PNC

Nearly every large organization understands the value of analytics and data science in improving decision-making, and has made investments in personnel and infrastructure to that end. Yet many don't have a comprehensive roadmap for upskilling their teams.

Relying on external hiring alone to address skill gaps is cost-prohibitive, especially in today's job market, so developing internal talent is a necessity, and a systematic upskilling program is the best method to accomplish that. Internal hires often have company knowledge lacking in external hires, especially in the areas of business literacy and data literacy, which, along with technical literacy, are needed for maximizing return on analytics investment.

I have worked in analytics for different organizations in different industries for over twenty-five years, and I have been an analytics educator for over ten of them. In addition to learning from the organizations for which I have worked, my students have shared their own workplace experiences with me in my capacity as an adjunct professor. From our combined experiences, I have learned of several ways that high-performing organizations are meeting the upskilling challenge:

• Creating an analytics/data science accelerator

As part of the team that developed the Quantitative Analytics Development Program at PNC Bank, I have seen firsthand how an analytics accelerator effectively grows talent. This program, now in its sixth year, alternates periods of formal education with immersive apprenticeship to multiple analytics teams in rotation. Participants graduate from the program having familiarity with products, business processes, data sources, as well as the technical agility to apply what they have learned.

• Sending your analysts to analytics conferences - internal or external

Hands-on offsite conferences can be effective for the right analysts, but as an upskilling method, this is too costly to scale. As an alternative, consider having internal conferences just for your own teams. Whether demonstrating how to code a random forest model in Python, or teaching data users where and how to find information in the corporate data lake, or explaining why Simpson's Paradox can result in an incorrect marketing analysis. These are opportunities for your analysts to teach and learn at low-to-moderate cost. Keeping the conference virtual can help manage costs even more.

  ​Relying on external hiring alone to address skill gaps is cost-prohibitive, especially in today's job market, so developing internal talent is a necessity, and a systematic upskilling program is the best method to accomplish that 

• Scheduling periodic learning meetings

Much less costly and labor-intensive than conferences, weekly one-hour meetings can cover presentations on specific topics, or serve as something like a professor's office hours, where people can ask questions and get answers.

• Creating internal online forums

When questions cannot wait for the next regularly scheduled meeting, posting them internally and getting answers from knowledgeable peers can help improve skills and productivity.

• Offering online training

Enrolling in degree programs is neither feasible nor desirable for everyone, but online training has become much more robust, especially during the pandemic. There are many effective and affordable offerings in this space.

• Offering instructor-led training

While more challenging to scale cost-effectively, instructor-led training is probably the most effective form of training for most analysts. This option requires a greater commitment of time and funding than many others, and determining those for whom it is most appropriate will take some thought.

• Creating a data dictionary/metadata repository

Analysts are sometimes asked to analyze data that they cannot precisely define, or possibly worse, build predictive models using that data. Defining key data elements in a centralized, easy to use and accessible format is critical. At a previous employer of mine, I was on an analytics project team consisting of forty people. We wasted two weeks working on a piece of that project almost entirely because we were not defining the relevant terms consistently. We stopped spinning our wheels once we realized that, and then used consistent definitions thereafter. It was a costly lesson. While it is possible to succeed without a centralized repository for this information, having one increases speed to market, decreases waste and frustration, and reduces the risk of supplying incorrect information to decision makers.

• Adjusting workload to allow time for upskilling

This is the bottom line, where organizations can show that they really "get it." Yes, every hour spent on upskilling is an hour not spent on "real work", but with an effective upskilling program, one hour of real work today might need only thirty minutes of work in the future. Upskilling must be viewed as an investment in the organization, and the required time must be competitively prioritized against other activities. A realistic cost-benefit analysis of the program can size the opportunity for decisionmakers, and ideally will convince them of the advantages of a formal upskilling program.

• Recognizing teachers, including those without formal teaching responsibilities

Finally, those who teach must be recognized for the value they add to the organization by helping others work more effectively and expanding their capabilities. This is especially important for highly knowledgeable analysts who are not in a formal training role. Penalizing them for spending time helping others instead of devoting all their time and effort to their "real work" is detrimental to the organization. Instead, recognize them for their expertise and make training others part of their "real work", to be prioritized along with their other responsibilities.

Many cost-effective educational options exist for improving an organization's return on their analytics investment, and most of them scale very well. Most of these options have the added benefit of transforming individual analysts into a more cohesive analytical community. Once the commitment is there, gradual implementation of these options is relatively straightforward.