Learn From Millions Of Email Experiments (Without Having To Run Them)

When I ask email marketers what their major projects are for the year, invariably I hear a list that includes some combination of efforts to build more contextual, triggered messaging programs and a desire to build lifecycle messaging. The typical approach to building these types of programs is to build a message (or series of messages) that are modeled on an innovative competitor or other email marketer’s program: the “template.”

Marketers typically find a template that they like and “make it their own” by rebranding and adding other features that they think fit their situation. If they have the time,  marketers will iteratively test and improve the initial version of the program until they see diminishing returns from the changes. Then the program is considered fully baked.

This approach presents some problems. Specifically:

The template program may not have been optimal for the originating company’s subscribers. The template that you choose may have only been minimally tested, and may not even have performed well for the originating company..



Once you have borrowed your template from another company, you may not have time to do a lot of incremental testing. Since testing takes a lot of work, you may not have the time to test program parameters  such as number of messages, time between messages, the classes of offers, CTAs, and other elements.

The template program was optimized for someone else’s subscribers. It’s very rare for another company in your industry to share more than 50% of subscribers on your list. Even if the template program has been fully optimized, that doesn’t make it optimal for your program.

So what can you do instead to develop successful email programs?

My suggestion is to take a look at the data available from email panel providers to understand the programs that work best for your subscribers. I recommend a three-step process:

1. Develop a list of program types (e.g., birthday, abandoned cart, winback, others) that you are interested in adding. This is the easy part. It's the rare marketing team that doesn’t have a “wish list” of new message types they want to build.

2. Use a combination of panel data and informed intuition to pick the program type that you work on. The primary approach I’ve seen is to make this decision based on the message types that drive the highest read rate (as measured in the panel data).

A more strategic approach is to take a look at lifecycle engagement. In other words,  what percent of subscribers in the first (second, third...) week of life on your list have engaged with you in any way over the last 30/60/90 days? From there, you can determine the points at which subscribers are typically disengaging.

By using panel data, you can compare your “lifecycle curve” to that of other leading marketers. If you find a company that’s performing better during a certain part of the lifecycle, take a look at the kinds of programs they are sending during that phase to retain engaged subscribers. Intuition and knowledge of your business also plays a big role here. 

3. Once you’ve picked a program type you want to build, choose a template from the messages that perform best for your subscribers from panel data. You can learn a lot about which templates work best in general. However, assuming you have enough panelists, picking as a template those programs that work best for your customers will provide a more optimal set to choose from.

By following the steps above, you’ll be learning from millions of experiments run by many thousands of other marketers before you start your own optimization.

Next story loading loading..