IBM Service Corps uses predictive analytics to find lead in drinking water pipes

By and Laura Gilligan | 3 minute read | August 17, 2020

On #NationalNonprofitDay, IBM Service Corps celebrates its work together with CNT and BlueConduit to reduce lead in drinking water. 

On a hot summer’s day, a cold glass of water refreshes and energizes.  But what if you couldn’t trust the water that came out of your tap? What if you weren’t sure if the pipes leading into your house were made of lead? The CDC has established that there is no safe level of lead in drinking water.

Sadly, even in this day and age, many cities lack accurate records of lead service line placement and do not have the resources to gather this data.  On top of that, identifying and replacing lead service lines is a slow and expensive task.  A Chicago-based IBM Service Corps team partnered with the Center for Neighborhood Technology (CNT) to help collect data on the location of lead service lines and document a playbook of best practices in two Chicago suburbs, Hazel Crest and Flossmoor.  This playbook will serve as a guide for communities on how to tackle finding, predicting, and documenting lead service lines.

Our IBM Service Corps team brought together data scientists, business analysts, project managers, UX designers, and Enterprise Design Thinking practitioners to document the playbook that can be used in Hazel Crest and Flossmor.  It was vital that the playbook be a series of repeatable processes for other communities.

The teams joined forces with BlueConduit to apply the lead service line predictive analytics method they developed and deployed in Flint, Michigan following the 2014 drinking water crisis.  The IBM team used the predictive analytics method as a starting point to create our own lead service line predictive method that could be used in the Chicago suburbs.

All three organizations jumped in, ready to share our expertise and work together.


We kicked off the project with a Design Thinking session focused on outlining the work ahead of us, from clarifying what data to collect, how to collect it, all the way to documenting the process. As we were preparing to meet again, COVID-19 shut down our workplaces, so we quickly transitioned what work we could to Slack and Zoom.

The data from the communities public works departments are in paper documents of various size, age, and condition.  The CNT team was in the process of scanning the data into a digital format, but had to stop as the offices closed. This gave us some data to use, but not as much as we had planned.  The predictive model relies on large amounts of data in order for it to be truly predictive. The IBM data scientists tried different model options and were successful in creating one that worked well, and would allow CNT to use and expand upon once they are able to collect and scan additional records.

Putting smart to work

While the data scientists were working on the scanning and model portion, the rest of the team focused on the more consultative deliverables.  We were successful in framing questions for community interviews, outlining key considerations for the development of tools to help the public works team visualize possible locations of lead service lines, and developing a playbook that CNT can use to repeat this project in other communities.

The predictive model created and the playbook will be invaluable as CNT and the communities work with additional data.  The more data captured, the better the model will get in predicting lead service lines.  With greater success in prediction, this will lead to the identification and removal of lead in the drinking water.

The playbook provides a comprehensive process that CNT can build upon as they expand into more communities in need of assistance to determine where there are lead service lines.  We have emphasized the importance of data accessibility and quality, understanding the data and its relevance, defining clear goals and setting limits for the project, and being able to make rapid decisions to keep the project moving forward.

The teams were able to accomplish a preliminary predictive model and a playbook to allow CNT and BlueConduit to continue their mission of helping communities ensure that all residents have safe and clean drinking water.  Data analytics and technology can play a role in detecting lead service lines and facilitate in mitigating this issue.  IBM was proud to partner on this project and use its technology, resources, and people to make a difference on this local Service Corps project.


IBM Service Corps is an innovative social impact program that develops IBM leaders, while contributing IBM talent and technology to local communities and non-profit organizations looking to tackle challenging problems.