Agami Insights

Lightspeed Retail data warehouse diagram: a gift card sale whose total, calcTotal, calcSubtotal and calcPayments all read zero, beside displayableTotal carrying the hundred dollars.

Lightspeed Totals a Gift Card Sale at Zero, Correctly

A customer hands over a hundred dollars for a gift card and Lightspeed writes the sale down as zero, on purpose. The semantic model is where your warehouse learns that is deferred revenue.

Square net sales vs total sales diagram: two adjacent payments columns, one including tip_money and one excluding it, beside the five named lines of Square's sales summary.

Square Puts Tips on Their Own Line and Replication Folds Them Back In

Square's report puts tips on their own line, because tips aren't sales. Replication lands fifteen money columns and none of the definitions. The semantic model is where revenue gets defined.

PeopleSoft ACAD_PROG EFFDT EFFSEQ data warehouse diagram: every dated row of a student's program stack, beside the one current row PeopleSoft Query returned by default.

In PeopleSoft, a Student's Current Program Is Not a Column

Every program change is still a row in PS_ACAD_PROG, and the tool you reported from dropped them by default. That default doesn't replicate. The semantic model is where the current row gets defined.

Two SyteLine objects for one entity, the all-sites coitem_mst table a connector copies including its syteline site_ref column, beside the single-site view it leaves behind.

Every SyteLine Table in Your Warehouse Holds Every Site's Rows

Every SyteLine table holds every site's rows. The object that filtered them to one plant was a SQL view, and your connector copied the table. The semantic model is where the site becomes a key.

An Infor LN table company number, decoded out of the physical table name tdsls401300 into package, module, table number, and company, beside the tables the replication landed.

How to Query Infor LN by Company When the Company Number Is in the Table Name

One Infor LN table lands in your warehouse as many tables, one per company, and not one of them has a company column. The semantic model is where the company becomes a column you can group by.

Oracle Fusion BICC data warehouse amounts: the four journal line debit and credit columns beside the one column whose name ends in Amount, and the conversion rule that catches each.

Oracle Fusion Balances to the Cent. Your Warehouse Can't Prove It.

Oracle's ledger balances to the cent because the amounts are decimal. The connector lands them as floating point, and the semantic model is the only place that records they used to be exact.

IFS Cloud code part dimensions in the warehouse: nine ledger dimensions named CODE B through CODE J, beside the named dimensions on the same general ledger fact.

The Only Place IFS Says What Code Part B Means Is a Power BI File

IFS ships nine ledger dimensions named B through J, and never writes down which one is the cost centre. The semantic model is where that finally gets said.

SAP MATNR leading zeros: the readable form is declared on the type, replication copies only the column, and the padded value lands beside a trimmed copy that joins to nothing.

Two Columns, One SAP Material Number, and the Join Uses the Padded One

Your SAP warehouse stores material 4711 as 000000000000004711 and ships a trimmed copy beside it. Only one of the two joins to anything. The semantic model declares which.

Workday RaaS data warehouse grain: one data source fixes one row per primary business object inside Workday, and the warehouse gets those rows under a table name somebody typed.

Only Your Semantic Model Knows Whether a Workday Student Row Is a Student

A Workday Student table is one custom report's output, at a grain Workday fixed inside the tenant. The table name was typed by hand. The semantic model is where that grain gets declared.

Infor M3 custom fields data warehouse: the item master lands complete with no customer-defined field, while the values sit in CUGEX1 behind a literal table name in a column called FILE

In Infor M3, Your Custom Fields Are Rows in a Table Called CUGEX1

Infor M3 keeps every customer-defined field in CUGEX1, found by matching the extended table's name in a column called FILE. The semantic model declares that join and its mandatory filter.

Oracle EBS org_id in your warehouse: a session variable sets the operating unit and a view filters on it, while a replication has no session and lands every unit in one table

Oracle EBS Secured Every Operating Unit in a Session Variable. Your Warehouse Copied the Table.

Oracle partitions EBS by operating unit through a view reading a session variable. A warehouse has no session, so every unit lands in one table. The semantic model makes org_id a required dimension.

Kantata data warehouse hours by client: workspace_association has two columns and both are the primary key, so a project in three groups has its hours counted three times

The Kantata Link Table That Counts the Same Hour Once Per Group

Kantata has no client table. A client is a group with a box ticked, a project can sit in several, and the obvious join counts its hours under each. The semantic model declares the path that totals.