Name the question.
Test a specific idea: the message, the proof point, the visual treatment or the call to action. Write down what you expect to learn.
Keep the comparison useful.
Avoid changing everything at once when the aim is to understand one variable. Make sure each version still expresses the same offer accurately.
Choose the right measure.
Judge the work against its objective. An attention metric answers a different question from an enquiry or purchase metric.
Read the context.
Consider audience, placement, spend and timing before drawing a conclusion. A small or unbalanced sample can make a result misleading.
Bring learning back to the brief.
Record what the test suggested and what remains uncertain. Use the next round to sharpen the question instead of repeating the same activity.