Baseline window
The baseline window is the before state of the experiment.
It shows what prompts, answers, sources, and competitors looked like before the change. Without it, the experiment result is hard to judge.
Clean baseline period
Use a clean period before the main change.
The baseline does not need to be perfect. It needs to be clear enough to compare. Avoid using a period where many unrelated changes happened.
Baseline window
The baseline is the before state for mentions, answer wording, sources, and competitors.
Data before start
Look for the main signals before the experiment begins.
| Signal | Baseline question |
|---|---|
| Mentions | Was your brand named before the change? |
| Sources | Which pages were read before the change? |
| Competitors | Who appeared before the change? |
| Answer wording | How did the answer explain the topic? |
Measurement effect
The baseline keeps the test honest.
If results improve later, the baseline helps explain what changed. If results do not improve, it helps show whether the test was too small, too early, or pointed at the wrong question.