How Design Collective used map-based community input for a downtown park redesign

How they won the project

It started in the proposal. Before the project existed, Design Collective included a working Senf survey in its response to the solicitation: a live link decision makers could click and try, in place of a paragraph describing an engagement plan. The client tested the survey before the job was awarded.

“We used the survey as a differentiator and as bait to tease the client and give them a reason to remember us. Of course this meant a little upfront effort on our part.”

Brian Reetz · Principal, Design Collective

A downtown park, and a lot of ground to cover

Carroll Creek Park is the linear creek-side park that threads through Downtown Frederick, lined with restaurants, galleries, and the seasonal installations the city is known for. Design Collective led the public-space and hardscape scope of the refresh: shade structures, a reimagined planting palette, and repairs to keep the City’s investment looking at its best.

The survey did double duty. Design Collective built it to capture more than its own scope strictly required, so the client could draw on the same dataset across arts, safety and other workstreams. That decision shaped everything downstream, because it meant the engagement had to hold up for several audiences at once.

A question at the beginning classified the participants: residents of the City of Frederick living outside the custom Downtown Frederick boundary were the largest group at 39.1%, followed by 25.3% from elsewhere in Frederick County, 24.5% who own a business, work at or volunteer downtown, and 19.1% living inside the boundary itself. A further 9.2% picked none of the above. Because people could choose more than one, the figures add up to more than 100%.

Participant profile

InputOutput

How the engagement ran

Collecting the input was the straightforward half. Reading it was the part that scales badly: when open-text comments are each tied to a spot on a map, reading every one and inferring its location by hand stops being workable somewhere in the first few hundred.

“If we would have tried to do this survey with SurveyMonkey, 35 questions, I don’t know that we would have gotten this level of participation. I think people would have been bored ten questions in. With Senf the interactive mapping made this survey dynamic, interactive and memorable.”

The survey ran from February 1 to April 10, 2026, combining map-based, standard, and image-preference questions. Residents could drop pins, draw routes, and mark specific areas on the map. Multiple-choice, rating-scale, star-rating, and open-text questions captured the rest. Conditional logic used each response to guide participants to the follow-up questions most relevant to them. For shade structures and planting palettes, image-preference questions let people choose between visuals—more like sorting photographs than completing a conventional survey.

Route feedback

InputOutput

Pin feedback

InputOutput

For analysis, the team leaned on heat maps instead of hand-coding. Rather than reading each comment and guessing where it belonged, Brian could slide across the map, watch where comment density clustered, then zoom into those areas to read what people were saying.

“Just by moving across the result maps, you could immediately see where most comments clustered. Then you could zoom into those areas and read what people were saying—directly in the dashboard.”

What the data changed

The engagement produced more than a record of public sentiment. It gave Design Collective and its client a shared, place-based evidence base for deciding where the park needed intervention and which design directions had the strongest public support. Those findings moved from the results dashboard into stakeholder conversations, design strategy, and the next round of project decisions.

Navigation and safety

InputOutput

The heaviest concentration of “difficult to navigate” comments sat at the vehicular-pedestrian crossings where the park meets traffic — residents wrote about crossing times too short for anyone with mobility difficulty, garage routes with no marked crossing, curb cuts that force a detour, and cars that don’t stop. That gave Design Collective a defensible basis for focused conversations with City stakeholders about exactly where to intervene. The geography did the prioritizing, so those meetings started from evidence.

Design direction

The image-preference results fed straight into the work. Design Collective built planting design strategy boards around the plant material residents selected most often, and shaped its shade-structure concepts from the visual preferences, so the engagement showed up in the boards the team was already producing.

Image-choice question

PlantsShade

“You even had 140-some 75-plus-year-olds. In my mind, that speaks to the simplicity of the interface for people to easily participate.”

The usage data filled in the rest of the picture. Women made up 65.6% of participants against 29.5% men, with 1.4% non-binary and 3.4% choosing not to answer. The most common reason for visiting was the restaurants, shops and museums along the creek, followed by the seasonal installations and the festivals. The largest group visits two or more times a week.

Where map-based engagement earns its place

Carroll Creek is a clean example of the project type where location is the unit of feedback: parks, trails, plazas, streetscapes, anywhere the useful answer is a place. A text survey makes the analyst rebuild that geography by hand. A map-based survey with heat-map synthesis hands it over directly, and the heat map stays legible enough that a stretched-thin principal can read it between other deadlines.

The second payoff is the record. Because every comment carries a location and the synthesis is visual, the firm walks into client and city conversations with an account of what the community said and where, rather than a stack of quotes. For a public-space project answering to a design team, a client and municipal stakeholders at once, that record is a deliverable in its own right.