Before you test, track!
It’s easy for an email marketer to develop and embark upon a comprehensive test plan before they even have a baseline—especially across their segments. For example, when it comes to testing subject lines, most email marketers simply do an A/B split across their entire database and use the open rate to determine the winner. The findings, if any, may or may not be applied to future mailings.
Before this type of testing even occurs, the audience should be segmented and tracked for at least four to six mailings to compare performance across primary segments, whether they are persona segments, recency segments, member segments, etc. This tracking can actually save some future heartburn if a particular email campaign does not perform well, either from an open/click perspective or from a conversion/revenue perspective.
Let’s say that an email marketer has been tracking top-tier, mid-tier and bottom-tier members, and now knows the average open, click and conversion rates for each. When a huge offer goes out, but conversions are low, it may be tied to only one single segment. Without having first tracked performance by segment, how would that even be known? Once this performance tracking has been done for a few sends, testing of subject lines, offers, or content will yield much more meaningful results.
The 2012 MarketingSherpa Email Marketing Benchmark Report references the fact that the majority of email marketers do not routinely perform testing as part of their email campaign process. This makes sense given how busy the average email marketer typically is. That said, the initial process of simply tracking key segments is not only reasonably done, it is an easy way to CYA in an economy where marketing performance is under increasing scrutiny.
Takeaway: The exercise of email testing may be futile without understanding how your key segments behave. What works for one segment of your database may not have been as effective for another. Before you embark on a testing plan, track the performance of your segments for at least three to four sends. When you do implement a test, make sure to monitor performance by segment to do a proper determination of your test winner.
