On the iPhone in 2008/9, before the iPad was out, the first version of the Facebook app was seen as one of the great examples how to build an iPhone app. It was developed by one person, Joe Hewitt, and parts of the app were open sourced as the Three20 project. When he left Facebook and stopped working on the iOS app, a new team took over which resulted in a complete rewrite for a universal app that also supported the iPad (November 2011). In spite of growing numbers, not a whole lot of users actually like the current iOS app. In the US iTunes store, the rating average is 2 stars, with out the 21,803 ratings, 11,839 1 star ratings (!). One star ratings are often a sign of frustration, and you can see that in the comments. For most of the apps an average of 2 stars is deadly, but we all use Facebook so yes we will all have to use its iOS app if you own an iPhone or iPad.

So what is wrong with the iOS app?

1. app is slow 
2. inconsistent information notification icons say there are new messages or responses, actual window does not show anything new.
3. app is slower than mobile web site while everybody is used to speedy apps, the Facebook mobile web site is faster than iOS app, and offers almost the same functionality.
4. tons of other bugs scrambled views, photo upload, text boxes disappear, no sharing.

What is the origin of these issues?

1. HTML and UIWebViews without Nitro JavaScript engine
I did some network sniffing (I like sniffing ) and found out that the data that the iOS app downloads from facebook.com is a mixture of REST (XML format, no JSON) and HTML. The HTML is used for your personal timeline, and profile and groups timelines. See for yourself, go to https://m.facebook.com/home.php and see that it is the initial view of the iOS app. Also from the screenshot above, you can see something went wrong while downloading the html and images/stylesheets/JavaScript. To display HTML in an app, a developer uses a Safari component called UIWebView. Very convenient, but also dangerous. The HTML downloaded is pretty big (15kb), and contains links to images, stylesheets and JavaScript. For a starter, caching of unchanged content cannot be controlled by the developer. The FB app downloads the whole timeline HTML every time, and it is up to the UIWebView to determine whether it needs to download images, stylesheets etc. again. Second, performance of UIWebViews is less than in mobile Safari. This has a lot to do with the absence of the new Nitro JavaScript engine in UIWebViews, apparently for security reasons. I ran some tests on my iPhone 4 with iOS 5.1.1, the JavaScript benchmark Sunspider running in Mobile Safari was 3 x as fast as running in a native app with a UIWebView. Also, to communicate from the UIWebView to the native app, a JavaScript bridge is needed. This is tricky stuff, slow and not really thread safe.

2. Different calls for similar information that is not in sync 
For notifications, messages and friend requests regular REST calls are done, returning XML data. First check is to see what number of new notifications are there (https://api.facebook.com/restserver.php), then the actual content is retrieved in a separate call (https://api-read.facebook.com/restserver.php). As far as I can tell, the Facebook service calls return inconsistent information. When you check too fast what new notifications are awaiting you, you don’t get the new information.

Why would Facebook use HTML technology inside a native iOS app?

1. HTML is easier for displaying fluid content. Objective-C really sucks when it comes to fluid display. An image with text around it, buttons with varying text labels are really hard to create yourself in Objective-C as you have to calculate dimensions and positions of all elements yourself. In particular for a timeline HTML will be much easier.
2. Creates code that can be shared across different platforms. iOS, Android, BlackBerry, Windows Phone are all different technologies and a developer’s nightmare. Sharing some content/functionality in the form of HTML makes sense.
3. HTML is much more in line with Facebook’s continuous deployment process. FB developers are responsible for their own QA, and part of that is to push code out to a limited set of servers, see results and then push it out to more and do this each day if not more often. With Apple taking as least a week of review, rolling back a code change is a nightmare.
4. They can get away with it. Yes Facebook is not a bank, there are no other iOS FB apps out there and we will still use the service as it has a virtual monopoly on social networking with 900 million users now. We just have to suck it up.
5. Feature phones is where growth is. A very high percentage of iPhone and Android users already have the Facebook app installed. The next frontier is feature phones, in particular in non-western parts of the world. These new users will first encounter Facebook on their mobile, and it will not be a shining iPhone.

I’d love to see a slick Facebook app like Path, I’m afraid it will not be soon . What do you think?

As this nice infographic from Apppromo.com shows, top earners spend an average of 14% on marketing and have an average marketing budget of $30,000 dollars. Yes, that means it is really hard to make money if you are a small team with no cash in the bank. But with so many apps out there and the iTunes and Play store not really facilitating discovery, you need to get your customers attention somewhere else.

