Create and Manage Data Products

This article provides you with a step-by-step guide on how to get started with Bobsled Sledhouse to create Data Products.

Prerequisites

• Bobsled Account: You must have access to Bobsled’s main application and the Sledhouse tab/feature enabled in your environment.

• To successfully create a Sledhouse Table, you must have at least one Data source and a Sledhouse Table preconfigured in Bobsled.

NOTE:
What you can see and do will differ based on your role and permissions.

Setup instructions

Step 1: Define your Data Product

1. In the Bobsled Application, locate and select the Data Products Tab, then select the “Define Data Product” button

2. In the wizard, choose how you’d like to define a Data Product.

TIP:
• The Query Designer is built with non-technical users in mind. Currently, it supports referencing only one Sledhouse Table.
• The SQL Editor is for power users, offering full flexibility and control. SQL-based Data Products can reference multiple Sledhouse Tables and even other Data Products.

Option 1: Via Query Designer (no code)

In this mode, Bobsled empowers users to choose to include or exclude certain columns from a sledhouse Table in a data product and use a no-code query build to control the rows returned in the data product. This enables fulfillment teams to control data entitlements without writing any code.

1. Choose a source Sledhouse Table to define a Data Product from and select continue.

2. In the query canvas, optionally select which columns you want to be included or excluded from the Data Product. By default, all columns are included, and you must choose which columns to exclude.

TIP:
• Use the arrow buttons in the middle to quickly move columns between sections. You can also drag and drop columns between the two squares.
• Hovering on object on the left-side panel will display additional metadata.
• To reference a different Sledhouse Table, select the pencil icon (edit) in the Selected Sledhouse Table section. Changing the table will remove any existing customizations.

NOTE:
• In this mode, at least one column must be included for the Data Product to be valid and saved.
• Columns used for partioning display a key icon. Bobsled recommends keeping partioning key columns, and will alert you if you exclude them.

3. Set your Data Product’s schema evolution preferences.

NOTE:
• By default, new columns added to your Sledhouse Table will are placed in the exclude list. To automatically include new columns, enable the toggle.
• Currently, Bobsled only supports additive schema changes. For more information, please contact your account representative.

4. Optionally apply row filtering, using an easy-to-use query builder that allows for defining filters (predicates in an SQL statement) that control what data is returned.

   • This feature will let you quickly define filters like “event_date is in the past 30 days”.

   • Multiple filters can be combined using AND or OR semantics to create almost any data product definition required. For more complex data products, the SQL builder and co-pilot option can be used.

5. Click the Run button to get a 100-row sample of your Data Product and a row count. Once happy with the shape of your Data Product, click save.

NOTE:
• You can define a Data Product while a Sledhouse Table is performing its initial sync.
• However, you won’t be able to preview the query until the sync finishes—the system will let you know when it’s still in progress.

Option 2: Via SQL Editor

1. In the SQL editor canvas, a built-in SQL editor lets you write transformations using Snowflake-compatible syntax.

2. Optionally select a Sledhouse Table to see metadata about it or its schema.

   • Reference objects by their name using {{ sht('sledhouse_table_name') }} syntax, or {{ dp('data_product_name') }}

   • Example:  

     SELECT
       *
     FROM 
       {{ sht('sledhouse_table_name') }}
     WHERE 
       state = 'CA'

2. Optionally use the AI-Assistant (Copilot) and type a natural-language request, and Sledhouse will suggest SQL statements. You can edit or refine these suggestions as needed.

3. Click the Run button to get a 100-row sample of your Data Product and a row count. Once happy with the SQL shape of your Data Product and no errors are observed, click save.

NOTE:
• You can define a Data Product while a Sledhouse Table is performing its initial sync.
• However, you won’t be able to preview the query until the sync finishes—the system will let you know when it’s still in progress.

SQL Data Products limitations

• Column references require explicit schema browsing or copy-paste

  • You can browse the schema in the SQL editor and copy column names into your query.

  • Bobsled does not automatically resolve columns without explicit references.

  • Improvements are planned to make table/column placement easier, but for now, column references must be manually inserted.

