Use visual experiments to test changes without App Store updates
This article walks you through how to create a Visual Experiment for your mobile app in the Taplytics Visual Editor. For how to create Code-Based Experiments, see our guide here.
If you're still reading this, that means you're ready to dive in and create a visual experiment. Awesome, let's go! Start by clicking the "Create New Experiment" button.
You'll be prompted to add a name for your experiment and choose the OS for your app.
Now that your experiment is set up, you should connect your device so you can see real-time changes as you configure your experiment. To do this, click the "Connect a Device" button.
There are two ways to pair your device with Taplytics. You can pair your device through development build pairing or you can set up App Linking. App Linking allows you to pair both development and production builds and it links your device via text, email, or ping.
While your app is still in development, you can pair your device by connecting it to your computer or by running your app on a simulator. Available devices show up in the pair list. All you need to do is click the "Pair" button next to the desired device.
If you have trouble getting your device to show up in the pair list, we recommend you use App Linking instead.
When you pair your device, an alert will pop up that asks you to connect the device to your Taplytics account
Tap "Yes" and the "Connected Devices" counter should show your newly connected device. A red box will also surround your app to tell you which experiment and variation you're viewing on your device.
If you haven't already, we recommend that you take a few minutes and set up App Linking so you can work on visual experiments in production mode as well. Plus it's an awesome way to pair your device! To do this, check out our App Linking guide so you can pair your device via email, text, or ping.
If you have already set up App Linking, click your pairing method of choice and follow the on-screen prompts to connect your device. When your device successfully connects, your app will have a red border around it and the visual editor will reflect the number of connected devices.
Now that your device is connected, you can set up your visual experiment.
Before adding elements to your experiment you should give your first variation a name. Here, we used "Great Variation," but you might want to use something more descriptive.
An experiment is all about creating variations of individual elements that are part of your app. By combining and changing a number of different elements, you can create extensive and powerful experiments. To start, click the "Add Element to Variation" button.
On the next page you will see a button that says "Edit the Look and Feel". Before you click, navigate to the part of your app that contains the element on which you want to base your experiment. Once you arrive, feel free to click the button.
When you click, the red box around your app will turn green.
This means that your device is ready for you to pick an element to test. Now, tap the element that you want to change. Yep, with your finger, right on your phone. Your screen will turn green while our SDK determines which element you selected and sends the information back to Taplytics.
Instantly, the SDK pushes the elements that it thinks you tapped back to the website where you can select the element you want to edit.
Select the element that you want to test and it will be added to your variation.
Feel free to add as many elements as you want. Repeat this process until you have added all of the elements you would like to test.
Once you have added the elements that you want to edit, you can select them one at a time to change their properties. The properties that can be edited are on the right of the element list.
Taplytics only shows you the properties that you can edit for that element. As you change the properties, the changes show up right on your device, instantly. How cool is that?
Why stop at one variation? Add as many variations to your experiment as you'd like. You can give each variation a name like we showed you in the previous steps.
By now, you should have your variations set with your elements edited the way you like them. If you're ready to move on, click the "Next: Configure Goals" button in the top-right corner or click the "Visual Goals" link in the left-hand navigation menu.
What's an A/B test without goals? Well, not a very good one. That's why the next step is setting your goals. Goals are set in the same way that you selected your other elements.
First, choose the type of goal that you want to set. You can choose button clicks or time on view. Then tap on the button or navigate to the view that you want to track. It's that simple.
You can add as many goals as you want. Click the "Add Goal" button. Feel free to track whatever makes sense, but we generally suggest you should only have one or two primary goals to determine your winner and a few secondary goals to add context. Check out all the goal options here.
Now that you have the perfect experiment with the right goals, you just need to determine how you want your A/B test distributed. First, select your filters. These filters make sure that no matter what distribution percentages you set, your experiment is only seen by a certain segment of your users. For example, you may only want users with devices newer than the iPhone 5 to see the experiment. If that's the case, set your filter to target that segment. After setting your filter, you can choose any distribution breakdown you'd like. We generally recommend an even distribution to start, unless you plan to do a roll-out. Check out the available distribution filters here.
The only thing that's left to do is crack a beer, order some snacks, run your experiment, and enjoy the results. To push your experiment live, click "Let's Run This Experiment".
Now that you know enough to get started with visual experiments, the best thing you can do is explore the platform. Try adding more elements and see what changing the different properties does. You can test out as many changes as you'd like internally, as long as you don't run the experiment.