So how can you do app marketing?

• Do PR: blog posts, youtube video’s, build an app web site or send out a press release to tech news sites and bloggers
• Buy ads from companies like Admob or Milennial Media
• Buy incentivized downloads from companies like Tapjoy, although these users will probably not become active
• Get promoted through platforms like Appsfire, Chartboost or Hooked

Edit 1: also very interesting thread on Hackernews about a story written on Ars Technical about this infographic.

]]> http://blog.mobtest.com/2012/05/for-success-you-need-a-great-app-and-marketing/feed/ 3 http://blog.mobtest.com/2012/04/rogue-app-syn-for-iphone-slips-through-apple-review-process/ http://blog.mobtest.com/2012/04/rogue-app-syn-for-iphone-slips-through-apple-review-process/#comments Tue, 10 Apr 2012 08:05:24 +0000 Dirk de Kok http://blog.mobtest.com/?p=99 Continue reading →]]> I got some annoying app requests on Facebook from this one app. They came via friends that I know click on too many apps and games, so I ignored them. Yesterday I saw a post about this very app called Syn on the Dutch iPhoneclub, where it was deemed a spam app. Worse, somehow it had made it to the number one position in the paid top 10 of all apps, in spite of an average rating of 1 star (!). My first reaction was that the company behind the app, Falkor, inc must have hired some shady firm to boost downloads to get that high in the ranking. But so many people complained about spam, I thought let’s dig deeper and do some research.

Syn asking for Facebook credentials

First, it’s a paid app (0.79 euro, $1), so reluctantly I had to pay. To make sure I would not hinder too many friends I created a new FB account that only myself friended. To see what traffic was sent down the wire, I used my favorite sniffer tool Charles, and set up my iPhone to use it. After installation the first screen already raised my suspicion: it asked for my Facebook credentials. With oAuth being the standard now, this is a clear violation of Facebook Platform policies: no collection of username and/or password. With Charles I could see what was done: Syn logged you in directly onto Facebook, which surprises me was possible. After comparing HTTP traffic with that of the Facebook iOS app itself, it turned out that Syn used precisely the same endpoint to login (https://api.facebook.com/restserver.php), and FB’s own API key (haha).

After logging in with my FB credentials, it downloaded my only friend (myself), and offered to sync the info to my address book on my iPhone. Right after I confirmed, I got an app request on Facebook. Clicking on the app request did only one thing: after a redirect via falkor.com the iTunes app was opened, deep linking to the Syn app.

So yes a rogue app. Facebook credentials are used in an illegal way, friends are spammed with a link to download the app. I suspect too many FB users are gullible and just click on an app request from a friend. I was looking if my address book was uploaded to the Falkor servers, that didn’t happen. So for sure Facebook should do everything it can from stopping this app by this shady company (web site lists another product called Spyder, a MySpace friend added bot). Apple also should have stopped this app, although strictly speaking the only violation I could see is tricking users to spam friends via Facebook.

The main thing Apple did wrong here is the top 1 position in the iTunes App Store, in spite of the average rating of 1 star. Apple is just asleep at the wheel, anybody can spot something is wrong. Fake or accidental downloads by tricked users gave it this nr 1 ranking, and this shows how broken the algoritm is. As I said before, we expect better from our beloved Apple.

Update 1 (4/18/2012):

I just tested the new version of Syn, version 1.0.2. Still it uses your Facebook credentials directly, which makes it still violate FB’s policies. I don’t understand why Facebook hasn’t blocked this app yet. On a positive note, it now asks for permission to spam your friends. The message is more like begging for money (see to the right). Still I would not recommend using this app.

Update 2 (4/25/2012):
I got notice from Facebook that they reviewed my report and subsequently have shut down the Facebook app.

Update 3 (4/26/2012):
And Apple has removed Syn from iTunes app store.

So what’s new and cool in Windows 8?

Metro style UI

Already present in Windows Phone 7 and XBox, the Metro UI design is now also available in the new version of Windows. So your apps that you build for Windows 8 computers will have a similar look as Windows Phone devices. On Windows Phone the Metro look has been well received as rivaling the user experience on iOS and for sure surpassing Android’s. Now the first major change in user interface is available for all devices, computers, tablets and phones. Having a consistent user experience across all devices we use nowadays is a major plus. Microsoft is taking this further than Apple, where Mac OS X is only slowly moving into the direction of iOS.

