After receiving the notification about Atlassian's new automation pricing, I am on a mission to make my automations more efficient so we don't exceed our limit. I am a bit unclear on how Atlassian is defining a "step" as it relates to conditions. If a condition (yellow) step/node has multiple criteria that are connected by ANDs/ORs (JQL or otherwise), does that count as a single step or one step per criteria? Similarly, within a trigger step, does each criteria/condition within that trigger count as its own step or is the entire trigger counted as a single step regardless of the number of criteria within it?
I found this article to be somewhat helpful but it did not fully answer my question regarding conditions: https://community.atlassian.com/forums/Jira-Cloud-Admins-articles/Automation-steps-start-billing-on-3-December-Your-allowance-what/ba-p/3285795
Does anyone have any firm information on this topic?
@Dayna Portillo ,
@Gor Greyan has the principle right and I want to add the source and the two cases it does not cover, because the difference between "I think" and "the doc says" matters when you are budgeting for December.
The unit is the component that executes, not the criteria inside it. Atlassian's definition: a step is a trigger, a condition, an action, an if/else branch, or one iteration of a loop, counted when that component finishes with a charged status (how usage is calculated). So:
A failing condition still charges for itself. The charged statuses are SUCCESS, NO_ACTIONS_PERFORMED, SOME_ERRORS, ABORTED and NO_MATCH. A rule that fires on "Work item updated" and fails its condition 95 percent of the time pays one trigger step plus one condition step on every rejected event. That is the single largest source of unexpected steps I have seen, and the fix is a narrower trigger (Field value changed for the specific field) rather than a broader trigger with a condition behind it.
SUCCESS
NO_ACTIONS_PERFORMED
SOME_ERRORS
ABORTED
NO_MATCH
Not charged: system and synthetic components Atlassian runs on your behalf, THROTTLED runs where nothing executed, and system-side errors. Rovo agent actions inside rules bill as Rovo credits, not steps.
THROTTLED
Verify it on your own rules rather than take anyone's word, including mine: open a space's automation list, More actions, View performance insights (or the global automation page for every space), which shows executions per rule and their outcomes (performance insights). It counts executions, not steps, so multiply by the number of components that ran; for a rule whose condition fails most of the time that is executions times two. Compare the total with the Automation steps meter in Atlassian Administration, Insights, Platform usage, and the arithmetic will tell you whether your reading matches the meter's.
@Trudy P Claspill link is the same page; the section titled "What counts as a step" is the one to bookmark. I wrote up the allowance table, the charged statuses and the five rule shapes that burn steps in one place yesterday, if the long version is useful: Automation steps start billing on 3 December.
Hello @Dayna Portillo
Take a look at this document that comes from Atlassian.
https://support.atlassian.com/cloud-automation/docs/how-is-my-usage-calculated/
I believe the relevant section is the one that starts with "A step is charged if it's execution results..."
Hi @Dayna Portillo
@Trudy P Claspill has already provided the correct Atlassian documentation.
To clarify your specific question, Atlassian counts the Automation component that executes, not each individual criterion inside that component.
So, if one condition contains multiple criteria connected with AND/OR, it is still one condition step, not one step per criterion. The same principle applies to criteria configured within a trigger.
However, if your rule has multiple separate Condition components, each condition that executes counts as its own step. Actions, branches, and loops are similarly counted when executed.
So for optimization, it can make sense to combine related conditions where practical and place restrictive conditions early in the rule to prevent unnecessary downstream steps from executing.Regards,Gor
I appreciate your thoughtful and thorough response! I was hopeful but didn't want to assume. I read the documentation closely but it was never stated explicitly, and the word "conditions" can be a bit vague whether it refers to the entire condition node or each criteria within (I often use the word "conditions" in place of "criteria"). You're exactly right -- being wrong would be very costly so I needed certainty. I will track my rules in the coming days! Thanks again!
Thanks, Gor!
Hi @Dayna Portillo , Sami's answer settles the counting, trigger criteria included. Since you mentioned tracking your rules, a shorter way to see steps per rule on some sites, and two rewrites the thread has not spelled out yet.
Steps per rule, without the multiplication. Sami's route, performance insights times the components that ran, works on every site. On some sites the Usage tab (Jira settings → System → Global automation → Usage) now also has an Upcoming usage model view with estimated steps per rule (Jira calls rules "flows" there) and a list of the highest-usage ones. If yours shows it, the multiplication is done for you; if it shows runs only, Sami's arithmetic is still the way.
Two rewrites the thread has not spelled out:
priority = High AND "Request type" = Incident AND labels = vip
To check that a rewrite worked, compare the rule's number for a week before and a week after: in the Usage tab if it shows steps, otherwise executions times components as Sami describes. A rule whose condition rejects most events should drop to roughly two steps a run.
One more option, disclosure: I'm part of the team behind Automation Map for Jira. It keeps itself current: a daily job of your own — a ScriptRunner or JMWE job inside Jira, or the Docker image or a script on your side — pushes the rules, this month's usage and the audit log (Advanced; usage and runs are beta). The job reads with your own token, the app only sees a key it issued, and nothing changes in Jira. It flags exactly these shapes rule by rule: conditions in a row that one JQL condition would replace, a condition placed after an action, a broad trigger with a narrow condition behind it. Each finding says how many steps per run it saves. https://marketplace.atlassian.com/apps/503651487
One rule: 9 steps if all of it runs, 2 after the fixes, each fix listed (demo data).
The Findings tab: each finding with the steps it saves (demo data).
It looks like you're new here. Sign in or register to get started.