About
Android, Linux, FLOSS etc.
Code
My code
Subscribe
Subscribe to a syndicated RSS feed of my blog.
Tue, 27 Mar 2012
Some success...
So I have had a little bit of success. In December of 2011, I was on
one ad network, Admob, and made $6.66 for the whole month from them. I
am now on three ad networks (Admob/Adsense, Millennial Media, Inmobi),
and I made $7.62 from them in the past two days, more than the whole
month of December. I would like to increase that in the future, but for
now, $100 a month coming in is great. Of course, I want to roll as much
of that as I can back into the business.
The breakthrough happened in late January. I have written Android apps from scratch like Bouncer and Love Poems, and I ported an open source Java library to Android with Panacea Database. Looking at a full-fledged open source Android project, FBReaderJ, I noticed some modifications I could make to it to improve it, and that would be for an audience without much overlap with the existing FBReaderJ audience. FBReaderJ is GPL licensed, which worries some people, but myself less. Anyhow, I released my version of the app, "Free books to download & read" on January 24th. By the last day of January, 2425 installs a day were happening, by February 5th, 11000 installs a day were happening. Daily installs ranged from over 8000 to over 11000 a day until February 20th. The install rate is still over 2000 a day. As is normal, the active installs in percent has been going down over time, but it is still over 35%. It currently has over 119600 active device installs. There is currently one ad - right before someone goes to a book - it has been requested from 13000 to 23000 times a day over the course of the past two weeks.
Having had success with modifying an open source project, I doubled down, and on February 12th I released a modified version of OI File Manager, another open source Android project. I chose it because it was open source, because I had thought of doing a file manager for a while, and because it had a wide appeal - it is not a niche product like Panacea Database or Bouncer, many people can find it useful. I wanted to release another app with wide appeal to ride the wave of Book Reader. And it did so, it has over 4239 active device installs, which for my five apps is second to only Book Reader. And has been achieved in six weeks, while I have been working on apps like Bouncer for ten months.
I do have my eye on one more Android open source project, but I have turned back to doing an original project. It uses Andengine, but is actually an app, not a game. It is original as far as I know, nothing else on Android does it in the manner mine will, which is much better than the handful of existing ones that are related to this app. I have to see how much work I am going to do on it before releasing it. It is more toward a niche product than a general one, but it is not a small niche. Anyhow, much work to be done on it, although I already have a decent prototype for one implementation of it.
Book Reader was making over $20 a day when the downloads were first flying. Also, I had an ad on the page seen when the app was opened for the first time, which I now do not have - although I may put that back. Anyhow, I rolled $100 of that Admob money into ads. While I was running my ads, Admob dropped their minimum ad bid to $0.01 a bid. So I dropped my bid to that. The money went mainly to buying ads in Brazil for the File Manager. Ads seem to boost downloads from the target market, even when they're not running, don't know all the variables which cause that although I can guess some of them. Anyhow, I know have over 1000 active users from Brazil for File Manager that I probably would not have had any how. Were they worth ten cents a head? Well, the initial buys were overpriced before Admob's price drop. Also, it was something of a test. Also, I want to roll my profits back into the business and couldn't think of a better thing to spend it on. Even with that $100 spent, I'm still getting over $350 from February Admob profits for Book Reader. Those kind of dollars came from the initial pop, I'm now more at the $100 a month level, as I said before. Although if I had more ads in the Book Reader app, I could probably make more. Although I want to avoid having ads over the actual book, as that is annoying.
In terms of running Admob ads - you can choose the devices to target, the SDK version, the country (and sometimes more specific location), whether to target mobile, wifi or both, gender and age group. Transfers of $50 or over from money I was owed to running ads gets you a small bonus of free ads. Each campaign is $10 a day minimum. Minimum bid nowadays is 1 cent a click. You can see conversion rates for app installation for app download ads.
The annoying part for Admob is the approval process. First you have to get approved to be able to transfer money from your balance to ad campaign budget. Then campaigns have to be approved. After I was approved for balance to budget transfers, I transferred $50 and submitted a campaign. A week later it still sat unapproved, so I sent them an e-mail, then it was approved. Contrast this to Millennial Media, who approved a campaign for me recently within hours. You'd think Admob would be more responsive to me wanting to give them my money.
