ACT Data Hub homepage with an interpretation guide and map

All Children Thrive - California: Shared data infrastructure for locally defined community questions

For California community coalitions, I proposed and built the ACT Hub to connect shared public-data resources with definitions, limitations, analytical applications, interpretation guidance, and project-specific support.

From scattered resources to a question-centered Hub

Project period: 2025–2026
Work: Hub discovery and information architecture, user research and testing, research-data systems, GIS and interactive maps, applied analysis, interpretation-guide design, documentation, and data technical assistance

All Children Thrive supported California coalitions working on housing, homelessness, alternative response, education, environmental conditions, mental-health resources, youth spaces, and other conditions affecting children and families. Coalition teams brought local knowledge, defined the questions they wanted to pursue, and set their policy goals. My role within UCLA’s data technical-assistance program was to help translate those questions into data discovery, appropriate measures, analysis, maps, interpretation, and communication support.

Through direct technical-assistance sessions and a review of where teams were finding maps, datasets, and guidance, I identified a recurring problem: sessions often returned to locating individual variables and rebuilding basic maps, leaving less time for scale, context, limitations, and what the evidence could support.

Based on that discovery, I proposed and built a central ArcGIS Hub that combined frequently requested datasets and applications with the interpretive guidance I had otherwise been repeating during assistance sessions. The aim was to let teams explore independently and bring a more developed question into direct assistance.

ACT Hub homepage behind a combined-variable exploration application
The ACT Hub brought shared data, applications, and interpretation guidance into one project environment. Open full-size visual

A Hub organized around how questions develop

The first version, built in spring and summer 2025, was a searchable catalogue. Datasets, maps, applications, and guides were grouped by resource type and linked from a central page. That structure assumed users knew what they were looking for and what it was called.

I ran the first formal testing round with five users, each working through a current project question. I asked them to locate relevant information, interpret it, and decide how they might use it. Users who knew the concept they wanted to investigate often lacked the dataset terminology needed to locate it, and search terms varied from project to project, so relevant variables went unnoticed. Place-name searches had a similar gap: a search for Big Bear did not surface the San Bernardino County map that covered it at the tract level. Teams also described what they needed in terms of what they had to produce, not data categories.

What I observed changed the design. The first information architecture began too far downstream. In fall 2025 I reorganized the Hub around the deliverables teams were working toward: conducting a needs assessment, educating decision-makers, and building a narrative. It kept a hub-and-spoke structure, now with an Experience Builder map application at the center as the single source for the maps, linking out to StoryMap tutorials, reference guides, and phase pages. I deferred further catalogue development, since better search would not help users who did not yet know which variables applied to their question.

The final round of testing showed that each part worked on its own, but users stalled between them. They could find a relevant resource and understand it, but were less certain about the next step. The missing piece was the connection between steps, not another resource.

That finding shaped the final design: a teaching thread running through the Hub along a progression of Find → Understand → Analyze → Communicate. Prompts on each page point to a next step based on what a user has explored and how well they understand it, moving from a mapped indicator to its definitions and limitations, then to combining indicators into an analysis of their own question, checking that analysis with data partners, and communicating the findings to decision-makers.

ACT information architecture in three stages: a catalogue grouped by resource type, a structure organized by deliverable, and a guided path through Find, Understand, Analyze, and Communicate
Each round of testing changed the Hub’s structure: from a resource catalogue, to paths organized by deliverable, to a guided path connecting each step. Open full-size visual

One analysis guide made that path concrete. I paired a school-performance map with “What the Pattern Suggests,” then structured the guide to move from a visible association into questions about schools, neighborhoods, other explanations, and possible action. The red-dot pattern becomes a starting point for interpretation and communication rather than a conclusion standing on its own.

ACT analytical map behind interpretation guidance with four questions
An analytical guide pairs a red-dot map with “What the Pattern Suggests,” carrying a visible pattern into interpretation questions and possible next steps. Open full-size visual

In spring 2026, an advisory subcommittee reviewed the redesign and how the reference guides were framed for non-technical users; their feedback was minor. A final round of 30-minute tests with project staff checked whether the thread held across the full path, and four users completed it. All 14 coalitions had access, and five active teams returned to the Hub beyond testing.

The combined-variable map became the Hub’s most-used component. As teams could examine several conditions before a session, the requests I received became less about finding a single variable and more about comparison and interpretation. In City Heights, for example, a housing-cost question expanded into demographic change, community history, collective efficacy, and community bonds.

Across the engagement, I produced more than 35 layers, curated datasets, maps, applications, dashboards, walkthroughs, and guides. To produce recurring Census-based maps for the guides more efficiently, I wrote a reusable Python workflow; one preparation task fell from about four hours to about 30 minutes.

The ACT project period closed in 2026. I documented the Hub’s information architecture, production routines, and analytical guidance so they were not dependent on continued operation of the public deployment.

Stockton: comparing youth and city-selected parks

An earlier city-aligned process had identified park-renovation candidates using income, rent burden, environmental conditions, green-space access, and other ParkServe priorities. Youth participants did not recognize many as parks they used. They assessed alternatives through their own knowledge of use, safety, broken facilities, tree planting, lighting, access, and other local conditions.

The youth and project team then asked how their selections compared with the ParkServe priority-area layer used in the earlier process. I performed that comparison using 10-minute walksheds. The youth-selected parks overlapped the layer by 67 percent, 92 percent, and 91.3 percent; the city-selected parks overlapped it by 78.3 percent and 72.4 percent. Two youth values exceeded both city values, while one was lower.

These percentages measure spatial overlap with one priority layer, not overall park quality, and do not replace the youth assessment. I built a focused map with park walksheds, priority areas, overlap percentages, and optional context layers, then prepared guidance for presenting the comparison in policy discussions. The project developed a recommendation for a funded youth role in future renovation decisions; it had not yet been formally presented to the city at the time of this account.

Stockton park-priority map showing selected park walksheds with one detail panel open
The open detail panel shows the youth-selected park with 91.3 percent overlap; the full five-park comparison is reported in the text. Open full-size visual

Get in touch: chris@christophernagel.com