HTML/Javascript for native apps

When building apps for Windows using .NET technology, you already had the option of C#, C++ or VB as language of choice, in combination with Xaml. Those are languages for real software engineers. With Windows 8, HTML/Javascript has been added so now web developers and designers can create native apps. As testified by a software engineer of Flixster, easy peasy. This all is built on top of the new WinRT runtime, which ensures same performance across different languages. With Visual Studio being the best IDE for software development, better than XCode and Eclipse, life of software developers is getting easier.

So that Nokia Lumia 800 running Windows Phone

Although the event was about the desktop event, Nokia and Microsoft are currently pushing hard to launch Nokia devices running Windows Phone. Nokia knows hardware, Microsoft software to this seems to be a great match. The Lumia 800 is not released in the USA, the slightly better 900 is. I won a 800 in the raffle so was able to lay my hands on one.

Being an iPhone user with a few Android phones on my desk, I had something to compare to. The Nokia Lumia hardware is slick, it’s a hip phone that stands out and feels great in your hand. The touch screen is really responsive. What interested me most was Windows Phone. The Metro UI is just stunning, and clearly designed for touch screens. Big squares you can’t miss with your big fingers. The performance of the interface is really snappy, clearly as much attention as for the iOS but much more than for Android has been paid to a responsive interface. The integration with Hotmail (ahum), MS Skydrive and Facebook & Twitter in the OS is tight.

The apps for Windows phone will determine if users will flock to the platform or not. Currently there are 80,000 apps out there. Not so much as the 600,000 for iOS or Android, but that is logical. While this might not be so good for users, it means there is an opportunity for developers. Making money on iOS directly from users has become really hard and requires 100,000′s of dollars in marketing, besides building a great product. Android is becoming very fragmented and no user pays for apps. If Microsoft makes the user experience of actually paying for an app a great one, revenue might be as good as in the early days for iOS. That means Windows Phone might prove to be a mobile platform that you can make good money on.

I think this is a step away of what they have been doing until now, which was all focused on the phase before launching. Also it is a crowded space: there are already mobile analytics companies like Flurry and app download companies like Distimo and Mobapp.

Meanwhile, we at Mobtest.com keep on focusing on our mission: to build a great marketplace where mobile developers can find testers. Developers need feedback on their apps before they push them out in the app stores. They need to get high ratings to do their marketing in the app stores and get interested users to actually download their app.

We are looking for beta testers of our own service, so sign up at mobtest.com to get an invite. First batch is 50 testers, so sign up quickly!

To be more precise, I think that app discovery in the app store sucks, and this hurts both users and developers.

Right now, the only ways users find apps are through New, What’s Hot or the top 25 of a category. With almost  600,000 apps,  that is basically an Office Depot with aisles that are 50 miles long. You look at the top 25, but that’s it.  App search is just really bad. Genius? Anybody ever used Genius more than once? Genius is a joke like Ping is. Apparently, Apple is just not good at social.

So how does an app get into What’s Hot, New or a top 25? An app gets into New, by being new (and a little bit lucky), but obviously only stays there a short while. An app gets into What’s Hot by being very lucky, like having extremely excellent design and an original app. So the important trick now it to get into the Top 25 of a category. This should be based on downloads, revenue and ratings. Unfortunately, too may bad apps, copycat apps or down right fraudulent apps get in the top 25 by playing this system. They do this by just cloning popular apps like Temple Run or using download bots to increase download numbers. And even good and legitimate apps have a hard time getting in a top 25. The problem is, as top 25′s are used so much, getting in there means staying there. It’s a self fulfilling prophecy that way. I see in my Dutch iTunes account that WhatsApp is still number 1 in Social Networks. Good for them, but this app is already like 2 years old.

As a result, app browsing is seeing the same apps all the time, and nobody uses search. So users don’t see new newsworthy apps and stop looking for new apps. For developers it is also bad. That $4 billion payout today might seem a lot, but not when you divide it by 138,000 something publishers/developer you get almost $ 29,000 over the last 3 years. The gold rush with stories about making tens of thousands of dollars with a simple game are way way behind us. Even worse, this top 25 failure means a complete winner-take-all distribution. My guess: top 5 % of app publishers do 70-90% of this revenue, and the rest of the publishers either have another core business (Facebook, LinkedIn, your bank) or are just doing bad. This means that the long tail and with that the whole ecosystem is doing bad, which will drive new developers away from the iTunes app store to different stores like Android Market or Windows Phone marketplace.

