Discovery score v2
Understand what we measure.
Rank Lab measures how your product appears in sampled API answers to your chosen questions. Gemini and OpenAI are supported providers. These results are not measurements of the consumer Gemini app, Google AI Overviews, ChatGPT, or total market demand.
The discovery score
A score from 0 to 100 combines the following signals from successful checks in the selected time window. Direct brand and head-to-head comparison questions are excluded to avoid inflating discovery. Failed checks do not count as zero visibility.
| Signal | Weight | Definition |
|---|---|---|
| Mention rate | 30% | Answers mentioning your product / successful discovery observations |
| Recommendation rate | 25% | Answers explicitly recommending your product / observations |
| Official-domain links | 15% | Answers mentioning your product and linking its domain / observations |
| Position | 15% | Average reciprocal listed position; unranked answers contribute zero |
| Share of voice | 15% | Your mentions / your mentions plus distinct competitor mentions |
Evidence and uncertainty
No successful discovery checks means “Not measured.” A measured score of zero means successful answers did not contain the scored signals. We show question coverage and sample size alongside the score. Confidence is a heuristic combining coverage, sample size, and extraction confidence, not a statistical confidence interval.
Links are copied from the saved model answer and may be inaccurate; they are not independently verified search citations. Extracted mentions, positions, and recommendations can be wrong. Always inspect the original answer.
Markets, monitoring, and comparisons
Country and language are supplied as prompt context. We do not simulate a physically located consumer session. Questions rotate in a durable database-backed schedule; daily and weekly settings are target intervals subject to provider and worker capacity. Pause monitoring at any time. Audit history consists only of saved measurements.
Changing active questions changes the score’s coverage and composition. Compare like-for-like questions and inspect sample sizes before attributing a score change to product improvements. Suggested actions are hypotheses, not guarantees of traffic or higher scores.
Privacy and independence
Workspace reports are private to your signed-in email. Audit prompts and brand context are sent to the configured model provider. Private projects do not create public listings or ownership badges. Sponsored directory payments never enter the visibility scoring formula.