‘How to split test your emails’ is probably the biggest email marketing question I’ve never written about.
Why not? I don’t know. You can and should split test your core email sequences. In my world that’s the sequence that goes out after people opt-in. My objective at that point is to sell someone a book. Which is an unclear conversion metric, because if someone buys from Amazon I can only see that if they respond to an offer within the book.
Split-testing can work at different levels. You can split test the subject line of an email, keeping everything else the same. You can split test an entire email, testing a completely different approach. Or you can split test an entire email series, including completely different emails.
Testing subject lines is akin to fine tuning. Testing entire sequences is more like fishing with dynamite. Although as with other forms of split testing, you can limit your risk by sending most of your contacts through your control sequence.
You need a reasonable flow of new contacts through a campaign for the results of a test to be meaningful. And as with all forms of split-testing, you need to document your tests. Otherwise you’ll lose track of the key learnings.
I sometimes find that people try to split test emails too early. Split testing only works when you already have a working control sequence. In many of the projects I work on, we’re trying to create an effective control sequence. You can only split-test once you have a baseline level of performance, and clearly defined metrics. (As discussed yesterday, I suggest you use Google Analytics as a good starting point).
As a rule of thumb, you want to create a balance between live production and core sequence split-testing. Most of my emails go as live broadcasts. I could conceivably test different subject lines, or even different emails. But the payoff from doing that in a single broadcast is relatively small.
Rather than muddying the water, I’d suggest you focus your split-testing efforts on your most critical email sequence. This sequence is always related to the biggest bottleneck in your business.
In my business, my primary metric is to convert cold prospects leads into book buyers within 30 days of opt-in. So it makes more sense to split-test that sequence rather than individual broadcasts.
Unless your email software caters natively for split-testing, you’ll usually need a third party service. With Infusionsoft for example you could use PlusThis, My Fusion Helper, or InfusionLabs.