Salesforce Source dbt Package (Docs)
- Cleans, tests, and prepares your Salesforce data from Fivetran's connector for analysis.
- Generates a comprehensive data dictionary of your Salesforce data via the dbt docs site
- Materializes staging tables which leverage data in the format described by this ERD and is intended to work simultaneously with our Salesforce modeling package
- Refer to our Docs site for more details about these materialized models.
- Optional: You can also bring in Salesforce history mode models utilizing Fivetran's History Mode.
- Connector: Have the Fivetran Salesforce connector syncing data into your warehouse.
- Database support: This package has been tested on Postgres, Databricks, Redshift, Snowflake, and BigQuery.
If you are using a Databricks destination with this package you will need to add the below (or a variation of the below) dispatch configuration within your dbt_project.yml
. This is required in order for the package to accurately search for macros within the dbt-labs/spark_utils
then the dbt-labs/dbt_utils
packages respectively.
dispatch:
- macro_namespace: dbt_utils
search_order: ['spark_utils', 'dbt_utils']
The history models in this package are materialized incrementally. We have chosen insert_overwrite
as the default strategy for BigQuery and Databricks databases, as it is only available for these dbt adapters. For Snowflake, Redshift, and Postgres databases, we have chosen delete+insert
as the default strategy.
insert_overwrite
is our preferred incremental strategy because it will be able to properly handle updates to records that exist outside the immediate incremental window. That is, because it leverages partitions, insert_overwrite
will appropriately update existing rows that have been changed upstream instead of inserting duplicates of them--all without requiring a full table scan.
delete+insert
is our second-choice as it resembles insert_overwrite
but lacks partitions. This strategy works most of the time and appropriately handles incremental loads that do not contain changes to past records. However, if a past record has been updated and is outside of the incremental window, delete+insert
will insert a duplicate record.
Because of this, we highly recommend that Snowflake, Redshift, and Postgres users periodically run a
--full-refresh
to ensure a high level of data quality and remove any possible duplicates.
If you are not using the Salesforce transformation package, include the following salesforce_source
package version in your packages.yml
Check dbt Hub for the latest installation instructions, or read the dbt docs for more information on installing packages.
packages:
- package: fivetran/salesforce_source
version: [">=1.1.0", "<1.2.0"] # we recommend using ranges to capture non-breaking changes automatically
By default, this package will run using your target database and the salesforce
schema. If this is not where your Salesforce data is (perhaps your Salesforce schema is salesforce_fivetran
), add the following configuration to your root dbt_project.yml
file:
vars:
salesforce_database: your_database_name
salesforce_schema: your_schema_name
It is possible that your Salesforce connector does not sync every table that this package expects. If your syncs exclude certain tables, it is because you either don't use that functionality in Salesforce or actively excluded some tables from your syncs.
To disable the corresponding functionality in this package, you must add the corresponding variable(s) to your dbt_project.yml
, which are listed below. By default, that is if none of these variables are added, all variables are assumed to be true. Add variables only for the tables you would like to disable:
vars:
salesforce__user_role_enabled: false # Disable if you do not have the user_role table
salesforce__lead_enabled: false # Disable if you do not have the lead table
salesforce__event_enabled: false # Disable if you do not have the event table
salesforce__task_enabled: false # Disable if you do not have the task table
salesforce__opportunity_line_item_enabled: false # Disable if you do not have the opportunity_line_item table
salesforce__order_enabled: false # Disable if you do not have the order table
salesforce__product_2_enabled: false # Disable if you do not have the product_2 table
The corresponding metrics from the disabled tables will not populate in the downstream models.
If you have Salesforce History Mode enabled for your connector, we now include support for the account
, contact
, and opportunity
tables directly. This will allow you access to your historical data for these tables while taking advantage of incremental loads to help with compute.
Customers leveraging the Salesforce connector generally fall into one of two categories when taking advantage of History mode. They either have one connector that is syncing non-historical records and a separate connector that syncs historical records, or they have one connector that is syncing historical records. We have designed this feature to support both scenarios.
