A metric has one owner
The contract is stored. Every name is checked against it, never prompted.
One definition of a metric, proven against the warehouse
Revenue means one thing in the semantic model. It can mean another in a workbook built in a hurry. An executive usually finds the gap, in a meeting.
Every answer uses your certified numbers.
You define a metric once in Salesforce Data 360, Snowflake, Databricks or the Palantir AIP Ontology. The AI names it; TabTotal writes the SQL from its own stored copy of your definition, using your identifiers. The number in a conversation is the number in your warehouse.
The definition lives in the warehouse
Define a metric once in Salesforce Data 360, Snowflake or Databricks. That is the definition. Everything downstream either matches it or is a finding.
- Read from the governed catalogue, never retyped by hand
- Metric contracts checked against what is stored
- Read-only. The warehouse is a source, not a target
Several warehouses, one logical object
The same table can sit in more than one warehouse. Here it is one object. Every pair is compared, so where any two disagree, you get a governance finding.
- Each warehouse syncs and approves on its own
- A difference between warehouses is surfaced, with the pair named
- Warehouses that match return zero findings
Lineage that reaches the column
A dashboard depends on a field. The field depends on a column. The column lives in a warehouse table. Lineage that stops at the table cannot say what breaks when that column changes.
- Dashboard to field to column to warehouse table
- Column inventories on every warehouse table
- A map that reached table level says table level
What changes
Impact is answerable
Change a column and see the dashboards that depend on it, by field.
Disagreements are listed
Where the warehouse and the workbook differ, you get the list.
Connect a catalogue
Point it at one warehouse, read-only. See the objects and the contracts it finds.