ZenboostThe four metrics that matter
| Metric | What it reveals | Useful cadence |
|---|---|---|
| Mention rate | The share of your buyer questions where your brand is named | Monthly |
| Share of voice | Your weight against competitors cited on the same questions | Monthly |
| Cited sources | The pages and domains holding authority on your topic | Quarterly |
| Description accuracy | What AI claims about your offer, right or wrong | Quarterly |
Keep the same question set
The classic trap is changing the questions at every measurement: scores stop being comparable. Freeze a representative list of buyer questions and only change it when your offer changes.
Reading a change
- A rise after publishing problem-solved pages: the explicit content is being read, keep going.
- A drop with no change on your side: a competitor published, or the model was updated.
- A score stuck at zero: the topic is still dominated by generalist sources — target more specific questions.
What is not worth tracking
The exact wording of a single answer: it varies every session. Track aggregates across a set of questions, not an isolated screenshot.