Five ideas cover most of what you do in Data Shepherd.
Rules
A rule is one check against one table, such as “the EMAIL column in
CUSTOMERS must not be empty”. You create a rule by choosing one of the 25
built-in rule types, the table and column it applies to, and its parameters.
Rules only apply to tables you have connected to the app.
Runs
Each time a rule executes, Data Shepherd records a run with its status, its counts and the full result. A run ends as Passed, Data issue when the rule found a problem, or Could not run when the rule itself failed - for example because a column it names no longer exists. You see every run on the Executions page.
Alerts
When a rule that has an alert destination finds a problem or cannot run, Data Shepherd records an alert. Alerts show on the Alerts page until someone acknowledges them. Delivering them by email or webhook is optional - see Alerts and notifications.
Schedules
A rule runs when you run it, or on a schedule you give it: daily, weekly, hourly or at an interval. One task owned by the app runs every scheduled rule, so you do not create a Snowflake task per rule. Scheduling ships switched off and costs nothing until you turn it on.
Quality dimensions
The dashboard groups results into five quality dimensions - completeness, uniqueness, validity, consistency and timeliness - and scores each one, so you can see which kind of problem is growing. A dimension’s score is the share of its rule runs that passed. Not every rule counts toward a dimension, and a dimension shows only when its rules ran in the range you chose.
Where it all runs
Everything above happens inside your own Snowflake account. Rules read your tables in place, and results, alerts and the audit trail are kept in the app’s own tables there. They are exposed as read-only views, so you can query them or build your own dashboards on them. There is zero data egress.
The usage guide has the detail and the SQL.