So on that Millennial Media campaign - I noticed a few days ago that the paltry sum I made in February from Millennial Media had been put into my balance. The sum was paltry because I was not even signed up with Millennial in the beginning of February. Anyhow, I took the dollar or two and put it into a campaign in Norway for File Manager. It was approved within hours, which was the good part. One downside was the minimum 5 cent bid - 5 times what Admob does. Also the targetting is not as precise for kinds of device and such. You can target to country though, which I did. I wonder if "Android" goes out to Kindles, Nooks and the like, I hope not as it would be wasted money. Anyhow, my $1.20 daily budget was filled and I got 24 clicks. I'll probably do a bigger one next month for MM when my March money clears, maybe for different countries. Another nice thing about MM is I'm not stuck with $10 a day campaigns! But unlike Admob, MM keeps the money you earn for two months plus instead of one month plus, so I may as well roll the money back into ads.
I signed up for Inmobi as well, but you have to talk to them or something to get approved to transfer money from balance to budget. It's not worth it at this point.
I also might do Adsense for mobile ads. I'll have to see. I should get the $350+ by the middle of next month, so I have some ideas for the money. I might spend some money for a contractor to do some work on Book Reader - which I plan on using myself and sending back to FBReaderJ as well.
I had used Admob as my sole ad network prior to January. One reason I chose them is they were known to be reliable about sending checks - in fact, they already sent me one last year. Also, they have a low check sending threshold - if you make $20 in a month, which I'm now easily doing. They also send the money within one month plus. If I made money on ads on January 1st, or January 30th, that money would get sent to me on March 1st and would arrive, usually around March 15th in Paypal. For Millennial Media and Inmobi, the amount of time is longer.
But anyhow I wanted other ad networks. For the sake of redundancy for one - if there was some problem with Admob, I'd still have two other sources of income. Also, perhaps I'd get some better deals, or extra functionality, which I have gotten. Also, I like the idea of keeping some competition open for the ad networks - it benefits developers to have a few competing ad networks out there. I read a report which said the top Android developers usually have as the top four packages includes - Adwhirl, Admob, Inmobi and Millennial Media. That dovetailed with what I had heard already so I went with Inmobi and Millennial Media.
Inmobi seems to do everything manually, and even over the phone. My app approval seemed to be in limbo until an e-mail back and forth. Then I had a phone conversation, where the rep said they wanted me to push up the number of requests I was getting as they thought it was too low. This conversation happened a month ago. I said my Book Reader got a lot of hits so submitted that. It was pending, then they said they wanted more info on my address etc., so I put that in and it is still pending. Not that I mind much, I submitted the app at their urging, to some extent. As I said before, to be able to transfer earnings balance to an ad budget requires manual intervention as well. Well, Admob and Millennial Media are more responsive without hassle, so I'll deal with them more in terms of buying and selling ads for the time being. Inmobi is still the primary target for File Manager ads though, with MM and then Admob as fallback, and 80% of traffic is directed to Inmobi via Adwhirl right off the bat. Aside from responsiveness, I'd need to make $1.67 a day from Inmobi to get a monthly check from them, and right now that is more like 28 cents a day, so I haven't even hit that minimum yet with them (or Millennial, which is about $1.03 a day).
I suppose eCPM, RPM, CTR, etc. are important in differentiating ad networks, but one overriding thing is fill rates. Admob and Adsense integration has been increasing as time goes on, other than it taking a day for clicks, CTR, eCPM and revenue to update (but not impressions or fill rate), the two are very integrated. And for normal apps, the fill rate for this is usually over 98%, if not 99%. As opposed to this, Inmobi has had a 21-54% fill rate for me over the past two weeks. Millennial, which is getting a fraction of the direct File Manager traffic Inmobi gets, but which does get its run off, has had a 77-86% fill rate for the past 9 days. The major slackoff from them is for countries like Brazil and Poland, they don't have the presence Google can afford there yet. But for the US, France, Germany, Japan etc., their fill rates have been on par with Admob's. With Adwhirl, lower fill rates are not as big a deal, but it takes seconds for Adwhirl to miss an Inmobi ad, and the Millennial ad, and then maybe even an Admob ad before putting up an Admob "Adwhirl" ad, and by that time the Activity with the ad may have been clicked off.
