Modeling Facebook Application Growth with Appaholic

So, you’ve written an awesome Facebook app and you think it’s going to take off. You get five users, submit it to the directory, and because your application works well and has no bugs it gets listed within a few hour. You have an idea of how fast you think it will grow, but you really need to know how fast it actually is growing.

Using Appaholic, you can create accurate models of your application’s growth. If you’re growing too slowly, you need to know so you can fix the problems that are keeping users away. If you’re growing too fast, you need to know so you can move the application to a host that can accommodate it. You’d also like to be able to set goals and know if you’re on track. Will I have 1.2 million users within two weeks?

I’m going to use Booze Mail as my example app and Excel to plot the data. First, go to Appaholic and navigate to your application’s page by entering its name or application ID in the left-hand input box. You will be taken to a page that displays various statistics about Booze Mail’s growth.

We’re going to be using the daily data because Excel doesn’t understand hourly time-based graphs. So, download the CSV (Daily) file from Appaholic and open it with Excel. Select the data you want to graph and click the “Chart Wizard” button. You don’t need to select the lines which have no user numbers — they correspond to the days before your app was listed in the directory.

Note: Make sure you only select date and number columns and not the column titles. Otherwise Excel won’t graph the data properly. In the Chart Wizard select the Line graph and click “next.” Customize your graph as you see fit and then click “finish.”

Now you have a nice graph in your spreadsheet. Right-click on one of the data points on the graph and select “Add Trendline…” Here you pick the curve you think best fits your graph. You will probably have to experiment to get the graph that fits best.

For Booze Mail I’m going to choose a third degree polynomial model. I don’t think is really accurate over the long run — I’d guess that most applications follow a logistic growth curve. But fitting logistic curves to data is beyond the scope of this article, so the third degree polynomial will have to do.

Finally, to get the trendline to extend in to the future click the “Options” tab. You should see a “Forecast” section. For our graphs one period corresponds to one day. So if I want to predict my application’s growth ten days into the future I can enter “10” in the “Forward” box. Since we want to predict Booze Mail’s growth by the end of July and today is the 15th I’ll extend the trendline 16 days into the future.

Click “OK” to graph the trendline. You might want to change the size and color of the trendline to make it stand out. Let’s see how Booze Mail is doing.

So, according to the graph, Booze Mail should hit one million users by July 25th. Anyone want to bet?


Recommended articles