Geographically speaking, Multnomah County is Oregon’s smallest county, but it’s also the state’s most populous. The majority of residents live in Portland, well known for its spectacular mountains, forests, food, and art scene. However, like many American cities, Portland is also experiencing a homelessness and housing crisis. Of the county’s 800,000 residents, approximately 15,000 are experiencing homelessness—and each year, those numbers increase due to a housing shortage that makes rent unaffordable for many residents. In Multnomah County, homelessness is a true crisis in need of effective solutions.  

While the numbers alone presented a staggering challenge, for those tasked with ending the crisis, the situation was far more layered—and individualized—than the public could see or understand. And the public had questions. With the record-breaking funding being raised to end the homelessness crisis, residents understandably wanted to know how dollars were being spent and when they’d see results. But without the right platform to access data on the county’s services, these were very difficult questions for Multnomah County to answer. Even though the county had done years of work to improve its data, its antiquated systems meant extracting and sharing that data was a cumbersome process that could introduce inaccuracies. 

“As someone who lives in Portland and has homeless encampments near my house, the most meaningful work for me was understanding what our orchestration looks like,” says Stephen Patch-Putt, a senior consultant at Slalom. “When I learned I’d be part of this project, I went to a homelessness consortium meeting. One of the common threads was the lack of data—of good data—that was available to researchers and providers.”

When Jenna Kivanc joined Multnomah County’s Homeless Services Department as data manager, she brought fresh eyes and an extensive background in public health to her new position. “I was trying to solve the problem of how we could modernize, get the data into a data mart so we could monitor the quality and improve it in order to make better decisions—because one of the things we’re trying to do here is be as transparent as we can possibly be,” says Kivanc. 

That’s when Dan Cole, the county’s IT portfolio manager, reached out to Slalom. Together, we embarked on a journey to create a data-exchange platform in Snowflake with Slalom’s Homeless Response System Accelerator. “To provide decision makers with faster access to information, we needed a more accessible and usable data solution,” says Cole.

As part of our discovery process, Slalom led two workshops—one exploring intake in various places like shelters, rehabilitation facilities, and hospitals, and the other focused on bidirectional reporting to ensure the county could present data in a way that kept key decisionmakers informed. The team came to understand not only how difficult it is to count a transient population but also how critical it was for Multnomah County to have good systems for both data collection and data cleansing. “It wasn’t a surprise that they were experiencing data-quality issues—in dire situations like these, data hygiene tends to take a back seat to boots-on-the-ground needs,” says Jack Murphy, principal at Slalom. 

Working closely with Multnomah’s skilled internal team, we learned that the data sources used for reporting were siloed in separate systems, and the county’s processes were intensively manual. So as dire as the crisis already looked on paper, the number of people experiencing homelessness was actually double what officials had believed it to be. “Once the Multnomah County team was able to explore our reports, they could drill down to the underlying data to gain more insight into trends and patterns within our local homeless services system,” says Patch-Putt.  

Up to that point, understanding local trends and patterns meant stitching together reports from different states over several years. According to Murphy, “The biggest challenge was creating a common language for multiple analysts who work on different styles of reporting and creating consensus on what their target measurements were supposed to be.” Snowflake served as kind of a foundation for all those different feeds into the county’s system, essentially taking snapshots of what happened every 24 hours.