Sun, 01 Jan 2012
Happy New Year
My New Year's started out the right way, one of my apps, Panacea
Database crossed the 5000 download mark. It's kept to a 40%+
active/net install base as well, hopefully with some of the updates
coming down the pike it will maintain, or even improve, that percentage.
Thu, 17 Nov 2011
Looked at Admob today, I finally pushed past $25 in payments from my Android applications. $25 was the one-time fee I paid to get on Android Market. So I've made $25.16 from my three mobile apps so far, and am now 16 cents in the black. Admob sends you money when you hit $20 for a month, so in December I should be getting a check for October and before. In addition to the Admob money, Samsung was also nice enough to give me a free $500 value 10.1 inch tablet to write tablet-sized apps on. And with my latest update of Bouncer out this morning, all three of my apps now handle "extra-large" displays, as Android calls them.
I was contemplating that I'm now in the black this morning, and felt good about it. My thought in terms of my business of putting out Android apps revolves around having no recurring capital costs, and if at all possible, no capital costs at all. Particularly in terms of some web page that an app must contact that I'd have to pay $10 a month or so for. Right now I just code the app, push it to Android Market, and collect the ad money. Aside from the slow wear on my keyboard, mouse, screen etc., the only expense is my time.
I wrote a framework for a spreadsheet, and did a number of spreadsheet features for it. Then I worked on getting pre-2007 Excel files onto it, which I did. Then I worked on getting Excel 2007 and 2010 (.xlsx) files onto it - and got stuck. There are two possible paths to fixing this, an easier one of I can get things down to less than 65,536 methods, and a harder one if I can't. I took a shot at the easier path, and that just might not be possible, as I got rid of a lot of methods. I may be able to pare down a few more. If not I'll have to go on the harder route. Anyhow, I put the code up on Github.
A month ago, I finished rewriting the layout of Panacea Database for all major (and minor) device sizes and screen densities. Then I added a feature to remember the last file opened. I did some testing and QA on the last file feature, but perhaps not enough, as it seems there have been some crashes since then which probably pertain to that. Which I am looking into. People seem to want column sorting, which I can work on implementing. I might throw in some SQLite stuff, depending on how easy it would be.
So all of my apps have decent layouts for all major (and most minor) devices, which I am happy about. So now I am on to my new apps, as well as fixing bugs and implementing new features in Panacea Database.
Sun, 09 Oct 2011
I released another Android application - Love Poems. It took off initially - by the fourth day there were 442 downloads, with 280 of them active installs. But then that slope of adoption leveled off, it fell in the Market rankings etc. Not sure what hurt it - I did an update allowing users to increase or decrease the text size, while someone gave the app a two rating. It then sunk in the Market rankings and downloads leveled off. A few days later I released an update with a few more poems, and also adjusted the text sizes a little. I will do updates in the future, in terms of both poems and display tweaking.
Android is continuing to gain market share. Here is the browser usage seen from various mobile operating systems, according to the web logs of the Internet's 7th most trafficked site, Wikipedia:
As the chart shows, the iPhone and iPad are doing well, as are Android smartphones. Windows Phone 7 is moribund - it only is 0.04% of traffic. There is more Android Honeycomb traffic on Wikipedia (0.05%) then Windows Phone. I guess we'll see how they do with Windows 8 and Mango which is supposed to launch in 2012, but they are way behind Apple and Google. The modern tablet market is newer than the smartphone market, so maybe they'll have a shot at competing there. I downloaded Windows 8 preview and developer kit and had a look at it. Their Store is free for developers, although applications are approved first.
I'm currently developing a fourth app. Won't reveal all details until it's released, but it uses Fragments and the ActionBar. Android's compatibility package does backward compatibility for Fragments but not ActionBar, so I am using Jake Wharton's ActionBar Sherlock for backward compatibility in ActionBar usage. I have that all implemented already actually. I haven't done all the happy stuff you can do with tablets and Fragments yet, we'll see about that, it's not an essential element to the project, but with all the usage of ActionBar and Fragments, redesigning it to do that will be easier. This new app may use SQLlite as well, so I may be looking into SQLlite.
