This page provides answers to frequently asked questions.
Q: Why can't I access Production, Dashboard, or other pages?
A: The onboarding wizard must be completed first. Navigate to Configuration and complete all wizard steps through the "Initialize Ruleset" action.
Q: Why is the Promote button disabled in Match Review?
A: Review constraints have not been met, OR you are not in the Operator role.
-
For review constraints: Check the status message below the button -- it will indicate how many more items need to be reviewed or what acceptance ratio is required. Adjust these thresholds in Settings > Review Constraints if needed.
-
For Operator role: Check the status message. If you are not an Operator, someone with the Operator role will need to promote the experiment, or change your role to Operator.
Q: What is the difference between Full Match and Incremental Match?
A: A Full Match reprocesses all records from scratch, which is thorough but slower. An Incremental Match processes only records that have changed since the last run, making it faster, sometimes much faster, when a small number of records change.
-
You must run Full Match after any rule configuration changes.
-
Use Incremental Match for routine processing of deltas.
-
Avoid Incremental Match if more than 0.2% of records have changed. It will be slower.
-
There is nothing wrong with running exclusively Full Match. It will take longer, but it may not matter to you.
-
Also, you must run Full Match periodically. The Incremental Match can accumulate subtle errors over time, and a Full Match will reconcile those discrepancies.
Q: Can I run an experiment while a production job is running?
A: No. Only one job (production or experiment) can run at a time. Wait for the current job to complete or abort it before starting another.
Q: What happens when I promote an experiment?
A: The experiment's ruleset replaces the current production ruleset. A Full Match job is typically needed to apply the new rules to all records. Manual overrides that were previously saved are retained.
Q: How do manual overrides interact with the matching algorithm?
A: Manual overrides express force-match (group together) or force-break (separate) decisions. These overrides take precedence over the automated algorithm during the next production match run. Overrides that become redundant (the algorithm naturally produces the same result) are automatically cleaned up.
Q: What does "Suppression" mean?
A: Suppression removes high-frequency data values from the matching process. For example, if a generic email like "info@company.com" appears in thousands of records, suppressing it prevents the algorithm from incorrectly grouping those records together.
Q: I see "No data available" on the Dashboard. What should I do?
A: Ensure at least one production match job has completed successfully. Dashboard metrics are populated from production run results. Try adjusting the time range selector to a broader period.
Q: The application shows a "Service Standby" screen. What do I do?
A: The Snowflake container service is temporarily suspended. Wait for automatic resumption. If the message escalates to indicate a long wait, contact your Snowflake administrator. Do not refresh the page repeatedly; the application polls automatically and will navigate away once the service resumes. This feature is configurable; refer to Installing & configuring the app for details.
Q: Why do I see an amber banner at the top of every page?
A: Your RIS instance is in evaluation mode. All features are functional. The banner is informational only.
Q: What is the difference between Abandon and Restart for experiments?
A: Abandon deletes the experiment entirely and returns to the empty state. Restart re-runs the experiment using the same configuration without deleting it. Use Abandon when you want to start a completely new experiment with different parameters. Use Restart to re-run with the same parameters (e.g., after data changes).
Q: How do I access Manual Overrides from an experiment review?
A: In the Experiment page Review tab, click the Manual Override button in the results detail panel action bar. This navigates to Manual Overrides with the review item records pre-loaded. A "Back to Review" button is provided to return after saving.
Q: When using RIS, when are my own compute resources being used as opposed to Redpoint’s?
A: RIS is a “data-in-place” application, and as such, your Snowflake resources are used for all data access, analysis, and match processing. Redpoint’s resources are only used for address data that is sent to the externally-hosted Hygiene service.
Redpoint creates a new database for the SHARED.INPUT and SHARED.MATCH_RESULT tables, as well as all of the “internal” tables in the PRIVATE schema. Application packages (SNAs) like RIS are built on a specialized database types as the deployment artifact. The RIS database does not represent a significant cost to you, because it doesn't do much more than hold storage for the tables. However, there are other aspects of the RIS SNA that have important costs implications:
-
A warehouse, which is a computational resource for executing SQL on the database tables
-
A small compute pool to run the RIS service and web app
-
A large compute pool to execute the matching processes