Tim Hoolihan, Practice Lead for Data and Analytics at Centric Consulting, speaks with Robert Lutton, VP, Sandhill Consultants, and CDO Magazine Editorial Board Vice Chair, in a video interview about his background, organizational alignment of data governance practices with AI and ML, organizational strategies to integrate data analytics in business, leveraging AI tools, and methods for estimating ROI.

Hoolihan started as a software developer. Passionate about leveraging data to optimize business value, he later worked with a company in the B2B space, working with massive amounts of data. After a couple of years there, Hoolihan circled back to consulting with Centric Consulting for data analytics.

Commenting on the organizational alignment of data governance practices with AI and ML, he maintains that data quality is the key. Having good, clean, and available data sets is critical to that and good governance leans into its availability and ensuring users do that.

Further, with tools getting better, domain users are looking at platforms to build their models that, in turn, enable innovation from business and domain users, says Hoolihan. Data governance supports that aspect as well, he notes.

Moving forward, Hoolihan sheds light on strategies to mitigate potential security risks. The first step taken was towards redaction and ensuring that folks are not presented with the data they do not need, internally or externally.

It is crucial to deal with sensitive data sets by bringing in good governance practices and understanding the issues that come with HIPAA and PCI compliances among others. The burning conversation around AI governance stresses leveraging tools such as ChatGPT wisely and ensuring training users to use AI technologies safely.

When asked about successful organizational strategies to integrate data analytics in business, Hoolihan says that analytics is closely tied to optimizing business value. As a strategy, organizations should start with the user experience of a report or another analytic insight and note that it is not worth building if it does not drive business value.

Organizations should strategize prioritizing the main business functions and figure out what is needed to generate better insights to do it, and then work backward into solutioning. With the challenge of finding qualified technical staff, he maintains that the existing staff should know what drives business value.

Adding on, Hoolihan mentions baking governance and security into the platform. It is about integrating business users, starting with the business value aspect, keeping them involved throughout the process, and enabling them with data and tools to generate insights.

Emphasizing effective methods to estimate ROI for AI and ML projects, Hoolihan highlights that rapid evolution has taken place in this domain. He recalls the era when deep learning was becoming practical for product organizations, and highly specialized individuals were hired for that.

On the contrary, now, there are a lot of out-of-the-box tools such as Azure ML on Microsoft Azure, SageMaker in AWS, and Einstein Copilot in Salesforce. These tools allow getting into AI without writing a custom model, he adds.

Taking an instance of supply-chain interruption, Hoolihan shares how these tools come at a lesser cost and help in producing output prediction 70% of the time. This can enable someone to take an actionable step and intervene with the business.

Additionally, investing in building a custom model depends on the business value and use cases, says Hoolihan. Even if the accuracy problem persists after doing it end-to-end, he advises investing in a customized model, but deferring until most data is accumulated.

In conclusion, Hoolihan states that even if it were to fail, using an out-of-the-box PoC that uses tools such as Sage Maker would fail fast before spending millions in building custom models.

CDO Magazine appreciates Tim Hoolihan for sharing his insights with our global community.

IT is the Time for the Data-Centric CIO

Related Stories