Hello Fivetran Team,

As described in your documentation, incremental syncs rely on reliable timestamp fields to detect changes, and for Salesforce you currently use one of the following fields:

• SystemModStamp
• LastModifiedDate
• CreatedDate
• LoginTime

If none of these fields are available, the table is fully re-imported.

Problem

In practice, this approach has key limitations:

1. SystemModStamp sometimes changes without meaningful business changes.
We see cases where SystemModStamp is updated even though no business-relevant fields have changed. These rows are still treated as updates by Fivetran, which increases sync volume and cost without adding real value. It is unclear what internal Salesforce operations trigger these changes.

2. Different objects require different incremental fields.
Our Salesforce administrator recommends using LastModifiedDate for some objects and SystemModStamp for others, depending on how those objects are maintained and updated in Salesforce. However, today this choice is not configurable, which makes it hard to align Fivetran’s behavior with Salesforce best practices for each object.

Because of this, it is difficult to trust that incremental syncs always represent meaningful business changes.

Feature Request

We would like to request one of the following:

Option 1: Configurable Incremental Column
Allow users to configure which date/datetime field is used as the incremental sync base field for each Salesforce object. This would let teams choose the field that best reflects real business changes and follow Salesforce-admin recommendations per object.

Option 2: Full Transparency of Incremental Logic
If configurability is not possible in the short term, please document and share the exact logic behind your incremental sync behavior, including:

• How the incremental field is selected for each object

• Priority order between SystemModstamp, LastModifiedDate, CreatedDate, and LoginTime

• Known edge cases and limitations

Why This Matters

• System-level timestamp changes can cause unnecessary re-syncs.

• Different objects require different change-tracking strategies.

• The current approach can lead to much higher cost.

Providing either configurability or clear documentation would greatly improve trust, accuracy, and usability of the Salesforce connector in real-world data pipelines.

Thank you for considering this request.