Revenue means one thing in every answer.
Agree what a number means once, with its rules and its time period. Every question that mentions it gets the same definition, whoever asks.
Semantic model
4 models · 3 relationships · derived from foreign keys
- PKcustomer_iduuid
- FKregion_idint
- first_order_attimestamptz
- channeltext
- PKorder_iduuid
- FKcustomer_iduuid
- placed_attimestamptz
- net_revenuenumeric
- PKproduct_idint
- categorytext
- unit_costnumeric
- PKitem_iduuid
- FKorder_iduuid
- FKproduct_idint
- quantityint
- customers1:Norders
- orders1:Norder_items
- products1:Norder_items
The definition argument, settled once.
Most teams carry three versions of revenue. This is where that argument gets had once and written down.
Starts from your data
Connecting reads your tables and how they relate to each other, so you start with a working draft instead of a blank page.
Named by your team
Rename fields into the words people actually use, hide the ones nobody should touch, and give each number the name your business says out loud.
Answers cite the definition
Every answer shows which definition it used and how that number is worked out, so a review can move past the arithmetic and on to the decision.
Owned and versioned
A definition has an owner and a history. When it changes you can see who changed it, when, and what the number was before.
Activity
Meridian Outfitters
- PR12m ago
Priya Raman pinned a chart Net revenue by region
Revenue analytics · Dashboard
- 07:00
Scheduled run completed Weekly revenue digest
6 tiles refreshed · 3.2s
- DOYesterday
Daniel Okafor saved a query New customers by channel
Verified · semantic
- Yesterday
Compiler refused a query Revenue by campaign
Ambiguous join path on campaigns
- MSMon
Marco Silva connected a source Warehouse
ClickHouse · 118 tables · read only
Why two reports agree
A definition the answers actually follow.
In most catalogs a definition is a document people ignore. Here it is the thing every answer is built from.
The level is declared
A definition says what level it is measured at. Ask for it by month or by region and iDash rolls it up correctly instead of guessing.
Each thing counted once
The catalog knows how your records relate to one another, which is what lets iDash count each customer once even when they have twenty orders.
Two paths, no guess
When there are two ways to connect a number to a breakdown and they disagree, iDash stops and asks which one your question means.
One entry, every surface
Chat, dashboards, scheduled runs and embedded views all read the same entry. There is no second version of revenue living in somebody's saved report.
The question
What was revenue by region last quarter?
- 1Used your definitionNet revenue, the way your finance team already defines it.
- 2Counted each order onceAn order with six items is still one order, so the total is not inflated.
- 3Grouped by regionFollowed the one clear link from an order to the customer's region.
- Only one way to connect orders to regions, so no ambiguity
- Each order counted once, whatever it contains
- Read only connection, capped at thirty seconds
Ask again tomorrow and it runs the same way.
It sharpens with use.
The catalog is the context an agent reads before it answers anything, and it grows as your team confirms answers.
Verified answers become examples
Mark an answer as correct and it becomes a reference for the next person who asks something similar. Only answers a person has checked qualify.
The vocabulary agents read
Anything built on iDash, including your own agents, looks names up here rather than inventing a meaning from a column heading.
Define once, answer often
The work of agreeing definitions lands in one place instead of in each analyst's saved report, so your data team maintains meaning rather than re-answering the same question.
Settle one definition this week.
Start with the number your team argues about most, and give it a single home.