Here Is How We Optimized Conversion Rate From %4.8 to %29.7 With Advanced A/B Testing
I do believe in instincts as well, but I should always see the data before moving forward.
This is part of the story about how I got half a million users within 3 months.
A/B testing is a hype term in the business environment, but if you ask in person for an honest answer to the people who are working with the product and marketing most of them probably haven’t done any real A/B testing in their lives.
With A/B testing (split testing or bucket testing whatever you call) I basically compared two or more alternatives between each other. I tested different sign-up onboarding pages for conversion rates, and I tested some features inside the app to decide if it works better than the previous one, before implementing. I’m always so careful about metrics and data, sometimes it might be misleading if there are other factors affecting results. And I always try to keep the main conversion goal.
Conversion Rate Optimization With A/B Testing: How A/B Testing Helped Us To Optimize Conversion Rates On Onboarding Pages
%123 Increase With Omitting Unnecessary Questions
At WhatsAround, there was only one single onboarding page for all signups when I started. And the conversion rate was %4.8. I wanted to make some changes, but I needed to be sure that it would work better than the previous one. We have removed some verification questions from onboarding, taking them inside the app ‘profile completion’ process. We run A/B test to decide on which one is performing best. And the result was better than my expectation. We have managed to increase the conversion rates %123 to 10.7 according to the previous one. Omitting not necessary information from onboarding made it easier for people to sign up without distraction.
Adding Descriptive Images To OnBoarding Increased %108
Instead of trying to guess what would happen, I tested each optimization step before implementing. I’m not a full-force data-driven person, I do believe in instincts as well. I have seen times in my life even though the data showed me the opposite doing what my instinct said helped me a lot as well. But this is a rare situation and even I should see the data before I move on.
I saw that people didn’t understand very well what we are offering. So I wanted to make onboarding a little fancier and informative. We designed some pictures giving a sense from the app. And run the test with and without pictures. The result was amazing, the pictures were getting %108 more conversion. And now with the new result, we reached a %22.2 conversion rate. I wanted to be sure about each step before moving forward that the metrics had a positive impact on growth.
Alternative Signup Pages For Different Interest Groups Increased The Conversion Even Better To %29.7
I saw that people were coming from different sources with different interests and expectations from the app. So I decided to modify and create alternative onboardings for different interest groups. To be able to do that we designed different images with different messages to users. And we showed the original designs (control) to half of the traffic and customized designs (variation) to the other half. We measured the experience difference between control and variation. With the analytics result, there was a %34 difference between tests. The customized onboardings were performing a lot better, increasing the conversion rates to %29.7. And after seeing the positive effect on user behavior, we have decided to implement new customized versions.
In the end, we were reaching out to almost 6X more users with the same ad budget. This was not always the same, we saw some decline when we were widening the focus target group. It’s another story that I want to mention how we found the most responsive focus group. And we had another issue with some users registering but not downloading the app. But always going through the focus questions about any change, we were able to succeed in getting half a million users within 3 months after working so hard on optimization for almost 3 months.