Documentation Index
Fetch the complete documentation index at: https://docs.mantrixflow.com/llms.txt
Use this file to discover all available pages before exploring further.
The Normalisation tab is in the Destination panel (⚙️ on any Destination node). It applies column-level rules to the incoming source data before it lands in the raw staging layer that the dbt Layer reads from.
Where it fits in the pipeline
Source (public.tablename)
↓
Normalisation ← rename columns, exclude columns
↓
Raw layer (raw.public__tablename)
↓
dbt Layer ← SQL model on raw.public__tablename
↓
Destination (analytics.your_model)
Two rule types
Rename
Map a source column to a different name in the raw layer.
| Source column | Rename to |
|---|
customer_id | customer_uuid |
email | email_address |
first_name | fname |
created_at | registered_at |
total_amount | order_total |
SELECT
customer_uuid, -- was customer_id in source
email_address, -- was email in source
fname, -- was first_name in source
order_total -- was total_amount in source
FROM {{ source('raw', 'public__customers') }}
Exclude
Drop a column entirely. It will not appear in the raw layer or downstream.
Common reasons to exclude:
- Auto-generated columns (
serial_col, bigserial_col) that are meaningless in the destination
- PII columns you do not want to land in the warehouse at all
- Large columns (e.g. raw
description TEXT) you plan to truncate in dbt instead
Excluded columns are silently omitted — no error is raised if a downstream query references them.
- Click ⚙️ on the Destination node in the canvas.
- Click the Normalisation tab.
- For each source column, either:
- Enter a new name in the Rename to field, or
- Tick Exclude to drop the column.
- Columns with no rule pass through unchanged under their original name.
- Save the destination configuration.
Naming convention for the raw layer
MantrixFlow stores normalised rows in a raw schema using this table name pattern:
raw.{source_schema}__{table_name}
| Source table | Raw layer table |
|---|
public.customers | raw.public__customers |
public.orders | raw.public__orders |
public.order_items | raw.public__order_items |
sales.transactions | raw.sales__transactions |
inventory.products | raw.inventory__products |
__) separates schema from table name. This is the reference you use in the dbt Layer {{ source() }} call.
Real-world normalisation examples
Customers table
| Rule | Source column | Target |
|---|
| Rename | customer_id | customer_uuid |
| Rename | email | email_address |
| Rename | first_name | fname |
| Rename | last_name | lname |
| Rename | date_of_birth | birth_date |
| Exclude | phone | — |
| Exclude | metadata | — (extracted in dbt) |
Orders table
| Rule | Source column | Target |
|---|
| Rename | order_id | order_uuid |
| Rename | customer_id | customer_ref |
| Rename | order_number | order_code |
| Rename | status | order_status |
| Rename | total_amount | order_total |
| Exclude | items | — (extracted in dbt) |
| Exclude | shipping_address | — (extracted in dbt) |
| Exclude | billing_address | — (extracted in dbt) |
Numeric types table
| Rule | Source column | Target |
|---|
| Rename | small_int_col | small_int_value |
| Rename | int_col | int_value |
| Rename | big_int_col | big_int_value |
| Rename | decimal_col | decimal_value |
| Rename | numeric_col | numeric_value |
| Rename | real_col | real_value |
| Rename | double_col | double_value |
| Rename | money_col | money_value |
| Exclude | serial_col | — (auto-generated) |
| Exclude | bigserial_col | — (auto-generated) |
| Approach | Best for |
|---|
| Normalisation tab | Clean column naming before dbt runs; excluding columns you never want in the warehouse |
| SQL Transform node | Full SQL reshaping, casts, aggregations, derived columns, filtering before the raw layer |
| Both together | Normalise first, then write richer dbt logic on the clean column names |
