SAMHSA / CBHSQ
Making behavioral health data usable
Eight years as UX strategy lead on the nation's behavioral health data — from a library of PDFs to DataTools.SAMHSA.gov, a working interface for analysis and visualization.
- Role
- UX strategy lead
- When
- 2016–2024
- Scope
- SAMHSA.gov/data, DataTools.SAMHSA.gov, data collection pages, survey metadata
Who the data is for
SAMHSA's Center for Behavioral Health Statistics and Quality (CBHSQ) runs the national surveys that tell the country what's happening in mental health and substance use: who's affected, what treatment they're getting, where the gaps are. Researchers, journalists, state health departments, and policy staff all depend on it.
When I came on in 2016, that data lived mostly in PDFs. Hundreds of them. Reports, codebooks, methodology notes, and the detailed tables that the actual numbers sat in. If you knew exactly which report you needed and which year, you could find it. If you had a question, you were on your own.
Eight years, three phases
Readability first. The early work was the unglamorous kind: overhauling the front end of SAMHSA.gov/data so the reports were readable on any device, cutting redundant content, and restructuring navigation around how people actually looked for things — by survey, by topic, by year — instead of how the publications had been filed. I proved the changes with the tools available: analytics, heat mapping, and user testing with the researchers and policy staff who were the real audience.
Data collection pages. Each survey needed one place that held everything about it — methodology, metadata, every output — so a researcher could start from the survey and find the report, instead of the reverse. We consolidated that into designed data collection pages.
DataTools. The last phase was the one the first two had been building toward: stop handing people PDFs and give them an interface to the data itself. DataTools.SAMHSA.gov launched in 2024 as a modern interface for analysis and visualization — pick a survey, pick the variables, see the chart, get the table.
The judgment call
Somewhere in the middle of this, SAMHSA's data properties needed to merge with the main site, and the design question was whether the data experience could survive that. Mockups weren't going to settle it: the issue was how navigation, search, and the data tools would behave together, not how they'd look.
So I escalated to a working model. I built a coded wireframe of the merged experience so stakeholders could click through the actual paths a researcher would take. That ended the discussion faster than any review deck would have.
The point isn't that code is always the answer. Most of the eight years was ordinary design work — IA, content, testing, iteration. The senior part of the job was knowing which decision needed a working model and which needed a sketch.
What this is evidence of
Long-horizon product strategy on a complex, high-stakes public dataset. Accessibility and multi-device as baseline requirements, not afterthoughts. And an instinct for when to escalate a design question to a working prototype.