Skip to main content
Question

workflow

  • May 7, 2026 1:44 PM UTC
  • 1 reply
  • 32 views

RoyWanyoike
Forum|alt.badge.img+13

How do I build idempotent workflows (so duplicates don’t get created)?

1 reply

Top Step
Forum|alt.badge.img+12
  • Observing
  • Nov 11, 2026 5:10 PM UTC

The main tool is a unique key plus an upsert, so running the same thing twice updates the record it already made instead of adding another one.

 

Pick a key that identifies the record you're about to create, for example the source record's Record ID# plus a dash plus the related record's ID. Store it in a regular text field that the pipeline fills in, and check Unique in the field properties so Quickbase itself refuses a second record with the same value.

 

In the pipeline, use Prepare Bulk Record Upsert with that field as the merge field, add your row with Add a Bulk Upsert Row, and finish with Commit Upsert. If a record with that key exists it gets updated, otherwise it's created.

 

A few other things cause most of the duplicates I see.

 

On New Event fires on adds and updates. If the step should only happen once, set the trigger to Record Created, or limit it to updates where a specific field changed.

 

For checkbox-triggered pipelines, have the pipeline clear the checkbox (or stamp a Processed date) as part of the run, and add a condition that skips records already processed.

 

When you link to an existing record, point at it by Record ID# rather than searching on names or dates, so a rerun always finds the same record.

 

The same approach works through the JSON API, where the upsert call takes a merge field ID.

 

https://developer.quickbase.com/operation/upsert

 

Want more detail? I wrote a full walkthrough here: https://www.topstep.tech/blog/quickbase-pipelines-prevent-duplicate-records