I was invited to the Android Developer Lab in New York on August 24th. It was good - I met some interesting people, and they pointed us in the direction of where Android is going, which helps me point my development in that direction.
I've been doing a bit of work on Panacea Database's layout. I moved a lot of stuff into XML. I'm using scale-independent pixels and density-independent pixels as much as possible, as well as adjusting the size of buttons by layout weight and that sort of thing.
One thing I've been doing - I change how many rows I display when fetching rows from the database, and the scale-indepedent pixel text size of the display, depending on what screen size I have, what orientation I am in, and to some extent, how many dpi are on the display. The way I've been doing this is putting a "gone" TextView in the XML, and from my code, reading the number of rows to display from that. Not sure if its best practices, but it works - if I find a better way I'll do that.
Sat, 09 Jul 2011
According to Alexa.com, Wikipedia is currently the 7th most trafficked web site. They are also one of the few large web sites to allow everyone glimpses of their web log analysis. I mention this in a previous blog post. In December 2010, Android devices made up .078% of Wikipedia's web traffic. At the end of May 2011 (June numbers are not done yet) that was up to 1.16%. So Android traffic on Wikipedia increased about 48% in six months.
Actually, the six month increase of about 48% from December to May was more-or-less matched by the one month increase from November 2010 to December 2010, which was a 47% increase in traffic. I guess a lot of people got Androids in their Christmas stocking, or next to their Hanukkah dreidels...
So anyhow, I released my second Android application, Panacea Database, on June 11th. I definitely followed the Release Early, Release Often philosophy for this one - I got the idea for it on June 7th, and by June 11th it was published.
I guess another party writing a nice Java library, which someone else posted a bug report, which was subsequently fixed, seven months before, that fixed all the Android bugs, helps. Thanks Miha Pirnat, wherever you are!
So what it does is iterates table rows and does searches for Microsoft Access style files on Android. Or Microsoft Access 2000 to 2007. With a lot of Access 2010 working. I actually just sent a patch in to the library people to fix a bug. Or implement a kludge to get around the bug anyhow - until I'm interested in dealing with Attachment data types, they'll have to write a fix.
So both my apps have passed through the 500 download point. Bouncer has a 41% active/total install ratio, Panacea Database has a 57% install ratio. Why is that? Well to quote a critic on the Android Market, Silas, "Move to SD card!!" The app has a lot of PNG's and JPG's and is 3.8MB. Maybe I will move some of that to the SD card, who knows? It's an issue I have to figure out how to deal with.
My Admob revenue for the last week is 79 cents, $1.52 for the week before that, and $1.28 for the week before that. My first goal is $100 a month in revenues. Whether that be by ads, sales or whatever, it does not matter.
Initially I thought of just tossing out apps left and right and seeing what stuck. But you put an app out and you have to maintain it. And I'm just one person. For now anyhow. I don't want lots of one-star ratings for my apps on Android Market. The lowest I've gotten were two three-star ratings for Panacea Database. One wanted me to fix the bug where next-lines in a text data type would make a button disappear. I've partially patched that already, and have a full patch for that (hopefully) that I will release, oops, I mean publish, soon.
Mon, 20 Jun 2011
A Guide for the Android Developer Guide
I wrote A Guide
for the Android Developer Guide which attempts to translate Googlese
into English
Tue, 31 May 2011
Bouncer, my first Android application
So, I have published my first Android app (the concept for which someone else described to me). What have I learned about Android development and such since then?
My first (unpublished) Android app was heavy on ListView. It was a tree of ListView's really - the top ListView went into sub-trees of ListView's, until a leaf/node on the bottom was reached, which might be something else. I filled out the onCreate method, and an onListItemClick method.
The first screen of my new app was initally going to be a GridView. I then gave up on that. I then created two activities which could go back and forth to one another via clicks (listened to with OnClickListener) via Intents. Then I had them pass information to one another in the Bundles. So now I can pass messages to my sub-trees via the Bundles, and they can be separate Activities.
