How your score is protected

1. 1. Harsh and generous raters are evened out

   Some people rate everyone a 3; others hand out 5s. Before anything is scored, each rater's pattern is statistically normalized against the organization (once they've rated enough for a pattern to exist), so being rated by a tough grader doesn't cost you and an easy one doesn't inflate you.

2. 2. No single person controls your score

   In 360 mode, different perspectives are weighted deliberately — peers as a group carry the most weight, and a perspective with too few raters (for example, a lone peer) is excluded entirely, with its weight redistributed to the perspectives that have enough voices. In advisor mode, by design a single named advisor rates the team directly, and score surfaces state which assessment mode the workspace runs.

3. 3. Too little data means no score — not a bad score

   In 360 mode, any behavior with fewer than 3 raters is suppressed: it shows as 'insufficient data' rather than a number, so no published 360 score is one a single opinion could have produced, and no one can reverse-engineer who said what. Advisor-mode scores are openly single-rater and labeled that way.

4. 4. Inflated self-ratings are automatically discounted

   If a self-rating is far above what everyone else consistently observed, the self-rating's weight is halved for that behavior. Honest self-assessment is never penalized — only statistically implausible gaps are.

5. 5. Every number carries its confidence

   Scores display with a confidence level (and, in 360 mode, a confidence interval) that reflects how many people contributed. Fewer raters = visibly wider uncertainty. The system never reports more confidence on less data.

6. 6. The AI cannot invent a statistic

   Every AI narrative is generated from — and validated against — the same computed numbers you can see. Each claim carries its evidence chips, and any AI-rewritten text is checked so it contains no number that isn't grounded in the cited data. If validation fails, the deterministic version is shown instead.

7. 7. Risk flags are reviewed by people, resolved by evidence

   A behavioral risk flag is raised by the scoring engine, acknowledged by a leader (with the action taken on record), and marked recovered only when a later cycle's scores show the behavior actually improved — never by someone clicking it away.

8. 8. Who can see your scores

   Your own scorecard is always yours. Your manager and your organization's leaders (executives and HR admins) can open individual scorecards — that's how they coach and support people. But peers who rate you cannot see your scores, and because ratings are anonymized and suppressed below 3 raters, no one — leader or peer — can tell who said what about you.

9. 9. Built for development, not decisions

   Kinetry is a coaching and development instrument, not an employee-evaluation system. It's designed to guide growth and conversations — not to be used as the sole or primary basis for pay, promotion, or termination.

Most of these run in the scoring engine itself (not policy documents) and apply identically to everyone in your organization; the last two are how Kinetry is designed to be used.

Related: how Kinetry connects kept coaching commitments to later score movement — including the floors and the limits of the claim — is published as the Kept-Commitment Method, and Kinetry's limits as a whole are stated in full at What Kinetry cannot tell you.