DEV Community

Sam Ellis
Sam Ellis

Posted on

Five mistakes we made building a city-wide scavenger hunt app

We built Ping Hunt - a photo scavenger hunt that generates a hunt in whatever city you happen to be standing in. You get a series of spots, you walk to them, and you match the scene in front of you with the camera to unlock the next one.

It is a small idea, and it has been interesting to find out how many things have to go right for it to work in a city that is not your own.

1. We assumed people know the city

Our first version was a list of trivia questions about nearby landmarks. It was fine in the neighbourhood we tested in and useless anywhere else. The whole premise of a travel app is that you do not know where you are, so a quiz is asking the user to have done the reading in advance.

Swapping the questions for photos fixed most of it. Now the task is "find the blue door with the number on it", not "which architect designed this building". The user does the looking, and the app does the checking.

2. GPS is not a position

Our acceptance radius started at 50 metres because that sounded tight. In old city centres, where the streets curve and the buildings are close, 50 metres can put you on the wrong side of a river. We ended up with a radius that grows with how specific the target is - a fountain can be checked tightly, a particular stretch of waterfront cannot.

The other lesson is about the map pin itself. The coordinates that make sense for a scavenger hunt are not the ones that make sense for a mapping app. We pin to the nearest sensible point a person could stand, not the exact centre of the thing.

3. Battery is the real constraint

A hunt that needs the screen awake for two hours will not be finished on a bad day. We had to make the camera flow do as much work as possible before opening anything heavy, and we let people pause and resume without losing their place. If the app has to choose between being clever and being finishable, finishable wins.

4. People will try to cheat

Not maliciously - just because the task is at the top of a hill and they are at the bottom. Once we relaxed the radius for the waterfront, we noticed people photographing the right view from the wrong street.

We stopped fighting it. The interesting thing is that being a bit loose makes the game better: you end up wandering, and you find things you were not looking for. The point was never to enforce a location, it was to get someone to look up from their screen.

5. We shipped without telling anyone what it was

For a long time the app had no explanation of itself. People installed it, saw a map with pins, and closed it. The first version of the description that actually worked was not marketing copy - it was a plain description of what happens in the first two minutes.

That is still true of the whole product. If someone cannot tell within a minute or two what the thing does, no amount of polish will save it.

Where it stands

We have written up the design in more detail, with real screenshots and the honest caveats about imagery coverage, on our docs page: pinghunt.com/docs.

The short version of what we would tell anyone starting something similar: pick the mechanic that matches how people actually behave, treat the location check as a suggestion rather than a test, and spend the time you have on making the first two minutes obvious.

If you have built something location-based, I would be glad to hear what broke for you.

Top comments (0)