Book a Demo

Test the same freight reference from two customers

Format: Company perspective.

A matching freight reference is not enough to merge two loads.

Before a TMS import, try this hypothetical test with made-up records. Create two customers, Account Cedar and Account Harbor. Give each a load with the outside reference REF-77. Use different internal load IDs and delivery sheets. These are two loads, even though the customers chose the same reference.

Expect two separate loads. A test user with rights to both accounts should find both by searching REF-77. The search must not join their files, terms, or history. A user with rights to only one account must not see the other account's load.

Define a compound business key: fields used together to tell records apart. Try the owning business, customer account ID, and customer load reference. Check whether that set is unique in your data. If a customer reuses references, keep the source record's stable ID or add an agreed field. Do not strip punctuation or guess a date to force a match.

Keep this evidence beside the test:

  • The two source rows and their different internal IDs.
  • The proposed matching rule.
  • The new record for each row.
  • The file linked to each new record.
  • A search result that shows both loads, still separate.
  • What a user with limited rights can see. Check that the other account's data stays hidden.

Then change just the spelling of a customer name. Keep its account ID the same. Does the import create a second load by mistake? That is a different fault from merging two customers' loads. Run both tests and keep their results separate.

Fail the test if one record replaces the other. Fail it if a file ends up in the wrong account too. Save the rule that failed. Fix it, then run the checks again.

This Core Agentic Desk Check is a proposed test. No real import was run. Bring the IDs and matching rules your team uses to discuss a custom TMS data move with Core Agentic. Book a Demo.

Sources and scope

Core Agentic TMS and design context. The operating steps and fictional examples are first-party recommendations, not verified installed controls or customer results.