A couple of weeks ago, I had the opportunity to interview Sharon Kratochvil, global data & analytics executive for Capri Holdings Limited. Capri Holdings is a global fashion luxury group, consisting of iconic brands (Michael Kors, Jimmy Choo, Versace) that cover the full spectrum of fashion luxury categories including women’s and men’s accessories, footwear and apparel as well as wearable technology, watches, jewelry, eyewear and a full line of fragrance products.

Kratochvil has an interesting background. Not only is she a chief data officer (CDO) who holds a PhD, she is a trained econometrician (someone who uses statistics and math to model and predict economic outcomes). I was impressed by her expertise when she spoke at a recent CDO Club event, where she advocated for two important new notions: There are data product managers and data products. She is truly on the vanguard of this trend and provides a valuable model for CDOs to follow. According to Kratochvil, the rise of data products and product managers has occurred because traditional business intelligence projects were too siloed.

What Is a Data Product?

Kratochvil’s perspective about data products contrasts with analyst firms. She argues that firms should have only a small number of data products. The goal for these products should be to support a strategic ecosystem, and each data product should include and integrate multiple data sources, with data enhancements (i.e. engineered features, algorithms, scores), and the platforms associated with the data product. One data product as an ecosystem is used across multiple functional areas to drive a set of business outcomes.

The move to data products makes data strategic. At Capri, their businesses consider data products to be strategic data assets. And this has led to a natural distinction between base data and data products. Base data is a single source. Whereas data products are always multi source and built for specific business function.

Capri has multiple data products. Capri’s first data product creates a single customer view. It combines CRM, purchase, SMS, and email data with loyalty information, as well as other customer interaction data. The goal is to integrate customer data sources. They call this data asset a single view of customer data asset. Kratochvil stresses that this is first-party data and individual instances exist for each brand; for example, Michael Kors.

Their second data product is about digital data. It combines website data, mobile application data, visitor identity, visitor segments, third-party platform integrations and consent management. This product combines data from digital channels, pulling selectively from their digital ecosystem. Compliance and customer experience are core focuses for this data product, as they need to both protect and leverage this data effectively.

On the protect side, they must conform to various digital privacy restrictions such as Apple’s App Tracking consent, and browser tracking prevention. Here, it is important to keep relationships in the center — merchants, finance, tech platform vendors, etc. It’s also important that they productize this data for other teams, so they can use it to power digital offerings in partnership with business groups focused on CX.

Related Article: Chief Data Officer: Blossoming Executive or 'Unsettled Role'?

What Role Do Data Product Managers Play?

Data product managers play a key role, argues Kratochvil. She believes this is an emerging role for talented folks in data.

In terms of skills, she claims data product managers need to have business knowledge, technical knowledge, and in her case, knowledge of the data in the subject area at or exceeding the volumes of data with which the team works.

It is also important that data product managers understand their enterprise’s business model and how it works. They should understand the problems that need to be solved with data and be familiar with the analytics and insights to be produced. Data product managers need to be great collaborators and communicators and comfortable acting as gatekeeper and shopkeeper for their business unit.

Getting Comfortable With AI and Machine Learning

As Kratochvil described the role, I realized that my role as a product manager for HP Software’s Data Products was very similar. Data product managers first need to be responsible for data quality and the methods for remediating it. They need to know how to look at data and detect issues, understand bad data and know how to validate data from a source. And then take responsibility for determining how to best remediate bad data.

Learning Opportunities

Data product managers don’t implement data models, but must understand how machine learning and AI work. This is what I did at HP Software with our advanced analytical models. Data product managers are also responsible for ensuring data is fit for purpose and is readily available in a usable form. This allows them to focus on vital tasks, like the data strategy, data roadmap, and enhancements to their data product based on the needs and priorities of the business.

To do all of this, data product managers must have technical skills and an understanding of source systems alongside the corporate business model. They also need to be able to manage vendors and analyze the effectiveness and efficiency of data collected. Organizations should have “a dedicated team focused on data governance and products,” argues CIO Pedro Martinez Puig. They also need a “transparent portfolio of data initiatives and the maturity level to expand geographically and look for cross-business opportunities.”

Related Article: CIOs Warm to the Chief Data Officer

Qualities of a Data Product Manager

Kratochvil claims that successful data product managers have several qualities together. First, they are curious about data. What does it say? What does it mean? Critical thinking is also essential; beyond this, they must understand the subject area. This includes big data and big data management. The job requires they work with terabytes of data. Data product managers must have a visceral feeling for data and what it means; they need to grok analytics and decision-making.

So, with the qualities defined, I asked who can be a data product manager. She suggests that anybody can grow into this kind of role. But she was clear: this is different than a data scientist role. They are not a data scientists; the more technical work is done externally.

Given this, legacy business analysts can become data product managers. They must understand data science and the role it should play. This includes feature engineering or extraction. Here, domain knowledge is critical to extracting features (characteristics, properties, attributes) from raw data. The motivation is to use these features to improve the quality of results coming out of a machine learning process.

Former CIO Isaac Sacolick agrees with Kratochvil: “Data has always needed an agile model with product management and delivery leaders, but their skills must include data management, DataOps and data governance + UX/CX. It also must go deep into the analytics and integrations customer need.”

Parting Words: Data Is at a Crossroads

Without question, data is at a crossroads. It is time for it and its processes to mature. A key element of this is to move from thinking about data as siloed and sitting separately within transaction systems and instead as a blend of data compiled into data products. And no one is better suited to manage this than someone skilled in data and product management.