From Fragmented Data to Shared Outcomes: Improving Homelessness Services Data in Albuquerque, New Mexico
Guest blog by Charlie Verploegh, IPP ’26
When I began this work, my implementation challenge seemed relatively straightforward: How can we better understand whether Albuquerque’s investments in homelessness services are actually improving people’s lives? At first, a dashboard seemed like an obvious solution. We invest significant public resources in homelessness services, and we should be able to clearly answer some basic questions: What did we spend? Who did we serve? What happened to the people we served?
What I learned through the PDIA process is that creating a dashboard is the easy part. Creating the conditions for the information on that dashboard to be accurate, meaningful, and useful is much harder.
Understanding the Problem Before Building the Solution
The fishbone exercise was particularly helpful because it forced me to break down what initially looked like a technology problem. Our analysis identified challenges across data and technology, provider burden, data quality, leadership and accountability, funding and capacity, program design, and communication. We had multiple systems that did not communicate with each other, inconsistent data collection, different definitions of success, limited analytical capacity, and no single shared dashboard. We were also asking providers to meet complicated reporting requirements while their frontline staff were understandably focused on serving people experiencing homelessness.
The exercise also exposed a more fundamental issue: programs did not always share the same goals or outcome measures. Contracts often focused more heavily on activities than outcomes, while the results of our investments were difficult to communicate clearly to policymakers and the public. This changed how I thought about the challenge. We did not simply have a dashboard problem. We had a data quality, shared outcomes, systems, capacity, and accountability problem.
Finding an Entry Point
PDIA helped me recognize that we did not have to solve every part of that problem at once. Instead, we needed an entry point that was both meaningful and achievable. We chose to focus on developing a short-term dashboard while simultaneously improving the data underneath it. Our fishbone work helped us identify several practical steps: determine the measures that should be shared across contracts, communicate those expectations to providers, train providers on entering and reviewing their data, determine where the information should live, and develop both internal and external ways of displaying results. That gave us something concrete to work toward while continuing to learn about the larger system.
Moving From Planning to Doing
Since developing that initial analysis, we have made meaningful progress. One of the most important steps was auditing our existing data. Rather than assuming the information in our systems was ready for visualization, we examined what was actually there and where there were gaps, inconsistencies, or errors. We then worked directly with service providers to correct their data and successfully upload it. This was important because data quality cannot simply be imposed through a reporting requirement. We needed providers to understand what was being requested, identify problems with their information, and work with us to improve it.
At the same time, we narrowed down the outcomes we actually want to communicate. Our original analysis identified shared measures including housing placement, housing retention, income growth, employment, and returns to homelessness. It also envisioned a public-facing dashboard organized around three understandable questions: What was spent? Who was served? What happened? That simplicity has become increasingly important to me. We can collect hundreds of data points, but more data does not automatically create more understanding.
Now that much of the underlying data has been cleaned and the primary outcomes have been identified, we have moved into another iteration: vetting dashboard visualizations and determining the best technical solution. We are currently exploring whether the dashboard should be developed through an internal City system or with an external data vendor. That means we have not arrived at the final product yet. But PDIA has helped me see that this is not a failure to finish. It is part of the process of testing what is feasible before committing to a solution.
What I Learned
Perhaps my biggest learning from this experience is that implementation requires spending more time understanding the problem than I naturally want to spend. There is tremendous pressure in government to produce something visible quickly. Our original fishbone even identified the “extreme urgency to getting something up quickly” as one of the pressures surrounding the project. PDIA challenged my instinct to respond to that urgency by jumping immediately to a solution. If we had simply built a dashboard using the data we had at the beginning, we could have produced something attractive relatively quickly. But an attractive dashboard displaying incomplete or inaccurate information would not have solved our problem. It might actually have made it worse by giving people confidence in information that had not been sufficiently validated. Instead, we audited the data, engaged providers, corrected problems, clarified the outcomes we care about, and are now determining how best to visualize and maintain that information.
How This Will Change My Approach
Going forward, I expect to use PDIA less as a formal exercise and more as a way of approaching complicated problems. I will ask: What problem are we actually trying to solve? What assumptions are we making? Who experiences the problem differently? What is within our authority to change? What is a reasonable entry point? And what can we learn by taking a small action rather than continuing to plan?
I have also learned to be more comfortable with iteration. Government often creates pressure to identify the solution, develop a plan, and execute it. Complex social problems rarely cooperate with that sequence. Homelessness certainly does not. The better approach is often to establish a clear purpose, take an informed step, learn from what happens, and adjust.
A Final Reflection for Other PDIA Practitioners
My advice to other practitioners would be simple: do not confuse producing something with solving something. A dashboard is a product. Better decision-making is the goal. The work underneath the dashboard, agreeing on outcomes, improving data quality, working with providers, understanding different systems, testing visualizations, and determining how information will actually be used, may be less visible than the final product. But that work is what makes the product meaningful. Our dashboard is still being developed. We are still testing, learning, and making decisions about what the final system should look like. But today we have cleaner data, clearer outcomes, stronger engagement with providers, and a much better understanding of the problem than when we started.
This is a blog series written by the alumni of the Implementing Public Policy Executive Education Program at the Harvard Kennedy School. 33 Participants successfully completed this 5-month hybrid program in September 2026. These are their learning journey stories.