So now Apple has acquired Chomp, an app search engine. Congratulations both! Some advise to the new owners of the app store:

1. improve recommandations and search. If Google can do it for the web and Amazon for shopping, so can you.
2. use other metrics than downloads and ratings for top 25s. Like number of times opened, home screen, amount of time used.
3. app refunds would be great (although free apps + inapp purchase also works), and you can do this as you control the whole ecosystem
4. create favorite app lists, and sharing of them with (Facebook?) friends
5. allow communication between users and publishers
6. allow publishers to respond limitless to reviews in the app store
7. give better statistics, including view, download and referral statistics. Keep statistics indefinite
8. profile users, and give aggregated and anonymised profile information to developers.
9. allow more dynamic pricing models, with business rules that automate this
10. offer A/B testing of pricing, icons, screen shots, etcetera

Let’s make the App Store a great place for everybody again!

My take aways:

1. Apple values great visual design over human interface guideline compliance
2. Assume nobody reads anything
3. Make choices for users (no 50 configuration settings please!)
4. Limit the number of clicks
5. Don’t force a workflow on mobile
6. Be consistent across different platforms
7. New trends #1: 5 tab navigation, where the center tab answers the question “What do I do with this app?”
8. New trends #2: hidden navigation is replacing tab bars
9. New trends #3: users expect “Pull to refresh”
10. New trends #4: iPad and iPhone are converging

Yes you should let others test.

The first weeks are vital in the life cycle of your app. If your users do not understand the app or it crashes on their phone, you will get a lot of one star ratings and sink deep in the ocean of 450,000 iOS apps.

OK, so how do I test. How do I get my beta to my users? iPhone beta distribution used to be a complete pain in the neck. You had to get the UDID of the iPhones of your testers, nicely hidden away. Then create an Ad Hoc Profile, send them the app + profile and pray your testers  would understand what to do with iTunes to install it. My experience: 20 % of installs didnt’ work, in particular updates of the beta app and then on Windows. That was a complete pain. Nowadays you can use Over-the-air (OTA). The distribution platform TestFlightApp helps you nicely with this.

OK, my testers managed to install the app, now what? Well, depending on what you want to test, give instructions. Ask the right questions. Can you run the app? Can you upload a photo? Do you instantly understand how to use it? Does it run on your specific version of the OS? Just sending the app will not get you that valuable results. Go in with a hypothesis, ask yourself and your testers the right questions and validate.

How do you do testing?

Testing is even harder, arguably also more important. Native apps crash completely on an error, not just one page. You have less screen real estate. Users have less time to use your app. And there are so many tools to choose from. So let me ask a few questions: what to test? And when are tools appropriate, and when do you need users?

The list of different kinds of testing we have here at Mobtest is long but a few first ones:

1. technical test: does the app run without crashing?
2. functional test: does the app do what it is supposed to do?
3. usability test: do users understand how to operate the app?
4. compatibility test: does the app run on all hardware required?

A first statement would be, that when implemented right software tools should be cheaper and faster. But what do tools detect, and what do users find? Tools are only as good as the people that operate them. Tools can detect imperfections in your code, execute tests that you have written yourself. Can they detect things you hadn’t though about?

Probably tools cannot detect all. In particular usability testing is by definition user testing. So it seems using both tools and users to test your app is the best way to go. What do you think?

Mobtest’s mission is to connect mobile app developers with users that understand apps and can test them

Why? Well for developers, we think that now it is too hard to do app testing, so the quality of the apps out there is not that great. How many times do you have an iPhone app crashing on you? Or received a lot of one star ratings because of a huge bug you or Apple didn’t find before publishing in the App Store? That should be a thing of the past. With over 450,000 apps in the iTunes app store and 250,000 in the Android market it is hard to stand out. Your app needs to be top notch to do so.

Besides that the market is growing and growing, over the next 4 years about 1,000 percent. We as app developers have to deal with that, crunch out more apps while same time improve the quality of our work. Not a simple task.

For testers, we want to give them early acces to the coolest newest apps out there, and give them a chance to help them improve these apps. And become good at testing them and adding value to the developers.

Engineers focus on development, testers will provide them with input what part of their work needs improvement. Find out why your app always crashes. Learn if your users find it easy to use your app. No more crappy apps! No more one star ratings!

What do you think?