This is a guest post from Karolina Jasvinaitė from Soundest.
When we start a business, all we have is a pile of assumptions, opinions, and various hypotheses. What I most like about A/B testing is the transition from “I THINK this can work” to “I KNOW this will work”.
What is A/B testing?
Testing your properties with the A/B method (also known as split testing) involves finding out which marketing solutions work best for your business. By creating different variations of your store/ landing page/ email campaign you can test a layout design and the effectiveness of a copy and determine which one performs better.
By monitoring A and B customer behavior on your versions A and B of call-to-action (CTA) buttons, product placement, or whatever else your goals are, you can ensure that particular changes will bring positive results. In this manner, continually testing and optimizing your properties will increase the numbers of sales, sign-ups, downloads, comments, etc.
If the idea of the objects to test has already born in your head, think about the related variables. For example, if you want to test the web form for gathering subscribers, you should include the timing when it appears, the colors, and a copy encouraging subscription. If it is a landing page, you should also care about background colors, text, and the placement of CTA buttons. It is very important that before testing has started, you have a concrete idea of what you want to clarify during the experiment.
What should you do if you like the current outfit of your store and you have doubts about whether different color buttons or different placements could change anything? The answer is – keep version A as it is now and then change only version B. It is not obligatory to set up both versions A and B from zero.
How many things can I test at a single time?
Using the A/B testing method, it is better to only analyze one thing at a time. If not, you will get confused whereas the button placement or the different headline has determined better performance. The other kind – multi-variate testing – is intended for testing more than one object at once. This test is an advanced level for testing.
What increasement can I expect while testing?
At the beginning of this post, I promised an increase of conversion rate as a result of A/B testing. So what should you expect and how do you know if you are moving in the right direction?
In the various case studies on the internet, you can find impressive examples with hundreds of percent growth. This is true. But till this happens, there is much still to do. In the beginning, your conversion rate should grow by about 7-10%. If your conversion rate is about 20%, the growth can be a little bit lower than this. This is totally normal. It depends on the case, but usually companies achieve hundreds of percent growth after several testing campaigns and changes. Be cautious.
What should I test?
You can test whatever business property you have. Most common A/B tests are provided for landing pages, product pages, ad banners, and email campaigns. If your business is a mobile app, then test your app. Now let’s talk in more detail regarding what parts of a particular marketing instruments should be tested.
Testing the website / landing page
Start with identifying in which conversion step you lose the greatest amount of customers. If the bounce rate is huge and very few people get engaged with your homepage – start from there. If customers find their favorites on your store, put them in the cart but abandon it afterwards – you should consider re-organizing and testing the check-out page first.
Consider testing this:
- The layout and colors of the page background and fonts.
- The headlines: are they catchy enough?
- The call-to-action button: color, size, text. Try even those colors from the personal “no way” category – you will be surprised by the results.
- Web forms and coupons gathering the contacts.
- The readability of your page body text: font, size, color, and content volume. Keep in mind that no one likes to read long texts next to the product. It should be short marketing style copy, unless it’s a washing machine or some other expensive item with a list of features.
- The page navigation: how many times do customers have to click to achieve a conversion? – The product page: product placement, image size, description text.
- The place and number of related products. Do they fit the screen without scrolling being needed?
Testing the email campaigns
When we talk about emails, the most important metrics are opens and clicks. High or very low numbers might be caused by different reasons.
Things to test:
- Sender’s name and email address. Try to send an email from a person, not a company. From an email personalization approach, this tactic makes customers feel that there is a real person behind the message. You may also consider inserting a sender’s photo and email signature into the body of the email.
- Subject line. Try different wording.
- Delivery time. Discover when your customers are most likely to open your email. There is no such thing as the best day or best hour for all newsletters. Take into account the difference in time zones, different holiday’s dates, etc.
- Layout of the email: colors, graphic elements, font size.
- CTA button placement.
- Responsive design, if this has not been used before. Keep in mind that 53% of all emails are opened in mobile devices, so a mobile-friendly/ responsive design is crucial for your email marketing campaign. Using responsive email templates that fit perfectly the recipients’ screen size will ensure that your recipients will see exactly what you want them to see. It prevents them from deleting emails without reading them.
Testing the mobile app
According to the Entrepreneur blog, by 2017, it is expected that over 268 billion downloads will generate $77 billion worth of revenue. This industry is growing tremendously, so you might be considering creating the mobile app for your business and it should be competitive. If so, you should know that in the mobile application the things to test stay pretty much the same: you can test everything from button colors to game physics.
Tools for A/B testing
A number of tools have been created for A/B testing. They are different depending on their scope and price. Here are a few recommendations to get you started.
- Google Content Experiments is a free tool from the search god. This tool lacks some advanced features, so it is good for beginners.
- Optimizely. This tool easily integrates with other web analytic tools like SiteCatalyst, KISSmetrics, and Google Analytics. It has a free limited plan and advanced paid features.
- Wingify Visual Website Optimizer is a testing and optimization suite. Along with an A/B testing tool, it includes tools for multi-variate testing, behavioral targeting, heat maps, and usability testing. The first 30 days are free for a trial period.
- Unbounce is a tool for building and testing high-performing landing pages. The tool provides real-time statistics on visits, conversions, and variant conversion data. Unbounce is starts from $49/month.
- Splitforce helps you test and easily improve everything on the mobile app. Splitforce is available on the iOS, Android, and Unity platform. A tool price starts from $19/month.
- Leanplum A/B testing tool seems like an app marketer’s dream. It has all the features needed and is easy to use. The best part about Leanplum is its availability on all platforms: iOS, Android, Unity, HTML 5, and it even has a REST API.
Author: Karolina Jasvinaitė is an online content marketer and blogger at Soundest. Everyday she discovers eCommerce latitudes and shares her knowledge and insights with readers @Soundest Blog | Marketing for eCommerce. Follow her on Twitter.