Hi @cpierson90 ,
This is a common queueing problem that you come across. There are many ways to trigger an automation:
- If a record meets a series of conditions - This is one of the more straight forward approaches and fastest to implement. However, what do you do when one fails? How do you cause the automation to run against this record again.
- If a record enters a view -- In this case, you can create filters on the view which mirror what you have in method 1. By changing the criteria, you can quickly cause the records to no longer meet the criteria and fall out of the view. Change it back and they will meet the criteria again. However, this also does not resolve the issue of how to reprocess things that didn't finish with out your intervention.
- Use a combination of 1 or 2, a checkbox and an automation based on a timer. This is an approach I have been using in my larger bases where I need to process a lot of things over time. Every 15 minutes, the timer runs and searches for records that meet the criteria for method 1. For those records, check the box that runs the automation. This method limits you to 400 records an hour due to the maximum of 100 records being returned in the find record automation. If you need more than that, you can scheduled more 15 minute automations off set by 5 minutes, essentially upping that number to 1200 records per hour.
Some advantages of method 3
- If I want to manually reprocess a record I can do it by checking the box.
- I can expose the checkbox in an interface and not have to be at the data or automation layer to retrigger the automation
- I can create a reporting interface that shows me how many items are checked still needing processing.
- Records that caused an automation to fail will be picked up and reprocessed once the data is fixed.
You can do something similar but more sophisticated if you use a single select. Set the single select to "START" in the 15 minute automation. Then in your main automation set it to "IN PROGRESS" at the start and "COMPLETE" at the end. If you have a bunch stuck in "IN PROGRESS" you'll know that the automation didn't complete for those records. You can add another feature to measure the time that a record went into "IN PROGRESS". If is is more the 2x -3x longer than what you expected show that in your interface.
