PCR Tech and IT retail, distribution and vendor news
As the old adage goes, ‘knowledge is power’, however, in relation to data science, the real power is in sharing knowledge. Too often, whether it’s an internal data science team or an external provider, the wider business only gets to see the results of data analysis. Information on how conclusions were reached and the algorithms or techniques used is often draped in secrecy. The result can be a lack of institutional knowledge within a business and a limitation in the understanding of the wider application of data science.
It’s easy to understand how data science becomes a ‘black box’ solution. As with many other business processes, there’s a fear that if there is a broader knowledge of how the system works and conclusions are reached, the value of the expertise of the data science team is diminished. If you teach a man to fish, he doesn’t need you to provide him with fish anymore. Of course, this is nonsense, there’s a vast difference between empowering people and training your own replacement.
If you’re a third party provider, such as a data science consultancy, sharing knowledge with your client is one of the best ways to strengthen a relationship. Not only does it build trust that your findings are based on a logical and understandable approach, but it also opens the door to widening your remit. It can jog the client into realising that your expertise could be applied to other areas of the business. It also helps increase the longevity of a relationship by ensuring that there are more stakeholders within the client that understand how, why and what you do.
For an internal data science team, sharing knowledge and providing training helps to increase buy-in throughout the business. It creates a virtuous circle where key players throughout an organisation understand what the data science team does and ensures that the capability can be more effectively used in the future.
Sharing knowledge is about more than just creating a nice PowerPoint presentation that explains how conclusions were reached and then presenting it to the business.
Companies should strive to be closely involved in every step of the process as a solution is developed. Training should be provided to technical staff to help them understand how various aspects of data science models were created allowing them to pick up and customise elements of the solution. To ensure that the knowledge share has more longevity and survives personnel change in organisations, clear documentation should be created that charts the entire process and results. This should allow future users to easily understand how various solutions and models work.
Knowledge sharing doesn’t need to be a tedious process, in fact, it should be seen as an important part of relationship building and management with a company. Thankfully, many data scientists do have teaching experience as a by-product of their academic study.
If you work with an agency who are unwilling to share how they reached their conclusions, you should naturally question their approach and how they perceive their relationship with you. Similarly, if you work with a business as an agency and they express no interest in you sharing knowledge with them, it may be that they don’t realise the full value of what you could do. Approaching the idea of knowledge sharing from the outset of a collaboration is the best scenario, however, it can also start to happen with longstanding relationships.
Many businesses aspire to be open and transparent, however, true transparency requires ceding some of the perceived power a monopoly of knowledge confers. It should be an intrinsic part of the way businesses work together because it has so many benefits for the parties involved. We judge the value of our partners, at least in part, by how much knowledge they leave behind when they leave the room.
