Separate analytics
Separate analytics help you read an experiment without the full website noise.
When an experiment is selected, the dashboard uses experiment questions and experiment chatlogs for the scoped views. This affects the way Analytics, Sources, Competitors, and History are read.
Experiment analytics need their own readout
Use the experiment readout to answer one question: did the test move in the direction we expected?
Do not try to explain the full website from an experiment. It is a smaller lens. It should help you read one test.
Separate analytics
Experiment analytics are a smaller lens for one test, not the full website truth.
Compare without polluting the main view
The main dashboard is still the full website view.
The experiment view can be cleaner because it has fewer questions. This makes it easier to inspect answer text, source movement, and competitor movement for one focused goal.
Signals worth watching
Watch signals that match the experiment goal.
| Signal | What it can show |
|---|---|
| Mention movement | Your brand appears more or less often |
| Answer wording | The model explains your offer differently |
| Source movement | Different pages are being read |
| Competitor movement | A competitor appears less, more, or differently |
Decision notes after the test
Write the final note in plain words.
For example: "The page helped source reads but did not change mentions yet." That is more useful than saying the experiment won or failed too early.