How to design a map-based survey, question by question

A map-based survey works when every map question has a clear reason to be on a map, a follow-up that captures why, and a short enough path that people finish it on a phone. Everything else in this guide is detail on those three rules.

We have watched a lot of surveys get built in MapX. The ones that produce useful data share a structure, and it is not complicated. Here it is, question by question.

Start with the decision, not the map

Before you add a single question, write down the decision the results will inform. "Where should we put the new benches" is a decision. "Understand how residents feel about the courtyard" is not, and it will produce a survey nobody can act on.

The decision tells you which geography matters and which question type fits. Benches need points. A new bike route needs lines. A zone where noise is a problem needs polygons. If you cannot say what shape the answer should take, you probably do not need a map question for it yet.

Pick the right map question type

MapX has five map question types. Each captures a different kind of spatial answer.

Marker: where, exactly. The respondent drops one or several pins. Use it for places: a favourite spot, a dangerous crossing, where they would like a bench. Set a maximum number of pins per respondent. Three to five is usually enough. Unlimited pins produce a few enthusiasts who dominate the heatmap.

Line: along what. The respondent traces a path. Use it for routes people take, where they walk at night, which streets feel unsafe over a stretch rather than at a point. Lines are harder to draw on a phone than pins, so keep them to one or two per survey.

Polygon: which area. The respondent draws a shape. Use it for perceived neighbourhood boundaries, areas they avoid, zones they consider "their" part of the estate. Polygons are the most analytically rich and the most demanding to draw. Give clear instructions, and consider whether a marker would do.

Transportation: how do you get from A to B. The respondent marks a route between two places and how they travel it: on foot, by bike or by car. Use it for commute studies, school route surveys, access to services. It replaces the old pattern of asking for a start address, an end address and a travel mode in three separate text fields.

Media marker: show me. The respondent drops a pin and attaches a photo, a video or an audio clip, whichever the question asks for. Use it for condition reporting, documenting what a place looks like at the moment someone is bothered by it, or collecting evidence for a planning submission. Expect fewer responses than a plain marker, and more useful ones.

Every map answer needs a why

A pin tells you where. It never tells you why. The single most common mistake in map surveys is a beautiful heatmap that nobody can interpret because the survey never asked what the pin meant.

MapX lets you attach follow-up questions to marker, line, polygon and transportation questions. When someone drops a pin, they are asked one or two short questions about that pin before moving on. Keep these to a single choice or a short text. "Why did you mark this place?" with five options and an "other" field covers most cases. The follow-up appears right after the pin, while the respondent is still thinking about that specific place, which is when the answer is most accurate.

Resist asking the follow-up at the end of the survey as a general question. "What makes you feel unsafe in the area?" asked once, after ten pins, gives you an answer that belongs to no particular pin.

Use geofencing to keep the data clean

If your survey is about one neighbourhood, draw that neighbourhood as a geofence and respondents can only place their answers inside it. This does two things. It removes the handful of pins that end up in the sea or in another city because someone zoomed out, and it tells the respondent, visually, what the survey is about before they read a word.

Geofencing also makes it easy to run the same survey in several areas: duplicate it, give each copy its own boundary, and each one collects data for its own area only.

Set the map up for the respondent, not for you

You know the area. Your respondent might not know it as a map, even if they walk through it every day. A few defaults make a difference.

• Centre and zoom the map so the whole study area is visible on a phone screen without panning.
• Turn on 3D buildings if orientation is hard. People recognise their own block from the shape of the building faster than from a street name.
• Put a one-line instruction above the map that says what to do: "Tap the map to mark up to three places where you would like more light."
• Make the first map question easy. A single marker on a familiar place, before you ask for a polygon.

Mix in ordinary questions, but keep the ratio right

A map survey is still a survey. You will want age, how long someone has lived in the area, whether they rent or own, how often they use the space. MapX has twelve non-map question types for this, from single choice to matrix grids to sliders.

The ratio that works is roughly one map question for every two to three ordinary ones, and never two demanding map questions back to back. Pins are quick. Lines and polygons take effort. Put a couple of easy questions between them so the respondent recovers.

Keep it short enough to finish on a phone

Most responses to a map survey arrive from a phone, often outdoors, often during a doorstep conversation or while waiting for a bus. Plan for five to eight minutes. That is roughly ten to fifteen questions including two or three map questions.

If you need more than that, split it into two surveys. Respondents can also leave and carry on later on the same device, without an account, but a long survey with a 40 percent completion rate gives you worse data than a short one everyone finishes.

Test with 20 people before you launch

The first 20 responses are free on every MapX survey, which happens to be the right number for a pilot. Send it to colleagues and a few residents. Look at the map. Are pins landing where you expected? Are the follow-up options covering what people wrote in the "other" field? Is anyone giving up at the polygon?

Fix it, then launch. The responses you collect after that are the ones you will actually use, and if the survey grows past 20 you unlock it once and never think about it again.

A worked example

A housing company wants to decide where to add outdoor lighting on an estate of 600 flats. The survey:

1. How long have you lived here? (single choice)
2. Which building do you live in? (marker, one pin, geofenced to the estate)
3. Mark up to three places where you would like more light. (marker, max three) Follow-up per pin: Why here? Dark path / I feel unsafe / Hard to see the entrance / Other.
4. Trace the route you most often walk after dark. (line, one)
5. How safe do you feel on the estate after dark? (slider)
6. Anything else we should know? (long answer, optional)

Six questions, two map types, one follow-up, five minutes. The heatmap from question 3, filtered to the pins marked "I feel unsafe", read next to the routes from question 4, is a lighting plan.

That is the whole method. If you want to see it built, the features page has an interactive preview of every question type, and you can start free and build the example above in about ten minutes.