JSM automation design and test plan
Jira Automation for Customer Votes: Flag Demand Without Spam
Use Wezy vote fields in a scheduled Jira Automation rule that adds a review marker, preserves other labels, and avoids treating demand as incident priority.
In short
Start with a scheduled query and one reversible action: add a demand-review label. Check the rule actor, preserve existing labels, and test repeated runs before adding notifications or other side effects.
In this guide
Use customer-vote thresholds to flag a request for human review, not to automatically promote it to an urgent incident. A safe first integration is a scheduled Jira Automation rule that adds a dedicated label to eligible requests. It can be disabled and the label can be removed without changing votes, request sharing or the customer’s original conversation.
This is a proposed integration using native Jira Automation and Wezy’s read-only demand fields. Wezy does not install this rule for you, and this guide does not claim that the rule has been run in your tenant.
Define the contract before the trigger
Use a narrow contract that someone can verify:
On this project, unfinished requests with at least five published active customer votes should receive the label
wezy-demand-review-v1. A reviewer decides the next action. The rule must not change priority, notify customers, edit vote fields or replace existing labels.
Five votes and the label are illustrative configuration choices. They do not imply a statistically reliable threshold or a built-in Wezy workflow. The label marks that the request qualified for this review cycle; it is not a live assertion that the vote count will remain above five forever.
This distinction avoids a noisy loop in which a request is repeatedly marked and unmarked as support changes.
Choose a scheduled query rather than assuming a vote event
Atlassian’s Scheduled trigger can run actions on work items selected by JQL. That is a defensible starting point because Wezy distinguishes the customer’s vote confirmation, publication of Jira fields and Jira indexing. There is no documented instant “customer voted” trigger promised by this article.
For a limited pilot, choose a cadence matching the team’s review needs and available Automation allowance. Do not begin with a high-frequency scan of every project.
project = "YOUR_PROJECT_KEY"
AND statusCategory != Done
AND cf[ID_CUSTOMER_VOTES] >= 5
AND (labels IS EMPTY OR labels NOT IN ("wezy-demand-review-v1"))
Replace the project key and custom-field placeholder first. The explicit empty-label branch matters: a negative JQL condition alone can miss empty fields. See Atlassian’s operator reference and the Wezy JQL queue guide for field mapping.
The query scopes unfinished work, not necessarily currently open voting. Add the validated business restrictions your review requires; never infer Wezy enrollment from statusCategory alone.
Configure one reversible write
Create the rule with a project administrator and inspect its actor. The Automation actor’s permissions govern what it can see and change. A successful search in your own browser does not prove that this actor has equivalent access.
After the Scheduled trigger, use the available re-fetch action before conditions where appropriate. Recheck that the review marker is absent and the currently available vote value meets the threshold. This reduces avoidable stale decisions; it does not make the rule transactional or bypass Jira’s publication/indexing behavior.
Use Edit work item to add the label. Preserve all existing labels rather than setting the entire label list to a single value. Atlassian documents the supported edit and re-fetch actions in its Automation actions reference.
Keep the first version free of comments, emails, external web requests and priority transitions. Those add failure and duplication modes without being necessary to create a review queue.
Test both the first run and the second run
Prepare a small authorized set with published values. Do not change app-managed Wezy fields to manufacture the fixtures; use the documented customer journey or select suitable existing test records.
| Case | Expected observation |
|---|---|
| Four active votes, no label | Not marked |
| Exactly five active votes, no label | Marker added |
| Five votes and an unrelated label | Marker added; the unrelated label remains |
| Five votes and the marker already present | No new marker action required |
| No published vote value | Not silently treated as five |
| Completed request | Excluded by this policy |
| Actor cannot access a request | No unauthorized change; inspect the rule result |
| Support falls after a successful marking | Marker remains as a review record until the team resolves it |
Run again after publication and indexing have converged. Check both the work item and the Automation audit log. Record the rule version, actor, request key, timestamp and observed actions. These are acceptance tests to perform, not completed customer test results.
Locate a failure before changing the rule
Use the audit log to identify the first stage that did not match the contract:
| Stage | Diagnostic question | Controlled response |
|---|---|---|
| Scheduled selection | Did the tested JQL select the expected accessible request? | Check actor, scope, field ID and publication/indexing separately |
| Recheck condition | Does the request still meet the threshold without the marker? | Stop the action when the current check fails |
| Label edit | Was the marker added while unrelated labels survived? | Inspect edit permission and action configuration |
| Later execution | Was the already marked request avoided or left unchanged? | Investigate delayed selection or parallel runs before adding side effects |
Re-fetch refreshes the work-item data used by the flow; it is not a command that forces Jira’s search index to finish updating. A failed edit should remain a failed test even when the initial query looked correct. Keep the first version’s one reversible action until all four stages are understandable.
“Idempotent outcome” is not “exactly one execution”
Adding the same label does not create a second distinct label value. That makes the chosen outcome easier to tolerate on retry. It does not guarantee one Automation execution, one billable operation or an atomic check-and-write sequence. Concurrent executions or delayed search updates can still revisit the same work item.
This is why the first version uses a marker rather than sending a message every time the threshold matches. Before introducing notifications, decide how to reconcile a partial run: a message might be delivered even if a later marker update fails, or vice versa. Do not advertise exactly-once delivery without a separate mechanism that actually provides it.
Operate the rule as a review aid
Create a human-owned queue for the review label. Record whether a request is being investigated, deferred or declined in the team’s chosen process. Avoid deleting the marker automatically merely because the score drops; it would become eligible again and restart the cycle.
To end a campaign, disable the rule first, inspect its recent executions, then decide how its markers should be retained or removed. A new review cycle can use an explicitly new marker after the team agrees the scope. This is deliberate re-evaluation, not an accidental loop.
Wezy Customer Voting supplies customer support and organization-demand signals inside company-managed JSM Cloud projects. Keep service priority and product demand separate: a customer-voting integration should make a decision easier to review, not make that decision invisibly.
Published by the team behind Wezy Customer Voting. Product capabilities are linked to their official documentation; worked examples are illustrative.
Let customers vote on your JSM portal.
Install Wezy from Atlassian Marketplace, open one request for voting and see the first vote in Jira.