NorthQuad for 120Water. Product designer, led the 0-to-1 design.

Sample Manager

Keeps water utilities on their sampling schedule. I led the 0-to-1 design, then helped rebuild it around real usage.

Role: Product designer, led 0-to-1
Platform: Desktop web app (SaaS)
Timeline: Sept 2024 to 2026
Tools: Figma, Figma Make, Claude Code

15organizations live, in 18 states

11email types, UX co-led

4beta-spec assumptions that usage overturned

Most violations are missed samples, not bad water

29,703 US water systems missed a test or report in 2023, about 1 in 5. 6,045 broke a health standard.

Source: EPA, 2023 National Public Water Systems Compliance Report (opens in a new tab)

One rule becomes hundreds of tasks

80 coliform samples a month in summer, 10 otherwise: 400 tasks a year from one rule.

Requirement

What, where, how often.

Sampling period

One window of that rule.

Task

One sample, one owner, seven stages.

A clickable prototype of the rebuild

My own rebuild, with invented data. The rules underneath are real. Best on desktop.

LiveFull prototype, 10 screensOpen in new tab

How I built it. Figma and Figma Make for direction, Claude Code for the working prototype. The rules come from the specs and the data is invented.

Four assumptions in the beta spec did not survive real usage

I read 13 specs as data. Every piece below is live.

The four assumptions, what usage showed, and what I changed
Decision	The spec assumed	Usage showed	What I changed
Time	Task counts follow the calendar month.	The date switcher gave confusing and inaccurate task counts.	Count from today. The month picker only drives the charts.
Done	Done means the state received the result.	Finished tasks stayed flagged as overdue.	Admins pick the stage that counts as done.
Scale	A utility samples at usually around a dozen sites.	Sampling points were a hard-to-navigate comma-separated list.	One tabbed page per facility, with a searchable table.
Lab results	Labs send results by API or file upload.	Manual import by staff was the top gap across 9 sales demos.	Staff upload the PDF themselves, and every row must pass validation.

Task counts start from today

Deadlines don't care what month it is.

LiveSettings · Task thresholdsFull prototype

Try: widen a window, then Save. The brief below follows.

LiveDashboard · Daily brief and month pickerFull prototype

Try: switch the month. The brief doesn't move.

Admins choose which stage counts as done

Finished work kept showing as overdue.

LiveSettings · Completion stageFull prototype

Try: pick Collected, then Save. The counts below drop.

LiveTasks · CountsFull prototype

Each facility gets its own page

Some facilities have hundreds of sampling points.

LiveFacility · North Utility GroupFull prototype

Try: open the Tasks and Sampling points tabs.

Staff upload lab PDFs themselves

Lab results arrive as PDFs, in every format.

“We probably would not be able to move forward without this step.”
UX research, on the PDF compare panel

LiveSampling Data · Where import startsFull prototype

Try: open a flagged result. Import isn't built yet.

Eleven email types, and rules for when not to send them

I co-led the UX. The risk was an inbox so stale people stop opening any. The prototype builds six.

Admins choose which emails can send, and each person chooses which they get

LiveSettings · Account and personal notificationsFull prototype

Try: switch off “Log Your Sample Collection” on the left.

Seven days out means different things for monthly and annual samples

LiveSettings · Reminder timingFull prototype

Try: change a reminder window, then Save.

Every email re-checks the task before it sends

No changes, no email.

Success metrics from the specs

These are targets, not results. The facilities brief had none, so I'd track searches and round trips.

Results page views per 30 days, baseline to target

40%

Accounts using self-serve lab upload, 90 days after launch

Lab imports needing staff, 60 days after full launch

99%

Extraction accuracy on known formats, 95% on new ones

Live with 15 organizations in 18 states, still being reshaped

First facility fixes shipped in March and April 2026. The rest is scheduled.

What I took from it

A spec is a hypothesis and usage is the answer key. Before designing, I ask what data we have. If there's none, I go get it.