Email Analysis

The Most Effective Way To Use Split Testing

You will have been told that you should test everything you do, and then test again. An email marketing campaign that doesn’t include a split test somewhere in it is a waste. You’ll never get that opportunity again.

Split testing is a simple system of problem solving. You identify something that is of concern, for example, your open rate. It might have been increasing at a steady rate for some months but now appears to have plateaued. 

There are any number of possible causes, such as Subject Line. Subscribers might be getting a little fed up with the same style all the time. Or it might be timing. You don’t know. Split testing will identify a cause.

As with all things email marketing, guidelines are just that; guides. You can ignore them if you wish. However, the one rule is to have a good reason for doing so. 

1/ Test only one thing at a time

Split testing, otherwise known as A/B testing, is used to show whether a single alteration works. It returns dependable data although some interpretation might be required. Testing more than one feature is known as multivariate email testing and has different guidelines which you can also ignore, although at your peril.

2/ Test something that is likely to increase your RoI

Don’t test for testing’s sake, although every metric can be improved. It’s more a case of prioritising. Pick the aspect of performance that is likely to improve your RoI the most. 

The Most Effective Way To Use Split Testing3/ Split your list randomly

Let’s accept that there’s no such thing as random. However, you can do it in a manner that makes the results statistically dependable. If you want to test 10% of the list, then every tenth person is about as random as it gets. Do not pick a factor, such as age, that is personal to subscribers.

4/ Know the limitations

If you test on a free offer, the result might not be applicable to a campaign to sign up to a seminar. Results can be specific. 

5/ Keep the numbers up

There is no fixed percentage of split to give a dependable result. If you have a small email marketing list, then 50:50 might be an essential. If you glory in a big one, then 5% might be more than enough.

6/ Give it time

Run the test for a sufficient time. The smaller your email marketing list, the longer you need to run it to obtain dependable data. 

7/ Remember where you started

Just comparing one side of the test is not enough. Check your data before testing and see if there’s been any overall improvement. 

8/ Believe your data

It should not need saying but instigate changes on the results from the test, otherwise there’s no point in split testing. If the difference is minor, and you are not certain it is significant, then test again, perhaps on another email marketing list. 

Split testing is a skill that must be mastered to get the best out of email marketing. It should also be continuous. 

WizBot

EMAIL MARKETING FREE TRIAL

30 days full functionality - No credit card required - INSTANT ACCESS