2. Embrace the roles and behaviors of various stakeholders
To be successful within the last mile of analysis, close collaboration between stakeholders with the right skill sets – data scientists, business units, IT and DevOps – is essential. Lack of interest in managing the implementation and management of analytics in production and leaving it to only one team (such as IT or DevOps) or not having the right incentives for all stakeholders to communicate does not add value to your analytics or AI initiatives.
For data scientists, development of analytical assets should only be initiated with implementation in mind, while IT or DevOps teams need to understand the integration requirements, operational data streams, and data preparation for model installation and retraining. The role of business stakeholders is equally important. They are the ones who need to clearly define what benefits are expected from the analytical models and collaborate with data scientists to understand the results after the models are put into production and monitor the results continuously.
3. Establish a systematic operational process
Finally, the only way to ensure the value, integrity, and transparency of analytics models is to establish a process for operational analysis. Many organizations have a well-defined process for the analytical development phase of the analytical life cycle. However, a lack of process-centered understanding about the life cycle model installation and management phase is an important barrier that must be overcome.
A well-defined process, with correct templates and workflows, must validate that the model developed using training data still functions as intended in the real world and integrates and executes the same model against operational systems or processes. Some organizations make a mistake and stop here. To fully realize the value, the models in production must be provided on an ongoing basis.
It’s no surprise that this last mile of analytics – putting models into practice – is the hardest part of digital transformation initiatives that organizations can master, but it’s nevertheless critical if they experience real benefits from AI and analytics investment. To systematically realize the full potential of data and analytics initiatives, organizations must early involve IT and DevOps in the data science project so that analysis analysis is not an overhaul; agree on the quantifiable results before building analytical models; and have a clear understanding of the steps, roles, processes and deliverables involved, from data preparation and model development to putting analytics into action.