In brief
A split test compares two versions of a campaign or an ad group by changing a single variable, to find out what works best. The variables offered are the creative, the targeting, the placements, optimization and bidding, or a custom test (two campaigns or groups of your choice). Automated rules carry out an action when a condition is met.
Why it matters
Without a clean test, it's impossible to know why one version works better. Rules save you from monitoring all day.
The steps
- From Create, choose Create a split test, or open the Experiment Manager.
- Choose the variable to test (creative, targeting, placements, optimization and bidding) or a custom test.
- Define the two versions, the duration and the success criterion.
- After the test, compare the results at equal budget.
- For rules, open Automated rules: use a template (schedule delivery, increase the budget of high-performing ads, stop those whose cost is spiraling…) or create your own.
Pitfalls to avoid
- Testing several variables at once.
- Stopping the test too early.
- Comparing versions with different budgets.
- Creating rules that are too aggressive and stop ads during the learning phase.
Frequently asked questions
Which variable to test first? Often the creative, because it has the biggest effect.
Are rules mandatory? No: they’re a convenience.
Key takeaways
- One variable, one test.
- Same budget, same duration.
- Rules = simple automation, to be monitored.
Practice in TRIALO
In Create, find the split test entry; then compare two ad groups that differ only by audience.