• SQL is ANSI standard, but not full platform-specific SQL

  • You can use ANSI SQL operations (joins, WHERE filters, column selections, aggregates).

  • Destination-specific SQL extensions (e.g., Snowflake-only functions) may not work as expected.

  • The current SQL dialect is Snowflake, so consider that while stick to ANSI SQL standard functions.

• Formatting is strict

  • Queries must respect Bobsled’s formatting rules for table references, {{ sht('...') }}.

  • The SQL editor or AI Copilot can help generate correctly formatted queries.

  • Pasting a query directly from Snowflake or another warehouse will usually require formatting adjustments.

• No native schema migration

  • If a Sledhouse Table schema changes, the SQL definition won’t auto-update.

  • You’ll need to review your Data Product definition and backfill if necessary.

SQL Data Products best practices

• Use the Copilot: It can auto-generate joins and queries with the correct formatting.

• Start with small queries: Validate joins and filters on narrow datasets before scaling.

• Paste & clean up: If you have an existing warehouse view, paste it in and use the Copilot to reformat it for Bobsled.

• Keep SQL lightweight: Complex transformations are better handled upstream; Data Products should focus on joins, filters, and projections.

Step 2: Save your Data Product

1. After clicking Save, and if no errors are found, a modal will appear where you can add extra details:

   • Name: A unique, internal name also used to reference the Data Product in SQL.

   • Target details: Set the target schema and target name your consumers will see. These can be overridden later during fulfillment.

   • Description and tags (optional): Internal notes and tags that are not shared with Data Consumers.

2. Select publish. Data Product successfully defined! You can now share it with a Data Consumer or set one up to begin fulfilling your data.

NOTE:
• If a SQL-based Data Product references more than one Sledhouse Table, Bobsled will use the first identified Sledhouse Table in the SQL statement to track changes for incremental replication.

Managing Data Products

On the Data Product page, you can observe the following tabs:

• Settings:

  • This lets you revisit the configuration (e.g., the SQL and source Sledhouse Table) and make changes

  • Backfill a Data Product and all of its fulfillments downstream

  • Delete a Data Product

• Data Preview: An on-demand preview of the Data Product with a 100-row sample and a row count.

• Schema: Displays all columns set for this Data Product

Previewing Data Product data

1. Locate and select the ‘Data Preview’ tab. Bobsled will render a sample and the total rows, and the last time it was refreshed.

2. Optionally select the refresh button to get a fresher preview of the data.

Editing a Data Product

1. Locate and select the ‘Settings’ tab in the Data Product you wish to edit. In the Data Product Configuration section, click the Edit Data Product button.

2. Perform the desired changes to your Data Product definition, click save, and then Publish.

NOTE:
• Bobsled will recommend a backfill when changes to the Data Product definition would benefit from it This ensures all your Data Consumers receive the latest columns and filters without requiring a new fulfillment.

• You can choose to only backfill the Data Product itself, or include any customizations derived from it—such as additional filtering for a specific Data Consumer.

Manually backfilling a Data Product

1. In the Settings tab of the Data Product you want to backfill, select Backfill Data Product.

2. Review the backfill details. You can optionally choose to include any downstream Data Products—fulfilled customizations derived from this Data Product.

3. Select Yes, backfill. Your request will be queued and will start shortly.

TIP:
• While an upstream backfill—triggered at the Sledhouse Table—can also trigger a Data Product backfill, running the backfil at the Data Product level is more focused.
• This is recommended if substantial changes have been made to the Data Product definition or if you want your Data Consumers to receive a fresh copy.

NOTE:
• You can currently observe a Data Product backfill only in the Data Consumer logs for any fulfilled Data Product.

Delete a Data Product

1. Locate the Data Product you wish to delete, and click on the name to enter its detail page.

2. In the ‘settings’ tab, scroll down.

3. Click on the ‘Delete Data Product’ button at the bottom.

NOTE:
• Bobsled prevents the deletion of any Data Product that has any fulfilled Data Consumers by it.

• To proceed with deletion, all related fulfilled Data Products must first be unshared.

• For your convenience, Bobsled lists and links all dependencies to help you manage this process efficiently.