(US & Canada) VIDEO | Don't Invite AI Talent to the Party Without Building the House — Hitachi Solutions Data Analytics VP for Canada

Simon Nuss, VP of Data and Analytics, Hitachi Solutions, Canada, speaks with Jason Brandt, Managing Partner and Commercial Officer at Stagwell Technologies, in a video interview about his journey in the data and analytics space, leveraging data to meet business objectives, challenges in deriving monetary value of data, application versus analytics layer, bringing in AI talent at the right time for the right job, the challenges faced by data science communities, and how to build better analytics departments.

Nuss has been in the data and analytics space for close to 15 years, beginning his career with dimensional data modeling (data warehousing). He has worked with organizations around the world, such as Facebook, Deloitte, banks and investment funds, as well as Hitachi Solutions. He is also the moderator of the Power BI Subreddit, which is close to hitting its 90,000 members milestone.

At Hitachi Solutions, Nuss leads a team of consultants specializing in data collection, governance, report development, training, data engineering, and data operations. He also mentors non-technical leadership while discussing and analyzing the data stack, including performance optimization, SSDLC, and architectural best practices.

Commenting on data usage, Nuss states that many organizations nowadays include data in their strategic pillars. Many leaders have a good sense of the effective use of this data to meet their business objectives.

However, actually executing a data warehouse and proving its monetary value is challenging, and many organizations are not measuring the success of their efforts. Thus, many are trying, but few are hitting the target, he asserts.

Further, Nuss states that the present economy has had major impacts, which are very evident in the budgets. He notes that it is difficult to prove the value of what is being done and therefore justify the budget.

As such, it is important to measure and report what is done to make a case for the budget the following year, suggests Nuss. Otherwise, even if one is successful in proving their worth, the organization may not have the money for it as the current environment is quite difficult.

Moving forward, Nuss suggests that when it comes to budgeting and allocating resources, the application layer is prioritized more than the analytics layer. He notes that when there is a financial crunch, the C-suite may not have as much power or influence.

In addition, he maintains that an organization will not cease to function if the data warehouse does not refresh or the CEO does not get their weekly report. For example, McDonald's will still be able to take orders even without sales analytics.

Furthermore, Nuss suggests that an organization should not invite its AI talent to the party without first building the house. He says that executives often say they use AI without realizing how it should operate within their organization.

In continuation, Nuss notes that AI can be fantastic for software-as-a-service (SaaS) companies but it may not have a place in a pie shop. Therefore, he warns that hiring data scientists before building momentum and use cases can have a downside.

Additionally, he observes that data science communities in forums such as Reddit complain that when hired, they are promised to work with interesting data sets and to execute Python scripts. But in reality, they are locked in a basement and end up instead writing SQL and creating VBA Macros.

Delving further, Nuss says that data scientists are expensive resources that need to have the right types of jobs for their skills. To resolve this, he suggests a professional consultant do the groundwork of defining staffing requirements, JDs, and data platforms.

He elaborates that the consultants can also have conversations with the business about what they need and map it to data science use cases. Nuss further advises establishing governance and guardrails and getting some momentum and appetite for AI/ML before bringing in a team of data scientists.

Thereafter, he discusses his way of building better analytics departments by following a simple formula. He starts by creating a "shiny POC" (proof of concept) — something of value that the business has not seen before, such as a Power BI report or real-time alerts.

Once created, the POC should be taken on a roadshow to build interest in it from business users and use it as a basis to capture use cases and pain points. This can then be presented to leadership as part of a budget request for the operational expenditure to hire headcount.

In his closing remarks, Nuss confirms that the entire process can take time depending on when one joins the organization, but it has been a success every time.

CDO Magazine appreciates Simon Nuss for sharing his insights with our global community.