Email Marketing Bible · 9 of 16
Testing and optimisation
Run tests that tell you something: one change, a measure chosen first, and a record of what you learned.
What you will be able to do
- Design a valid A/B test
- Pick a measure beyond opens
- Keep a test log
Rules
Change one thing, choose the measure first, split randomly, send at the same time and use enough recipients.
Measures
Clicks and replies beat opens, which can be inflated by image preloading. Tie the measure to a business result where you can.
Small samples
Differences between small groups are often chance. Repeat the test before acting on a marginal result.
Log
Record the hypothesis, the groups, the result and the decision.
Your output: checklist
- ☐ Write the hypothesis before testing
- ☐ Choose clicks or replies as the measure
- ☐ Keep the groups random and equal
- ☐ Repeat a surprising result
- ☐ Log every test
Related