Citizen Data Scientist Programs Can Help Close Your IT Talent Gap

Are you struggling to fill technical roles in your organization? Is it difficult to find the digital and data talent required to enable the business transformation programs your business leaders are demanding, and that you’re driving?

You are not alone. There’s a simple reason why the skills gap is more pronounced today than during previous technology booms. The current boom is creating a talent gap because “existing skills in older analytics concepts and technologies are insufficient for working with real-time data from Internet of Things sensors, social media, mobile computing and other modern sources of data.” 

There are different steps CIO can take to close the skill gap, but one that I have been focused on recently is citizen data science programs. Citizen data science programs empower business teams to self-develop and serve analytics without having to engage data and analytics specialists, developers or IT. 

With citizen data scientist programs, the analytics are more likely to be adopted by end users, and the burden on IT to develop and maintain reports is reduced. It’s not a practice that can be instituted overnight, but the rewards can be significant, especially when it is so difficult to find outside data talent. 

Risky Data Landfills 

Most CIO recognize the issues with how their organizations currently analyze data. Many have investments in last generation BI tools that require significant IT skills to administer, develop, and support. These reports are often ill suited for today’s business leaders looking for more real time reports with data integrated from multiple enterprise and external data sources. 

When business users don’t get the data they require, they resort it to DIY (do it yourself) methods which usually amount to extracting data from a source system, analyzing it using spreadsheets, and developing presentations to report on it. Many CIO cringe at the size of their data landfills with manually developed analytics that are inefficient, error prone, and that raise security and privacy concerns. 

Then there are the Big Data infrastructure and data science programs many of you have invested in, and from which some are extracting value. These are another source of the skills gap as they often require Hadoop developers, cloud engineers, and PhD data scientists to configure the infrastructure, wrangle the data, and program the analytics to develop ongoing insights. While this approach is needed for the most strategic enterprise data initiatives, it is expensive and often under-resourced. 

Necessity is the mother of invention 

I stumbled on citizen data science when I was confronted with the challenge of upgrading the analytics of operations and marketing teams. They were overwhelmed with data, but only a few people in their organization had access to it. The data was often outdated, poorly integrated with other data sources, and suffered from various quality issues. These departments were mandated to become more efficient, improve operational quality KPIs, and increase the quality of leads generated for the sales group. 

They had the desire to use data driven practices to close the gaps, but the IT team was completely dedicated to improving the user experience of our customer facing products. Not only didn’t we have people with the skills available to address these needs, we didn’t have a budget for outside help. 

Or did I really need it? Someone in the operation’s group was using Microsoft Access to pull all the operational data together to report on it. Someone in marketing was using Excel to grab digital marketing data from multiple sources to determine what activities were generating the best leads. Was it possible to retool these individuals, teach them modern data practices, and evolve data governance so that they would be successful self-serving the analytics required? Could they develop dashboards for their colleagues to drive better and more frequent decision making?

Absolutely! 

The individuals we engaged to be citizen data scientists were taught new technologies and how to evolv the usability of their dashboards and analytics. As the CIO, I provided them with access to technologies and training by internal experts. If they got stuck getting data out of enterprise systems in the right format, then I brought in database specialists from the technology team to lend a hand. When they needed to do more advanced data modeling, analytics, or programming I helped bring in more experienced data scientists and technology experts. 

I partnered with their business leaders to ensure there was agreement on the priorities and solutions. When development milestones were reached, demos were organized so their business colleagues could see progress and provide feedback. Their users embraced the easy-to-use dashboards and were happy with the quick turnarounds to their requests. I then helped the business leaders institute completed dashboards into their business processes.   

As the programs matured, we rolled out data governance, visual design standards, and formalized agile processes. We also took steps to grow the program to other departments. At one business we scheduled a company-wide town hall to showcase the results and to recruit more departments to participate in the citizen data science program.

Lessons Learned From Instituting Citizen Data Science Programs 

Having rolled out these programs a few times, let me share a few lessons learned. 

1. Citizen data science programs are best instituted bottom up rather than top down.
   Unfortunately, your leadership team has been exposed to many reporting solutions failing to deliver on promises. Find a department with a meaningful business problem, select individuals with some basic data and analytics skills, and ensure the department is willing to embrace change. Achieve success with one or two departments before trying to announce a program and bring executives on board. 

2. Get involved in reviewing tools.
   There are many SaaS providers claiming their analytics tools are for business users. Unfortunately, some are too simple to address real business data issues, or they are too complex for business users to be successful. But there are a sufficient number of tools to choose from. Your selection criteria should consider factors such as data complexity, types of analytics, volume of data, number of users, and skills of the targeted citizen developers. 

3. Expect to spend a significant amount of time re-engineering the data services offered by your IT team.
   DBAs have largely thought their role was to keep operational data warehouses up and performing, but citizen data science programs require them to virtualize data sources into consumable resources. Process leaders also have to consider how to take practices such as version control, quality assurance, and release management into business processes followed by citizen data scientists. 

See my related blog post: How to Kick Off a Citizen Data Science Program.