RIS Glossary

Abandon Experiment

Terminates a running experiment (or deletes one that hasn't been run) and returns to the empty-state creation screen.

Acceptance Ratio

The percentage of reviewed items that must favor an experiment before it can be promoted to production. This ratio can be modified in Settings.

Alert

A system-generated notification surfaced in the Dashboard Alerts tab and the Notification Bell. Alerts communicate operational events requiring attention (e.g., record count thresholds, license expiry).

Cortex AI Model

A Snowflake Cortex AI language model used during onboarding to generate a SQL mapping prompt for your source data.

Coverage

The percentage of records that have a non-null value for a given field.

Evaluation Mode

A deployment configuration for trial or evaluation instances, indicated by a persistent amber banner at the top of all pages. All features remain functional.

Experiment

A test configuration that runs against the same data as production, allowing side-by-side comparison before committing changes. Refer to Experiment for more information.

Frequency Limit

A threshold that suppresses data values appearing more often than the specified count, preventing common values from creating false matches. This limit is configurable in Settings.

Full Match

A production match job that processes all records from scratch.

Goal Indicator

A status bar in the Experiment Review tab showing whether the current experiment meets promotion criteria (green) or what remains before it can be promoted (amber).

Group Brittleness

A grouping parameter controlling how many parts transitive groups tend to be split into, when confronted by incompatible group members. This can be thought of as, “if a group is going to split anyway, is it going to split or shatter completely.”

Group Reach

A grouping parameter controlling how strongly the algorithm attempts to keep together transitive groups that are “stringy” or “ropey” and contain incompatible group members.

Group Stickiness

A grouping parameter controlling how strongly records remain in their current group, in the presence of incompatible group members.

Incremental Match

A production match job that processes only new or changed records since the last run.

Manual Override

A user-imposed decision to force-match or force-break specific records, overriding the automated algorithm. Refer to Manual Overrides for more information.

Mapping Mode

The approach selected during onboarding for preparing source data: Scan Customer Source Tables (AI auto-scan), Manually Complete Prompt (template edit), or Skip (no AI).

Match Group

A set of records determined by the algorithm (or manual override) to represent the same real-world entity (individual or household).

PII

Personally Identifiable Information: data fields such as name, address, phone, email, and social identifiers used for matching.

Promote

The action of accepting an experiment's ruleset and applying it as the new production configuration.

Reject

The action of discarding an experiment's results and reverting to the existing production configuration.

Restart Experiment

The action of re-running an existing experiment from scratch using the same configuration, without changing the ruleset.

Ruleset

The complete set of matching parameters (tightness, rules, frequency limits, grouping settings) that defines how records are matched.

Service Standby

A state where the Snowflake container service is suspended due to inactivity. RIS redirects all navigation to a holding screen and auto-resumes when the service comes back online. This is sometimes referred to as “Sleep mode.” Refer to Enable Service Standby for configuration and management details.

Suppression

The exclusion of specific high-frequency data values from the matching process to prevent false matches. Refer to Working with suppressions for details.

Tightness

A parameter controlling matching sensitivity. Lower values produce more matches (looser); higher values produce fewer, more confident matches (tighter).

Transitive Groups

Groups formed through chains of matches (A matches B, B matches C, therefore A-B-C are grouped). Such groups always survive unless they contain incompatible members that were brought together by transitivity. Transitive-group splitting is controlled by stickiness, brittleness, and reach parameters.