17 Proven AB Testing Techniques for Marketers

AB testing is the “science of marketing”. You have to rely on the quantitative data you’ve acquired from the experiment to make future decisions that’ll make or mar your marketing/growth campaign.

AB Test is a marketing experiment where you test at least two variations of the object or campaigns in order to see which campaign performs best.

AB Test in marketing is usually used to improve every stage of the growth funnel. I.e. (Acquisition, Activation, Referral, Revenue and Retention). However, the most common use cases are for improving engagements, customer acquisition, activation, and revenue.

---

Here are 11 majors ways/channels to conduct AB Test in Marketing

AB Testing Marketing Examples

1. Landing Page
   
   Running a marketing campaign? You can direct traffic to two or more landing pages to see which variations convert better.
   

   

   
   Example of a long landing page (left) vs short landing page (right)
   
   Landing Page A
   New Visitors: 10,000
   Conversions: 1,500
   Conversion: 15%
   
   Landing Page B
   New Visitors: 10,000
   Conversions: 5,400
   Conversion: 54 %
   
   Landing Page B has higher conversion. One can then decide to run the rest of the campaign on landing page B because of its higher conversion.
   

2. Website forms
   You can test forms for higher conversion. Create two signup/contact-us, etc forms.

   

   Form A(Left) - With 2 or 3 fields
   Form B (Right) - With 8 fields
   Test for which form variation will have higher conversions provided every other thing remains the same.
   

3. Audience/Device Targeting
   This is popular with digital advertising. I.e. (Google or Facebook Ads Manager). You can create two campaigns with different targeting to see which performs better. Targeting can be based on;
   
   - Device: Mobile/Desktop
   - Device Type: Android/IOS
   - Audience Behaviour
   - Other persona or demography.
   

4. Marketing Channels
   You can test the effectiveness of marketing channels. I.e.
   
   - Does Facebook Ads bring me more leads than Twitter Ads?
   - Lead from Facebook vs Twitter Leads: Which is more quality?
   - Which channel has the lowest customer acquisition cost?
   - etc.
   

5. Marketing Copy & Design Creative
   You can test;
   
   - Landing page copy
   - banner copy & sizes
   - Ads asset type (image, gift or video)

6. Email Copy
   You can test which email title or general copy works best for your email contact.
   
   - You can divide your customers into two and send them emails with the same content but different headlines.
   - You can change the content of your email copy to see which style makes them take more action.
   

7. CTA Button
   You can test;
   
   - CTA button copy
   - CTA button colour
   - CTA button sizes
   

8. Verification code
   Should I send the verification code/OTP to;
   
   - Customer’s email
   - Phone number
   - Both Phone number & email
   Which will give a higher success rate?

9. Timing

   You can test for;
   
   - The time or days your ads campaign performs the best. Some campaigns are known to perform X2 better on weekends than on weekdays. This will help you determine how you spread the campaign budget across days.
   
   - Sending Email newsletter for 6 am on Monday or 8 am on Tuesday? Which has a higher open rate or conversion?
   

10. Social Media AB Testing Opportunities
    - Text type
    - Text vs Image
    - Gif vs Image vs Video
    - Video length
    - The best time to post
    - Number of posts
    - etc.

11. Lead generation magnet/popups

AB Testing Opportunities to Drive Core Growth

• Dashboard onboarding and experience

• Checkout process

• UX testing

• Email Onboarding

• Signup - Activation Process

• Landing Pages - Signup Process

AB Testing Guidelines

• Define every AB testing objective before running any experiment.

• It’s best to change only ONE Variation at a time
  Testing 3 different marketing channels with 3 different landing pages might be confusing. It can be hard to determine the major cause of changes in conversion.
  
  If you want to test marketing channels, test only that.
  If you want to test landing pages, test only that.
  

• The larger the test sample, the better the result.
  Historically, large samples tend to decrease the margin of error and improve confidence in the results. You’ll get a clearer picture if you send 10,000 visitors each to Landing Pages A & B than sending 100 visitors.
  

• Have your tracking & reporting tool in place
  Identify the best tools for tracking & reporting to measure success, set them up, and test them. The fate of your test depends on your ability to collect and interpret the right data correctly.
  
  You can depend on marketing and product tools like Fullstory, Hotjar, Google Analytics, Mixpanel, Heap, Appsflyer, Google Ads Manager, Facebook Manager, etc. are pivotal to success.
  

• Avoid external influence as much as possible
  External issues such as page speed, server overload, etc. can affect the result of the test. Especially if these issues were present at Test A and absent at Test B.
  
  A sudden news mention of your brand can also skew the result of your test. Make sure to create a controlled testing environment that will be unaffected by external factors.

The Best AB Testing Models for Marketers

There are three (3) proven AB testing models that are best for marketing experiments.

1. Random Testing Model

This is when you conduct simultaneous and random tests.

• You spend $100 each on Facebook Ads, Google Ads, and Twitter Ads at the same time. And all the resulting traffic to the same landing page, with the same copy, design, targeting, and campaign time. or

• Or when the system randomly shows visitors any of the two signup pages you’ve created for AB testing.

2. Spilt Testing Model

This is when you split your audience into segments, your landing pages, design or copy. I.e.

• When you divide your customers into two segments and send them a mail to see how each segment engages with it.

• When you send visitors from the same source to different landing pages to see which converts better.

3. Standard Testing Model
This is when you complete Test A, before starting Test B.

• Onboarding Flow A - Used for 2 weeks

• Onboarding Flow B - Used for another two weeks.

The standard model is the most difficult to protect against external influence.

Imagine if you had a news mention of your brand when you were testing Onboarding Flow A, which drove tons of users to your site, and caused server overload. And then two weeks, later, when you were testing Onboarding Flow B, there was neither an abnormal usage nor server overload?

This scenario above is just one of the factors that can skew AB Testing results especially, with the Standard Testing Model.

Conclusion

AB testing is the “science of marketing”. You have to rely on the quantitative data you’ve acquired from the experiment to make future decisions that’ll make or mar your marketing/growth campaign.

If you have to prioritize AB Testing for your company, start from where you have the largest drop in the AARRR growth model or where there’s a potential to get an immediate increase in revenue or whatever test has huge potential to increase your North-Star Metric.

For most start-ups, it’s always better to get a 10% increase in revenue than a 50% increase in acquisition numbers.