According to a recent McKinsey study, the majority of companies are using artificial intelligence (AI) in a single business process, and thus gaining only incremental benefits. One contributing factor for many organizations is siloed environments in which data scientists and engineers are hired for functional and regional areas of the business, and therefore lack a consistent and comprehensive view.
If this sounds familiar, consider the steps outlined below to better connect your teams and build competitive advantage.
1. Create Better Visibility Between Projects
It’s not uncommon for organizational boundaries to create friction between technical teams that prevents understanding of project overlap and redundancies. To address this issue, many companies establish a searchable feature and model catalog, enabling data science teams to reuse valuable data features and save both time and processing power. This solution also engenders greater efficiencies because data scientists can check the catalog before creating anything new.
2. Spotlight Select Cross-Functional Projects
Creating a select group of focused initiatives around key business areas that would have a cross-functional impact typically draws the interest of stakeholders across the organization. Think top-level predictive key performance indicators (KPIs), customer initiatives, or next best product outputs. One way many organizations kick-start these projects is through hack-a-thons. These events lay the framework for the collaboration required to strengthen the data science initiative and also expose data scientists to other approaches and areas of the business.
3. Strive for a Single Data Science Production Environment
Establishing a single environment that addresses all data scientists’ needs and can be adopted enterprise-wide significantly reduces friction between disparate departments and business units. Introducing DevOps processes creates greater efficiencies in numerous ways, including consistent model validation and monitoring standards, peer reviews of code and methodology, and version control.
4. Create New Connection Points Internally and Externally
Particularly in large, distributed environments, many data scientists may work alone in business units or functional teams without the support of peers with similar skill sets. Creating a regular meeting or forum for these individuals to connect, share challenges, and highlight good work can make a huge difference, especially for junior team members.
As with any initiative, building a strong data science community is a journey with many challenges, obstacles, and small wins unique to the organization. Regardless of where your company may fall today, the next step in the journey requires greater alignment of the people, processes, and technologies needed to get there.
For more on the steps outlined above, take a look at this Enterprisers Project article.