If you are gathering data from both standard Salesforce as well as Salesforce History Mode, and your target database and schema differ as well, you will need to add an additional configuration for the history schema and database to your dbt_project.yml
.
vars:
salesforce_database: your_database_name # salesforce by default
salesforce_schema: your_schema_name
salesforce_history_database: your_history_database_name # salesforce_history by default
salesforce_history_schema: your_history_schema_name
Perhaps you may only want to use the Salesforce History Mode to bring in your data. Because the Salesforce schema is pointing to the default salesforce
schema and database, you will want to add the following variable into your dbt_project.yml
to point it to the salesforce_history
equivalents.
vars:
salesforce_database: your_history_database_name # salesforce by default
salesforce_schema: your_history_schema_name
salesforce_history_database: your_history_database_name # salesforce_history by default
salesforce_history_schema: your_history_schema_name
IMPORTANT: If you utilize Option 2 and are also using the dbt_salesforce
package, you must sync the equivalent enabled tables and fields in your history mode connector that are being brought into your end reports. Examine your data lineage and the model fields within the salesforce
folder to see which tables and fields you are using and need to bring in and sync in the history mode connector.
The History Mode models can get quite expansive since it will take in ALL historical records, so we've disabled them by default. You can enable the history models you'd like to utilize by adding the below variable configurations within your dbt_project.yml
file for the equivalent models.
# dbt_project.yml
...
vars:
salesforce__account_history_enabled: true # False by default. Only use if you have history mode enabled and wish to view the full historical record of all your synced account fields.
salesforce__contact_history_enabled: true # False by default. Only use if you have history mode enabled and wish to view the full historical record of all your synced contact fields.
salesforce__opportunity_history_enabled: true # False by default. Only use if you have history mode enabled and wish to view the full historical record of all your synced opportunity fields.
Daily account, contact and opportunity history tables that are created from these history tables are available in our dbt_salesforce
package.
By default, these models are set to bring in all your data from Salesforce History, but you may be interested in bringing in only a smaller sample of historical records, given the relative size of the Salesforce History source tables.
We have set up where conditions in our staging models to allow you to bring in only the data you need to run in. You can set a global history filter that would apply to all of our staging history models in your dbt_project.yml
:
vars:
global_history_start_date: 'YYYY-MM-DD' # The first `_fivetran_start` date you'd like to filter data on in all your history models.
If you'd like to apply model-specific conditionals, configure the below variables in your dbt_project.yml
:
vars:
account_history_start_date: 'YYYY-MM-DD' # The first date in account history you wish to pull records from, filtering on `_fivetran_start`.
contact_history_start_date: 'YYYY-MM-DD' # The first date in contact history you wish to pull records from, filtering on `_fivetran_start`.
opportunity_history_start_date: 'YYYY-MM-DD' # The first date in opportunity history you wish to pull records from, filtering on `_fivetran_start`.
To ensure maximum value for these history mode models and avoid messy historical data that could come with picking and choosing which fields you bring in, all fields in your Salesforce history mode connector are being synced into your end staging models. That means all custom fields you picked to sync are being brought in to the final models. See our DECISIONLOG for more details on why we are bringing in all fields.
To update the history mode models, you must follow these steps:
- Go to your Fivetran Salesforce History Mode connector page.
- Update the fields that you are bringing into the model.
- Run a
dbt run --full-refresh
on the specific staging models you've updated to bring in these fields and all the historical data available with these fields.
We are aware that bringing in additional fields will be very process-heavy, so we do emphasize caution in making changes to your history mode connector. It would be best to batch as many field changes as possible before executing a --full-refresh
to save on processing.
By default, this package builds the Salesforce staging models within a schema titled (<target_schema> + _stg_salesforce
) in your target database. If this is not where you would like your Salesforce staging data to be written to, add the following configuration to your root dbt_project.yml
file:
models:
salesforce_source:
+schema: my_new_schema_name # leave blank for just the target_schema
If an individual source table has a different name than expected, provide the name of the table as it appears in your warehouse to the respective variable:
IMPORTANT: See this project's
dbt_project.yml
variable declarations to see the expected names.
