IE 11 Not Supported

For optimal browsing, we recommend Chrome, Firefox or Safari browsers.

Washington, D.C., Uses Data to Deliver on Right to Shelter

The district is using data and technology to transform how it connects people experiencing homelessness to shelters with beds during hypothermia season. Calls to a key service hotline have dropped significantly as a result.

A person in a suit holds a glowing orange orb with three human silhouettes on top of it. Dots around the image represent data.
Image credit:
tuiphotoengineer - stock.adobe.com
Technology is playing a key role in Washington, D.C.’s work to more efficiently shelter people experiencing homelessness this winter.

In Washington, D.C., its own longstanding right-to-shelter mandate requires the government to provide expanded shelter access and related services during hypothermic and hyperthermic weather conditions. The district releases an annual plan ahead of winter to better serve and protect constituents. Winter weather conditions create dangerous — and sometimes deadly — circumstances for the unhoused, but some experts argue better data tools can help address this issue.

The district’s new shelter census tool officially launched in March 2024, enabling shelter staff to provide more information about availability to the United Planning Organization (UPO) drivers who transport those experiencing homelessness to shelters.

Because it is a right-to-shelter jurisdiction, the goal was the same before and after launching the new tool: to provide shelter for every individual during hypothermic conditions, Adam Gerstenfeld, the senior data scientist for the Washington, D.C., Family Service Administration, explained.

“But the rate at which we’re sheltering people and the efficiency with which we’re doing it is markedly different this year,” he said.

Prior to the tool’s creation, there was no technology involved in the process besides a phone, Gerstenfeld said. Staff had done hand counting with pen and paper and submitted the information nightly. Now, the dashboard leverages Esri’s ArcGIS Survey123 solution to enable hourly occupancy rate updates directly reported by shelters.

UPO drivers previously had to call shelters to get up-to-date information after picking up an unsheltered individual who called the Homeless Services Hotline for transportation services. During hypothermia season, defined by D.C. as Nov. 1-April 15, both the hotline and transportation service are available 24 hours a day. Now, hotline dispatchers can spend less time determining where to route someone for shelter. Per Gerstenfeld, since October, calls to the hotline by shelter staff decreased by 55 percent, down nearly 9,000.

“One thing that did was it made it more efficient just to get the client to the actual location,” Gerstenfeld said, noting that the dashboard frees UPO drivers’ time to transport more clients to shelters. And the reduced call volumes allowed staff to perform other important tasks, such as screening interviews. As Gerstenfeld put it, “everything gets done so much faster.”

The district works with The Community Partnership (TCP) — a nonprofit corporation that coordinates D.C.’s Continuum of Care — in a collaborative effort in which TCP is the entity that created the survey, trained providers and led on-the-ground engagement. Meanwhile, Gerstenfeld’s team created the dashboard and the Python code that enables daily reporting.

The Survey123 tool is commonly used in human services agencies in jurisdictions throughout the U.S., Gerstenfeld said, but emphasized the hourly counts taking place in D.C. set the district’s strategy apart.

One way the district hopes to build on this work is by creating another specialized Survey123 tool to better serve unsheltered individuals by informing the creation of individualized case plans; the goal is to determine better methods of connecting clients to different services they may need.

With this winter season’s end approaching, the district is already exploring how to improve the tool with new features for the next hypothermia season. This will likely include updating the display to provide a quick visual indicator of shelters with availability; for example, using different colors for shelters with availability could help drivers more quickly plan their routes, Gerstenfeld said.

In related data work, the district has a separate project called the hypothermia shelter prediction model, he noted. The data that was input into the machine learning model includes the district’s weather pattern information going back to 2007, annual Point-In-Time count data, and homeless shelter occupancy rates. This tool aims to help inform resource distribution so stakeholders can take a data-informed approach to prepare for predicted hypothermic weather conditions.

Washington, D.C., tracks the number of hypothermia nights it has had, and Gerstenfeld said this past winter season was “particularly brutal” due to a high number of hypothermia alerts. The question for future winters, as he put it, becomes: “How can we create an even better model in the future?”

Last, a dashboard of voucher-sponsored sites is in development, Gerstenfeld said; this will be used in annual performance oversight hearings with the D.C. Council to inform planning and decision-making based on where voucher clients are living.
Julia Edinger is a staff writer for Government Technology. She has a bachelor's degree in English from the University of Toledo and has since worked in publishing and media. She's currently located in Southern California.