This afternoon we saw the second Weekly Wednesday Lunchtime Lecture, hosted by SourceTag. The speaker, Raul Portales, talked about his experiences in marketing his games on the Android Marketplace. Although I’m an iOS guy, the challenges we face on the App Store are even greater on the other side.
Raul presented several graphs showing his total installed base over time as irrefutable evidence to back his assertions, and I took away several lessons that I think are relevant to all platforms, and to product engineering and marketing in general.
When we talk about the business model of your app, we’re talking about whether you’re selling it outright, or giving it away for free and trying to monetize with advertising.
Also common are the hybrid “freemium” models. You can have two versions: a lite version and a paid version. Or you can use the old shareware model, letting people download and try the app for free, then purchase it in-app to remove adds, unlock features, and remove limitations.
Raul proposed a new model, which can be used alone, or in conjunction to other models, which I call the “freebie” model. You have the main, full-featured app available to purchase. You then have a number of simplified spin-off apps you give away for free, which contain ads for the main app.
There are several advantages to this model, and Raul’s numbers prove that it works. His freebie app was featured by Google, which brought in some 80,000 new users. Although the absolute numbers are an order of magnitude lower, the growth curve for the paid app was an exact match to that of the freebie app.
It was no accident the freebie got featured. Raul shrewdly made a stripped down seasonal variant, in this case for the Easter holidays. Not only are customers looking for seasonal things to try, stores are looking for seasonal things to feature.
This is genius. Seasonal apps are a niche market. The number of entries is smaller, meaning the chances of getting picked are much higher. Getting featured is the fastest way to pick up new users, and making the app free makes downloading it a no-brainer.
Not only do a number of users follow the cross promotion to the main app, the quality of the users is particularly high. Every app has a certain “bozo penalty” of users who give it undeservedly low ratings for asinine reasons. The less your app costs, the more bozos end up trying it, with free apps suffering the worst by far.
The freebie model uses the free app to absorb the bozo penalty, and only sends over the users who actually liked the app. This means that not only are your sales higher, your reviews are higher than they would have been with similar growth from other sources, like being featured directly.
I’ve always preferred the shareware model, but this bozo shield effect is a compelling counterargument in favor of the lite model. If my app skinned easily, I think I would eschew them both and just use a pure freebie model.
Raul’s other observations included the superior performance of niche apps over apps for a general audience. Even though the market is smaller, penetration rates are much higher, meaning you capture a larger percentage of possible users. Ratings are also higher.
Why is that? There are a number of possible explanations, but the one I prefer is based on the fact that real users differ from the models of them we use in creating our products. Assuming you know what you’re doing, the closer the actual user is to the model, the more they will like the app.
By extension, the farther the user is from the model, the more they will dislike the app. Users tend to like products more as they get used to them, because mastering the product moves them closer to the model user.
When you make an app for a general audience, you are casting a wide net, and are going to get a lot more people who are so far from the model as to fall into this bozo category. Because they are focused, niche products tend to attract people already closer to the model, leading to better overall satisfaction.
Another problem Raul mentioned is that people who dislike an app are more motivated to rate it than people who actually like it. This causes all reviews to trend downward. This effect is exaggerated by the fact people overuse the lowest rating. Anyone who ever got an F knows what one low score can do to an average.
Andy Ihnatko once made this point about people leaving one-star reviews of his book on Amazon for relatively minor complaints. I am paraphrasing here, but the idea is that if the book contained nothing but insults for the reader, racist jokes, and fond remembrances of Hitler, you would also give it one star.
I think the solution to both these problems is for the store or OS to prompt users after one month, or upon deleting the app, to simply give the app a thumbs up or thumbs down.
This is how we do our speaker ratings after each Weekly Wednesday Lunchtime Lecture. No star ratings. No surveys. We simply ask people to answer a yes-or-no question: would you like to see this speaker again? We get around 75% voter participation, which is pretty damned good.
Any speaker who receives at least an 80% approval rating will be put on the Appsterdam speaker list, which will be online so journalists looking for interviewees, conferences looking for presenters, and schools looking for lecturers know who to talk to.