Data Scientists aren’t Domain Experts

A benefits-realisation process for data science projects

Data-based decision-making is becoming a competitive necessity. But expectations surrounding Data Science – the burgeoning field that has grown out of what used to be called ‘business analytics’ – are being blown out of proportion. For data science to really work, you need a multi-skilled team (not just data scientists) and projects (not just data experiments). Furthermore, to generate business value, you need to connect the world of data scientists to that of domain experts.

Data scientists and domain experts are two different animals. Domain experts – buyers, merchandisers, product managers, and so on – operate in the business domain. To uncover hidden business opportunities, data scientists must capture the business domain in the model domain in the form of concepts, models, measures, and hypotheses that are checked for their fit with available data.

Then, any insight uncovered in this model domain must find its way back into the hands of the domain experts to be put to good use. In today’s business environment, however, data scientists are separated from domain experts – and the challenge is to bring the two together.

Benefits-realisation process

Success in realising the benefits of data science will require a process that connects the business and model domains. But this process isn’t just a matter of collecting or sharing data and information – it’s also about social connectivity, which has been under-appreciated.

Prof Stijn Viaene, a full professor of ICT and business process management at Vlerick Business School and KU Leuven, has published an article in which he proposes a benefits-realisation process comprising the following activities:

  • Modelling the business (representing a business idea in the model domain regarding how to use data to improve the business)

  • Discovering data (searching for insights by running data experiments – involves iterative data-based model analysis and synthesis in the model domain)

  • Operationalising insight (transferring the insights gained in the model domain into the business domain and embedding them in the organisation’s work systems)

  • Cultivating knowledge (promoting best practices for using data and a culture of decision-making based on data and analytics to maximise the return on your data science investment)

Constructive conversations

This benefits-realisation process should form the skeleton of all your data science projects. The four activities have one feature in common: they’re guided by constructive conversations between data scientists and domain experts. These conversations are characterised by interpersonal interaction and dialogue, so that the project participants can appreciate ‘the other side’s’ perspectives.

Creating a digital business ecosystem

In the end, your analytics platform should form the basis for a digital business ecosystem. It should stimulate business productivity by opening up the world of data and analytics to the entire business community – and making it easy for internal and external parties to connect and innovate with data.

The platform – whose attractiveness lies not least in the continuous incorporation of new data sources and data science technologies – should also be an open invitation for ecosystem participants to venture into new domains with analytics.

Source: ‘Data scientists aren’t domain experts’ by Stijn Viaene. Published in IEEE IT Professional, 15 (6): 12-17, 2013.

This article was made possible by the support of the “Bringing IT to Board Level” Deloitte Chair at Vlerick Business School. 

Discover our executive programmes for IT professionals and managers.
Big Data & Analytics is part of the Mastering IT Management programme. Or, as C-level executive, consider the Executive Summer School in Digital Transformation and Digital Leadership.


& Rankings

Equis Association of MBAs AACSB Financial Times