you need to know what data you have, what you need, and what is still missing. Once that is clear, those datasets can be unlocked via a central point - a data warehouse, for example. From there, data sources can be matched, enriched and analysed with data products such as a dashboard. The final step is using the outcomes in decision-making.'

'That sounds like there is a lot involved in data-driven work. That also seems like one of the disadvantages - there is simply a lot of work to do before you can work data-driven…'

'You certainly do not do this in an afternoon, but it is always worthwhile. The advantages of incorporating data into business operations are abundant. Precisely by combining data sources, you arrive at insights you could never have obtained otherwise. All sorts of companies have even fully aligned their business model with data - think of Google, Airbnb and Booking.com. That is their business model. It works!'

'You give examples like Google and Airbnb - those are obviously the really big names. These are organisations that are already very advanced in data-driven work. Suppose you are an organisation that is less far along. What are the concrete advantages of data-driven work?'

'Very diverse: it could be efficiency improvements for internal processing, cost reduction or revenue increase, applying better customer segmentation so you can better align your products with customer groups. But also better identifying doubtful debtors.'

'How do you tackle this practically as an organisation?'

'Everything starts with collecting and properly organising data. Only then do data products come into view, such as a dashboard. This could just as easily be a one-off or recurring analysis, of course. More importantly: look carefully at the end user. Who will actually use the data product in practice? If it is a sales employee, they probably prefer a dashboard. A group of scientists, on the other hand, prefers working with a well-defined cleansed dataset for further analyses.'

'What do you need as a minimum as an entrepreneur for data-driven work?'

'First of all, the ambition to want it: you have to believe in data-driven work. Data-driven work is not a "holy grail", because not every data product delivers immediate profit or adds value right away. In short, no guarantee of success. What we have noticed so far is that everyone gets more out of it than they thought beforehand. That said, the first step is always: map your data sources. Often we then discover that organisations actually have much more than they had initially thought. Once you have done that, make a plan for how to unlock these data sources, in what order, and what data you need for that. And importantly, ask yourself how fast you want to go. In stages, or do you want to do everything at once? We often help our clients think this through carefully. Because one source is more complex than another. Especially when certain data elements are stored in different sources that do not communicate with each other - they can contradict each other, and which is the truth? You have to think about that.'

'It sounds quite abstract. Can you give practical examples?'

'Plenty. A probably very recognisable one is - without wanting to name names - a webshop. This retailer has an online shop, a stock system and a financial system. These three operate independently. The moment you link them together, you will encounter differences. Think of product codes that are not the same everywhere. You need to address that. Once brought together, you suddenly see that stock levels can be reduced, saving costs. But you also discover better customer segmentation, increasing revenue. More insight leads to higher margins.'

'This seems like a very nice and practical example. If I want to get started, what should I think about? And is it complex?'

'It is not difficult, but it does require a solid plan of approach. Think of a certain structure, a certain order of things. It is difficult if data is not directly "your world". You make something, sell something, or provide a certain service. You do not build data labs. That is precisely what we do. So we have done this before - for us, it is not difficult. You also have to consider that one organisation is not the other. You have larger and smaller organisations, organisations with a lot of data and organisations with complex data. In other words: it is bespoke.'

'Are there organisations that cannot benefit from data-driven work?'

'I have not been able to think of any, except perhaps organisations that are a bit too small in size. But in terms of industry - no, not really. Because even if you think you have no data, often you have more than you would think beforehand. You do not need to have all data yourself. You can obtain data "from outside". There is always data available to do something with.'

What is all involved when I think about technology, personnel and the organisation?

'Good that you mention all of them, because it is not just technology when you work data-driven. The way you make decisions changes fundamentally. You incorporate data, which also means people need to be taken on the journey. Training to be able to read data products such as a dashboard. And data products need to be created! The change can be profound - one organisation handles it differently from another. There are also organisations that see change as a challenge: it gives energy. There are organisations accustomed to working in a certain way - they find such a change more difficult.'

And regarding costs…?

'In principle, we virtually always see that the benefits outweigh the costs. Moreover, you cannot avoid it - every organisation must sooner or later get to work with data. Customers expect it too.'

'What does the customer notice when I start working data-driven?'

'That could be anything: a lower price, faster service delivery. But it could also be that they can log into their own portal and see a dashboard fully tailored to them showing expenditures.'

'Data-driven work is really nothing more than hiring a data analyst and ensuring they can get to work with the right technology - or am I wrong?'

'You are indeed wrong - it really is more than that. A data analyst actually already works with prepared data, the final step, approximately 25% of the whole of data-driven work. Three-quarters before that consists of collecting and cleansing data. But the most important step - which we still forget - is that the people who ultimately have to work with those data products must also be able to and want to.'

'Can you explain that?'

'Data-driven work means many changes. It is not just ensuring people know how to use a dashboard - you also have to understand what you can do with it. That takes time. Ultimately, you want people themselves to come up with what data products should show - so that they start asking questions of that data analyst themselves.'

'In essence you are saying that data-driven work means getting everyone enthusiastic about working with data products. Clear - but what does this cost and how long before I have this sorted?'

'I have said it often: one organisation is not the other. With a small organisation with a limited number of data sources, you can have your first data product within a few months. If you want your entire organisation to truly work data-driven - where you also address the human side - then you are looking at somewhat longer, up to a maximum of one to two years.'

Harmen & Jeroen: thank you for this clarifying conversation. Naturally this is generic - in practice things often go slightly differently. Perhaps you have specific questions, or you are enthusiastic about working data-driven yourself. Do not hesitate to contact us - it is always free and without obligation.