(US & Canada) Chris Eldredge, VP of Data Office at Paycor, speaks with Channie Mize, General Manager, Slalom Consulting, in a video interview about the various data skill sets required in a data team, the cost challenges with data, using standardized data, prioritization, collaboration with business and accountability, evolving trends in data, and building the data culture.

Paycor builds HR and payroll software for leaders and frontline managers to help build winning teams.

Data has various dimensions that require different skill sets, says Eldredge. The management of the data warehouse and data cloud platform has changed over the years, he adds.

To manage the data cloud platform, an increasing focus is required on the security, cost management, and performance fronts, as they create a huge impact, says Eldredge. When it comes to data engineering, there are two main focal areas, he notes.

The first one revolves around data movement, which involves creating and managing data pipelines. It is critical to streamline this process so that resources can support as many pipelines as possible. Further, with the rapid data growth, data ingestion becomes a continuous process that requires regular adjustments to data sources.

Therefore, the onus falls on data engineers to ensure that data can be moved reliably and consistently for it to remain effective over time. The second crucial area is data consumption, which involves preparing the data to be fit for use by stakeholders.

Regarding data marts, Eldredge states that while business intelligence and analysts operate on the front end, there is an intermediate layer of translating the data from the raw system data. Structuring this raw data derived from transactional and third-party systems into a consumable format for the downstream resources is crucial and requires skilled data engineers.

Moving on to business intelligence, Eldredge states that, here, the emphasis lies in structuring the data to tell the right story, also known as data storytelling. Coming to the data science aspect, he maintains that most data scientists do not learn to work with harmonized and certified data.

Instead, the practitioners build their features in tools like Panda rather than in centralized warehouses. Eldredge opines that these resources need to excel at aligning the model with the problems, wherein data plays a key role. He adds that a comprehensive data team must ensure that the right data and reusable features are made available for data scientists.

Moving forward, Eldredge says that the challenge with making real-time data available in the cloud faster is that it turns out to be more expensive. On top of that, manipulating the data adds to the expense, and even the objects managed within the data warehouse have different cost bases.

According to Eldredge, when considering modern data warehouses, the key questions should be around what effort is required to make the data consumable, and once it is consumable, what are the usage patterns and associated costs.

For example, if a finance team runs large queries on the data monthly or quarterly, having distributed data might be acceptable. However, if an API runs on top of that dataset a million times a day, costs can escalate quickly if the data is not optimized for consumption.

Delving further, Eldredge notes that many times the business side is not aware of the data sets other functions use. Therefore, it is necessary to ensure that certified and standardized data is clear and used first.

Analysts must be incentivized to use that data, which includes supporting these teams with appropriate alignment when they need it to be adjusted or extended. In other instances, it is also advisable to redirect them to a more appropriate data set than building a new one.

One of the things that Paycor ensures is setting aside appropriate resources and time to keep everything running and keeping things fresh. However, people working in data experience the need to show constant initiative.

This need demands prioritization and collaboration with business partners, making sure that the work is in tandem with the corporate initiatives. Additionally, Paycor emphasizes accountability, which can be done at various levels to assess the progress and ensure that delivery achieves the business value.

Speaking of evolving data trends, Eldredge mentions that the messaging around AI is to make it accessible for everyone. But, in reality, it increases the pressure on both data and applications teams to support AI applications, and the IT teams are already stretched thin.

To address this, it is essential to ensure proper resource alignment and support as these new capabilities are implemented. From a data standpoint, this involves accommodating new requirements such as enhanced security measures, the rollout of new types of applications, and establishing new connections to data warehouses.

Shedding light on building the data culture, Eldredge mentions establishing communities of practice early on, wherein everyone whose role involved any aspect of data analysis was invited to join. The primary focus was on education, understanding the tools, and becoming a trusted partner for their needs.

As 80-90% of the work then involved data wrangling, the biggest challenge was accessing and aligning the needed data. The focus was on eliminating the obstacles to shifting their efforts from data wrangling to true analysis and recommendations.

In conclusion, Eldredge states that the number one thing is trust in data, and to build that trust, it is essential to ensure data integrity.

CDO Magazine appreciates Chris Eldredge for sharing his insights with our global community.