A/B testing for optimizing applications: benefits, tools and tips

5 min reading
UX / 02 October 2015
A/B testing for optimizing applications: benefits, tools and tips
A/B testing for optimizing applications: benefits, tools and tips

BBVA API Market

Over 1.5 million. This is the current number of apps both at Apple Store and Google Play. Faced with so much competition, every effort needs to be made to optimize the product. In that race to attract the user's attention, A/B tests are one of the key tools for developers and designers. With these tests, putting an app on the market is much more than flipping a coin. 

There are currently many tools for conducting A/B tests in the process of optimizing apps to make them more competitive at the stores. Generally speaking, this work is known as App Store Optimization (ASO): the aim is to attract organic traffic and increase the volume of downloads by always appearing well positioned in the search results.

Application ASO or SEO depends on several elements:

– Keywords: this is a crucial factor in web positioning and within the search engine of app stores. Much attention needs to be paid to keywords in the app name, the product's description and also its location.

– Visual element: essentially icon and screen shots in the showcase.

– External factors: number of downloads, user rating and the comments on the app's features. Moreover, tracking the operation of keywords and changing them based on the results can improve ASO positioning.

Many of the apps with the largest number of downloads around the world depend to a great extent on proper optimization of all these elements. On this web page you can check the free and paid apps with the largest number of downloads at the Apple and Google stores and also for MacOS X.

Tools for A/B testing

1. Optimizely

Optimizely is possibly the best known tool for A/B testing, multivariable testing and multipage testing available on the market for both iOS apps and Android operating systems. It is a platform widely used by app developers for several reasons:

– It is really flexible and offers many possibilities.

– It is not a tool that requires great effort for its implementation. The IT team or the customer downloads the SDK, integrates it into the app and places the fragment of code in jQuery with each test of the components to be analyzed.

– These tests can be implemented through the Visual Editor, which is much more convenient than the more complex option, changing the code directly in the code editor. This code editor can be accessed directly from the Visual Editor itself.  

The Visual Editor enables the developer to move all the design and usability elements in an app. The developer can modify texts, change background images, color styles, fonts and size, copy and paste elements, edit dynamic elements like dropdown menus… It is a fast editing system for changing the graphical interface of apps and conducting tests. 

Developers can also change the styles using the Visual Editor, as can be seen in the first image, but also via the code editor, by entering new elements in the fragment of code in JQuery. With this second option the developer can make special changes that are not included in the default options of the Visual Editor. 

Changes directly in the code, two examples: 

– Specials CSS:

$("body").append('<style> YOUR_CSS_HERE</style>');

– Background image:

$("body").css({'background-image':'url("http://exampleurl.com/image.jpg")'});

2. Apptimize

Apptimize is another widely used application for testing iOS and Android apps. It has an extremely easy-to-use visual editor. On the web page itself there are some explanatory tutorials to help get started with the platform. 

Steps to follow:

– Install the SDK on the mobile app.

– Open a new experiment.

– Select visual editor and add a new variant. In this option the developer can include new buttons and change text, background images, font styles, colors, the size…

– Once finished, the user simply clicks the ‘Done’ button. If last minute editing is needed, click ‘Edit’.

– The user then goes on to the events page, where the key performance indicators are set; they will tell the developer which of the variants proposed in the app's design works best. 

Apptimize not only has a visual editor, but the test's variants can also be set directly from the code editor. This would be an example of real code for testing whether the Facebook or Twitter button would work best for logging into an app. The syntax can be coded from the editor itself using the Swift option

– (void)viewDidLoad
{
    [super viewDidLoad];

    [Apptimize runTest;@"Facebook v. Twitter" withBaseline:^{
        //Baseline variant "baseline"
        _loginWithFacebookButton.hidden    = NO;
        _loginWithTwitterButton.hidden     = YES;
    } andVariations:@{@"variation1": ^{
        // Variant "Twitter"
        _loginWithFacebookButton.hidden    = YES;
        _loginWithTwitterButton.hidden     = NO;
    }}];

3. Amazon A/B Testing Service

This tool is still in the beta version, but it is backed by a large company like Amazon. Like Optimizely and Apptimize, Amazon A/B Testing provides a testing system using variants aimed at enhancing the conversion metrics, optimizing the design and usability… for both iOS and Android apps. 

Some of its features include:

– Tool for controlling up to five versions at once.

– The service is totally free of charge, at any usage level.

– Amazon claims that the SDK can be integrated into the app and the first experiment configured in just one hour.

– A/B tests are launched thanks to the Amazon Insights API.

– The Amazon Insights SDK is connected to Android or Xcode (iOS) to make it easier to purge errors.

– It features a segmentation option for defining the type of user who will receive the experiment. A given segment can be defined using a filter or the SDK, through the UserProfile API. This option can also be used for creating subsets in certain groups, for example, to display test variants only in men.

There are many other A/B test tools, equally powerful and interesting: Taplytics, Splitforce, Mixpanel, Leanplum and Arise.

Other tips for optimizing apps

As a result of monitoring app performance in the stores, some guidelines for success have been identified for organic positioning:

– Repeated keywords have no effect in searches.

– Excessive use of keywords is counterproductive.

– Users look for short keywords using several combinations

– If figures are used, it is best to choose numbers rather than words. 4, not four.

– Users use plurals. Therefore, plurals should be used.

– Prepositions or articles are dispensable.

– Use the maximum number of characters permitted in keywords. In iOS, 100 characters. This field does not exist in Google Play.

– In the case of the app name, the first 25 characters are key for searches.

– It is not advisable to use special characters or duplicate keywords in the app name and the rating categories. 

Follow us on @BBVAAPIMarket

It may interest you