App promotion and usage statistics

Two weeks after launch my Utrecht Walking Tour iPhone app had sold about five copies. Since then I’m trying to figure out what’s causing this, in a systematic way. These are the four possibilities I can come up with:

1. irrelevant product
2. something crucial missing in the MVP (Minimum Viable Product)
3. bad marketing
4. too expensive

(1) It wouldn’t be the first time someone built something that nobody needs. Unfortunately when it comes to leisure time, what “need” means is not very clear. When there’s an oil spil, the oil company needs it to stop. Nobody needs an iPhone app to guide them through a city in that sense. In fact, as a tourist, they may not even need to be in that city in the first place. Of course restaurants, tourist shops and guidebook publishers make plenty of money for a reason, but this lack of need is one of the reasons I’m still looking for startup ideas outside the consumer “optional activities” space (travel, food, non mandatory education, games, generic literature, social networks, etc).
(2) I tried to build a minimum viable product so I could start learning as quickly as possible. I can travel for half a year with just hand luggage and I’m not afraid to leave out features in a product. In fact perhaps I was so unafraid  that left out too much. Or I might have picked an initial target audience that’s vanishingly small. The walking tour in the first version might only be convenient for a few dozen people a week.
(3) Even if people really need an app like this and there’s plenty of people that can be persuaded to walk this specific tour, they’re not going to find out about it automatically. Perhaps in due time the mayor will enthusiastically announce it on local television. But even in my most optimistic scenario I don’t expect word of my app to spread like wildfire on the first version. My initial marketing consisted of a short announcement and a retweet by The Railway Museum; the tour starts with a train ride to that museum. It would have been nice if that was enough to get traction of course.
(4) Given the number of tourists visiting the city and the percentage of the population with iPhones, I estimate that I need to charge at least €3.99 for the app and perhaps even more for future versions, to make it worth the time I and others would need to build a complete version. The best paper based guidebook costs €25 and they received government subsidy. That means it’s a tiny market, but given that they sold more than 15.000 of them it’s still interesting if you can keep costs low. Rumor has it though, that people aren’t comfortable paying more than €0.79 for apps and they tend to compare apps with apps, not travel products with travel products.
So how do I figure out what’s going on? I’m running a series of experiments. The first thing I did was to make the app free during the Easter Weekend. This led to the following encouraring chart:

461 downloads in one weekend! I made it all the way to #3 when searching the App Store for “Utrecht”, which is higher than the official app by the Utrecht Tourist Board (they announced their app when I was already working on mine). I made it to #15 in the travel category. This is going to change everything, man!
Now I know from experience that if you make a paid app free it gets featured on all sorts of price-drop websites and gets downloaded by a lot of people who just download everything that’s free.
So let’s look at what people actually do with the app once they download it. In comes Mixpanel. Almost half the people who downloaded the application did not launch it, at least not during that weekend.

The app shows several sights in Utrecht together with a short description and links to more information. Almost 30% of those who launched the app followed one or more of those links. That suggests that those people were indeed interested in the information itself. How about actually going on a walk?

These numbers are a little disappointing, especially because nobody has walked the complete tour so far. However the weather was pretty bad most of the weekend, so I can imagine that people preferred to read about Utrecht on a warm couch at home and didn’t really feel the need to go outside.
So now that I made it to the top of the charts and almost a hundred people actively browsed through the information, what happened next? I put the price back at its original level and the number of downloads dropped back to zero. Of course that’s not unexpected on weekdays and with bad weather.
Last weekend I tried another experiment: can I use Google Ads to draw people to my website and have them buy the app? With my previous app this seemed prohibitively expensive with most relevant keywords costing over €1 and Google ranking my landing page very low, but in this case things look better. I created a slightly nicer landing page with some teaser content on it, set a maximum price of €0.50 per click, picked a bunch of keywords that Google suggested and let it run for the weekend.
It’s difficult to track conversion with the App Store: Google tracks how many people click on your ad and with some effort you can track how many people click on the Available in the App Store link, but that’s where the trail ends and Apple’s black box begins. Here’s some of the keywords that resulted in the most clicks:


There were three sales that weekend and no guarantee that they came from Google. This may sound bleak, but it’s not. Consider that downloads only happened on Friday and Saturday. On Friday the conversion from clicks to sales might have been as high as 20%. Because Google’s lawyers don’t allow me to use the word “iPhone” in the ad text, I assume I can double my conversion in the long run by building an Android version. So if I tweaked the ads, landing page a bit more and had a more complete product to offer, it’s not unthinkable that I could buy a customer for between €1 and €2 resulting in a profit.
The next weekend with descent weather I plan to run a shorter but 5 to 10 times more intense campaign with slightly tweaked keywords. If that results in 10 to 30 sales I can be pretty sure adwords were responsible for them. If nothing happens, then it’s more likely a problem with the product itself or the pricing.
So I still don’t know which of my four hypotheses are true, but I’m working on it.