# dbt_project.yml
...
config-version: 2
vars:
salesforce_<default_source_table_name>_identifier: your_table_name
If you do not use the default all-caps naming conventions for Snowflake, you may need to provide the case-sensitive spelling of your source tables that are also Snowflake reserved words.
In this package, this would apply to the ORDER
source. If you are receiving errors for this source, include the below identifier in your dbt_project.yml
file:
vars:
salesforce_order_identifier: "Order" # as an example, must include the double-quotes and correct case.
The source tables Fivetran syncs do not include formula fields. If your company uses them, you can generate them by referring to the Salesforce Formula Utils package. To pass through the fields, add the latest version of the package. We recommend confirming your formula field models successfully populate before integrating with the Salesforce package.
Include the following within your dbt_project.yml
file:
# Using the opportunity source table as example, update the opportunity variable to reference your newly created model that contains the formula fields:
salesforce_opportunity_identifier: "my_new_opportunity_formula_table"
# In addition, add the desired field names as pass through columns
salesforce__opportunity_pass_through_columns:
- name: "salesforce__opportunity_field"
alias: "opportunity_field_x"
This package includes all source columns defined in the generate_columns.sql
macro. You can add more columns using our passthrough column variables. These variables allow for the passthrough fields to be aliased (alias
) and casted (transform_sql
) if desired, but not required. Datatype casting is configured via a sql snippet within the transform_sql
key. You may add the desired sql while omitting the as field_name
at the end and your custom pass-though fields will be casted accordingly. Use the below format for declaring the respective pass-through variables:
# dbt_project.yml
...
vars:
salesforce__account_pass_through_columns:
- name: "salesforce__account_field"
alias: "renamed_field"
transform_sql: "cast(renamed_field as string)"
salesforce__contact_pass_through_columns:
- name: "salesforce__contact_field"
alias: "contact_field_x"
salesforce__event_pass_through_columns:
- name: "salesforce__event_field"
transform_sql: "cast(salesforce__event_field as int64)"
salesforce__lead_pass_through_columns:
- name: "salesforce__lead_field"
salesforce__opportunity_pass_through_columns:
- name: "salesforce__opportunity_field"
alias: "opportunity_field_x"
salesforce__opportunity_line_item_pass_through_columns:
- name: "salesforce__opportunity_line_item_field"
alias: "opportunity_line_item_field_x"
- name: "field_name_2"
salesforce__order_pass_through_columns:
- name: "salesforce__order_field"
alias: "order_field_x"
- name: "another_field"
alias: "field_abc"
salesforce__product_2_pass_through_columns:
- name: "salesforce__product_2_field"
alias: "product_2_field_x"
salesforce__task_pass_through_columns:
- name: "salesforce__task_field"
alias: "task_field_x"
salesforce__user_role_pass_through_columns:
- name: "salesforce__user_role_field"
alias: "user_role_field_x"
salesforce__user_pass_through_columns:
- name: "salesforce__user_field"
Fivetran offers the ability for you to orchestrate your dbt project through the Fivetran Transformations for dbt Core™ product. Refer to the linked docs for more information on how to setup your project for orchestration through Fivetran.
This dbt package is dependent on the following dbt packages. For more information on the below packages, refer to the dbt hub site.
If you have any of these dependent packages in your own
packages.yml
I highly recommend you remove them to ensure there are no package version conflicts.
packages:
- package: fivetran/fivetran_utils
version: [">=0.4.0", "<0.5.0"]
- package: dbt-labs/dbt_utils
version: [">=1.0.0", "<2.0.0"]
- package: dbt-labs/spark_utils
version: [">=0.3.0", "<0.4.0"]
The Fivetran team maintaining this package only maintains the latest version of the package. We highly recommend you stay consistent with the latest version of the package and refer to the CHANGELOG and release notes for more information on changes across versions.
These dbt packages are developed by a small team of analytics engineers at Fivetran. However, the packages are made better by community contributions.
We highly encourage and welcome contributions to this package. Check out this post on the best workflow for contributing to a package.
- If you encounter any questions or want to reach out for help, see the GitHub Issue section to find the right avenue of support for you.
- If you would like to provide feedback to the dbt package team at Fivetran, or would like to request a future dbt package to be developed, then feel free to fill out our Feedback Form.