Having dropped the Gridview, I tried out the TableLayout, which I eventually went with. So now I had my grid-like table of letters on the first screen, able to pass which letter was pressed via a bundle in the Intent to another Activity. I used Buttons for these letters.
I then wanted there to be a tab on the front screen, with the table of buttons in the primary tab, but with people able to tab over to the "About" tab. So I made the first activity a TabActivity, and opened the Activity with the table with an Intent.
I then wanted to change the color of the buttons, but found out it was not all that simple, and learned about 9-patch drawables and the like. So I created my own buttons, which needed their corner rounding to be specified and the like.
Google suggests you put an End User License Agreement in the application. There is a standard class to do this, so I put it on the application.
Ultimately, I want my app to cover all 50 of the US states, as well as the District of Columbia (Washington D.C.) Currently, it covers 46 of the 50. I had the current ID for 46 of the states, at this point in development I started putting up older licenses that may still be valid.
Most of this time I was designing for a high density, normal size screen in a vertical position. About 17% of people using Android's use medium density however. Also, some people flip from vertical to horizontal mode, I even encourage this flipping in the application when the full image is about to come on the screen. So I did some work on making it at least function with medium density setups, and for high density setups when viewed horizontally. I get the display metrics, and then call different layouts depending on what the metrics are.
When to release is always an open question. "Release early, release often", agile development and so forth is the popular credo, and I agree with it for most applications. On the other hand, you can't release too early, especially since Android Market has a rating system and so forth. But at this point, I felt I had enough, and the last four holdout states it didn't look like I would get anything from them in the next few days, so I decided 46 was enough to be useful, that layout looked decent for most phones, and was at least usable for almost all phones. So I released.
One thing I did not do when releasing was release the initial version with ads. Why? Because Admob wants to know where it is on Google Market to give you an ad code, and I had nothing up there yet. I later realized I had misunderstood due to my unfamiliarilty with all of this, I could have put an ad in the initial version. Within a few hours of publishing version 1.0.0, I released 1.0.1, which contained Admob ads.
It's been 28 hours since I released the initial version, and 15 hours since I released the version with ads. Thusfar I have had 78 downloads of the app from Android Market, and have had 55 ad impressions served.
In subsequent versions I plan to improve the application. I will work to get the four missing states, and the District of Columbia. I will put more information about identification. I might put a bubble up announcing updates, but I wouldn't want it to be too annoying. I also have some kludgey stuff in the layout files which hopefully I can clean up, as I learn the Android API better these things can be more smooth.
Fri, 22 Apr 2011
I have been looking over Android's API and have been writing an Android application with Eclipse.
Android use has started to take off in the past months. I have looked at various metrics, one I like is from the Internet's 8th most trafficked sites, Wikipedia. It shows the growth of Android use over the past six months:
The graph y-axis is the percentage of all browsers coming in - mobile, desktop and whatnot. X-axis is the time period of usage - the past six months. The OS versions are listed in the key, although "Mobile other" is a catch-all.
In October 2010, 0.47% of all hits to Wikipedia came from Android phones. In March 2011, 0.98% of all hits to Wikipedia came from Android phones. So that has more than doubled within the past six months.
Wed, 30 Dec 2009
Android install
I am installing the Android SDK.
For Linux, it suggests the Eclipse IDE, at least version 3.4. I use
Eclipse 3.5, "Galileo". So I am installing the plug-in and get (in an
almost impossible to read output box)
Cannot complete the install because one or more required items could not be found. Software being installed: Android Development Tools 0.9.5.v200911191123-20404 (com.android.ide.eclipse.adt.feature.group 0.9.5.v200911191123-20404) Missing requirement: Android Development Tools 0.9.5.v200911191123-20404 (com.android.ide.eclipse.adt.feature.group 0.9.5.v200911191123-20404) requires 'org.eclipse.wst.xml.ui 0.0.0' but it could not be found
What this translates to is Android is dependent on another plug-in. So I go to install the webtools/wst xml plug-in, but it needs an EMF plugin. Then it needs a GEF plugin. Finally it will accept the webtools/wst plugin. Then the Android plugin can be installed. This sounds easy, but between Eclipse's junky and non-intuitive GUI and Android's documentation not mentioning their plugin had dependencies, it was not.