June 5, 2018
Nearly all firms want to do more with data science. However, executives and data scientists alike are frustrated by the difficulty of turning new initiatives into business impact at scale. Outside of well-established use cases — such as risk scoring or optimizing search results — data scientists regularly complain that their solutions can take months or even years to deploy or, worse, are never adopted.
Move Data Science From Artisanal Craft To Industrial Production
A lack of data and technical capabilities is not the key reason new data science initiatives underperform; it’s because most organizations approach data science as an artisanal craft. Here, cross-disciplinary, Leonardo da Vinci-style data scientists (chimeras) construct the full solution to each business problem from scratch — each using his or her set of bespoke tools. Organizations must instead approach data science as an industrial process: one that requires planning, division of labor, and cooperation. Not only does it require multidisciplinary data science teams, it requires that data science, IT, and business teams each execute their respective roles effectively. And it requires the right tools that enable both rapid data science development and deployment, as well as collaboration and management.
Five Best Practices To Scale Data Science Across The Enterprise
Successful data science projects are effective collaborations. The business is actively engaged and committed to implementing the results. Data science designs a business-first solution. IT swiftly enables data access, provides infrastructure and tools, deploys models into production, and builds applications so that end users can access the insights. To achieve this with new data science use cases (e.g., image or text analytics) and with parts of the business that are new to data science (e.g., sales or HR), successful enterprises leverage five best practices:
I’ll be delving into each of these best practices in a series of blog posts over the coming weeks. Forrester clients can also access my report on this topic, “Best Practices: Scaling Data Science Across The Enterprise,” as well as the vendor landscape of related tools, “Now Tech: Predictive Analytics And Machine Learning Solutions, Q2 2018,” or set up an inquiry for more information.
- advanced analytics
- analytics applications
- artificial intelligence (AI)
- machine learning
- predictive analytics