Collier Blueprint

01 The question

Residents want to hear about projects before the vote.

Picture 600 new apartment buildings proposed in your small town. They change your taxes, your traffic and the wildlife near your house. When would you want to first hear about it?

When residents want to hear

When most residents hear today

02 The problem

Most meetings sit empty. Then 200 residents arrive too late.

Township meetings are where residents give feedback before officials vote. On Feb. 23, 2026, 200+ residents showed up, but the plans had been in motion for years.

03 Collier

Collier, a town of 9,000, is growing in two directions.

A suburb 6 miles from Pittsburgh, known for quiet streets, parks and the Panhandle Trail. Its population grew 25% since 2010.

04 Research

We heard from Collier 224 times over seven months.

Residents, township staff, department heads and commissioners, across 4 stages of research and 13 product iterations.

• 155

  usability testing sessions

• 30

  qualitative interviews

• 39

  survey responses

• 6

  in-person visits to Collier

05 What we heard

No single reliable source told residents what was coming.

Residents pieced news together from word of mouth, signs and several social media accounts. By the time they showed up, decisions were already made.

06 Direction

Residents felt they and the township were on separate teams.

We started out trying to get more residents to engage. Research showed they already care. They need to know when to engage.

07 What we dropped

Residents rejected points and badges and chose stage-by-stage tools.

We put 10 rough prototypes in front of 49 users, from a points leaderboard to a printed newsletter.

08 Blueprint

We built Blueprint, one site for residents and township staff.

Blueprint drafts every township decision early and out in the open. Residents and staff see the same projects through two views.

09 Results: For residents

Residents follow each project stage by stage.

See where it stands

Each project as a timeline of stages, from planning to vote to construction.

2.2 → 4.4

comprehension rating out of 5, policy PDF vs. timeline view

11 / 11

residents rated the timeline higher

Speak up while it counts

A forum on each project. A live viewer count reminds commenters to be mindful, and AI helps make feedback specific and actionable.

12 / 12

residents in A/B testing preferred a community forum

“I would love to see the comments, but in my experience, it sometimes spirals out of control.”

10 Results: For township

Staff post projects with AI and turn feedback into grants.

Upload projects with AI-assistance

AI pulls the stages out of existing documents and shows how confident it is. Staff review before anything goes live.

“I like [the AI-assistance]. It would provide a huge value to the staff and public.”

Generate grant applications and reports

Reports pull from every project, comment and poll, so staff can cite resident input in grants, board briefs and annual reports.

“This is a source of unprecedented value for grant writing and operations.”

11 Tradeoffs

We narrowed three touchpoints to one and made AI optional.

Every feature had to help residents and stay something township staff could run.

12 Late testing

With one month left, we turned outside of Collier to supplement testing.

After we narrowed to the policy tracker, we had no time or access for new resident sessions. We kept the resident needs from earlier research and tested three other ways.

1. 01

   Analogous testing

   People outside Collier with the same needs ran through the prototype.

   “I rate it a 5 out of 5. The overview and timeline makes total sense, very easy to follow.”

2. 02

   More client calls

   We held extra calls with township staff to check each change against the new priorities.

3. 03

   Heuristic evaluation

   We checked basic tasks against usability heuristics, then finalized the UI and information structure.

13 Impact

• Comprehension

  15/15

  usability testers accurately summarized a project in their own words

• Readability

  4.55/5

  average readability rating

• Accessibility

  91.29%

  success rate on usability tests from 12 users, ages 55 to 84

• Staff workload

  9 of 9

  government staff foresee minimal administrative work

14 What I learned

I learned to commit early and let testing correct me.

Seven months and a small team left no time to study every option. Three moments that taught me how I want to work.

1. 01 Prove the direction with prototypes, not debate

   Our brief was a PRD titled Insightful Governance: use AI to strengthen data, transparency and community understanding. It could have meant anything. We put 10 rough prototypes in front of 49 users. Residents rejected points and badges and kept reaching for stage-based tools, so I pitched the policy tracker and built every prototype to prove it. I learned that a rough prototype in front of residents settles a debate faster than another week of discussion.

   “I like being able to see what steps the projects are on.”

   2.2 → 4.4

   comprehension out of 5, policy PDF vs. timeline view

2. 02 Own the build so quality survives the handoff

   I owned both design and development of the final prototype, built the design system in code, and maintained the GitHub repo we handed to Collier. Feedback from each test went straight into working code, across 13 iterations. I learned that a designer who can ship the fix protects quality without waiting on a handoff.

   “I found it to be quite easy to maneuver and navigate who isn’t that great with computers. It made a lot of logical sense along the way.”

   91.29%

   success rate on usability tests from 12 users, ages 55 to 84

3. 03 Ship a day-one version and a roadmap to the rest

   The full vision includes AI uploads, sentiment analysis and grant reports. Collier needed somewhere to start. The prototype I built works without AI from day one: staff post projects, and residents follow and comment. We paired it with a step-by-step roadmap, budget and timeline that add the AI layers as optional investments. That first step is what let Collier start building. I learned that a small version that works today moves a client further than a complete vision they cannot fund yet.

   1. Day one

      Blueprint without AI

      Project timelines, polls and comments, run by staff by hand.

      Open prototype↗collier-demo.vercel.app

   2. Next

      Optional AI layers

      AI uploads with confidence levels, suggested responses, sentiment, grant reports.

   3. Vision

      Full township adoption

      A mature system residents and staff use for every project.

   9 of 9

   government staff foresee minimal administrative work

15 My role

I found the direction, designed the system, and shipped it.

Co-lead designer and developer, with Laila Dancy (Lead UXR), Bosi Li(co-designer developer) and Jason Park(PM).

1. 01

   Direction

   The final policy-tracker path was my idea from ideation. I built all the prototypes to prove it.

2. 02

   Research

   Split interviews with residents and township staff, ran user testing, and explored ideas through rapid prototypes.

3. 03

   Build

   Coded components that made Collier’s web brand sleeker, and drafted the design and dev handoff with our PM.