Recently Android has displayed some amazing growth as preferred OS for Smartphones. This has attracted lots of developers into making Android apps. Readily available tools, documentation & ease of App release have made this attraction even more powerful. Though excited with emergence of a new avenue for their work, often developers complain that they aren't making enough money with Android apps. To address this issue, I will list out some of best practices that can really help in generating better revenues:
Two version policies
It is a good practice to release two versions of the app, one paid and another free. Majority of android users are keen on installing free apps, but there is a sizeable chunk that prefers paid ones too. There are multiple reasons why having two versions make sense. In case of high engagement apps like eBook Reader or Games, some users find the ads annoying and want to switch to a cleaner (paid) version. In utility apps, for example sms scheduling, users want to try out the free app first before spending for paid app.
In app billing
Users can be asked for payment in exchange of some extra utility/service within the app. pretty useful for game applications, where user may pay to buy new levels or powers etc. Effective use of this feature for monetization would require some creativity.
Multiple Ad networks
Current major mobile ad networks are: adSense, inMobi & millennium media. One mistake in our early development days was that we only relied on one ad provider. But as we switched to multiple networks, our ad revenues saw significant rise. Tools like adWhirl are needed to enable support for multiple ads.
Number of tablet users is rapidly raising in Android, thanks to Icecream Sandwich upgrade & Amazon's Kindle Fire. Once an app becomes popular, releasing a tablet version of it on Android Marketplace & Amazon Appstore makes lot of sense. Another advantage is, usually tablet users are willing to pay higher prices for their apps.
I would conclude by saying that best approach for monetization is more of an art than science. Going through repeated cycles of applying a new approach and study consecutive data requires patience, but is rewarding in long term. Suggestions I gave above would act as a good starting point.