Two-Way Sync Field Mapping Workbook

Separate identity, ownership, and transformation

Use one row per field relationship. Keep identity separate from the value being synchronized: a company can change its domain or name without becoming a new customer. Store source and destination record IDs, and document how the original pairing was established. A deterministic key is especially important when a create succeeds but its acknowledgment is lost.

A two-way integration does not require every field to be writable in both directions. For example, finance can own credit status while sales owns account notes. Use bidirectional rules only for fields where both applications are intended editing surfaces. This reduces conflicts and makes incident resolution easier.

Work through a customer example

Consider a CRM, an ERP, and a database used by an internal application. The example workbook uses a persistent customer ID for matching, an ERP-owned credit status, a shared contact phone number, and a separate contact-to-company relationship. These are proposed design choices, not default behavior of Stacksync or either vendor.

• For the customer key, reject missing values instead of creating a second entity with a guessed match.
• For credit status, translate internal enum values through an explicit lookup; reject unknown codes.
• For phone number, distinguish an omitted field from an intentional clear. Define what simultaneous edits should do.
• For relationships, identify the contact, company, and relationship label independently. Removing an association must not automatically delete its records.

A lossy conversion needs special treatment. If the CRM stores a date while the database stores a timestamp, writing the date back cannot recover the original time. Make that field one-way, preserve the original value separately, or document the accepted loss. Do not call a reversible-looking pair of arrows a reversible data transformation.

Turn each row into a test

For each mapping, create a normal case, a null case, an invalid value, a repeated delivery, and—if both sides write—a simultaneous edit. Record the observed values in both systems. A screenshot of a connected application does not establish that field ownership or duplicate prevention works.

• 01Create a test entity and record both systems’ identifiers.
• 02Update the owner-controlled field and verify only the intended destination changes.
• 03Send the same logical change again and confirm no extra entity or business action appears.
• 04Clear the value explicitly, then omit the field in a later update; verify the two operations follow different rules if specified.
• 05Change both sides before the next capture cycle and inspect the result.
• 06Attach evidence and the business owner’s acceptance to the row.

Keep mapping changes versioned. When an administrator renames a field or changes a picklist, review every affected row and rerun its tests. Stacksync’s configuration-update guide describes the product workflow; the workbook records why the new mapping is correct.