Skip to the guide

JSM worked scoring analysis

Calibrate JSM Demand Scores Before Activating Advanced

Test how Wezy coefficients and organization weights change a JSM ranking. Use a reproducible fixture, inspect rank reversals and record an activation decision.

Wesy November 11, 2026 5 min read

In short

Keep the requests and their demand inputs fixed while testing candidate formulas. Explain which signal changes the ranking, reject accidental double weighting, and activate a saved Advanced configuration only after checking its effect.

Runs on Atlassian
Company-managed JSM Cloud Existing JSM permissions apply
In this guide

Calibrating a demand score means choosing a defensible policy, not finding coefficients that make your preferred request win. Hold the observed requests constant, compare candidate formulas, and explain every important rank reversal before activating a configuration in Wezy Advanced.

The basic organization-weighting guide explains how an organization’s contribution is counted. This article addresses the next decision: whether a proposed weighting policy is stable, understandable and appropriate for the review you intend to run.

Fix the inputs before changing the policy

Wezy Standard uses S = V + 3 × O + 2 × R30. Advanced supports S = a × V + b × O + c × W + d × R30. V is active votes; O is distinct represented organizations; W sums their weights once each; R30 counts eligible active votes first activated in the last 30 days. The score is not a percentage. These definitions come from the official scoring documentation.

Record the project, accessible request set, observation time, edition, formula version and organization-weight policy. During a live comparison, customers may change their support. A later export is not automatically the same input set. For a clean policy experiment, keep a private snapshot or use the synthetic fixture below.

A deliberately conflicting fixture

These are fictional requests and inputs, designed to make trade-offs visible. They are not customer data or evidence that a particular policy improves business outcomes.

Request V O W R30 Situation represented
A 20 1 5 1 Many people from one higher-weight organization
B 8 4 4 2 Support spread across four equally weighted organizations
C 5 3 3 5 Smaller but entirely recent active support

All inputs are internally consistent: recent active votes do not exceed active votes, and represented organizations do not exceed the number of votes. A has one organization weighted 5; B and C use weights of 1. Never multiply W by the number of employees again.

Download the calibration fixture. Its columns are an educational worksheet, not Wezy’s CSV export schema or an import format.

Compare policies that express different choices

Candidate policy Coefficients (a, b, c, d) A B C Leading request
Standard baseline (1, 3, 0, 2) 25 24 24 A
More organizational breadth (1, 5, 0, 1) 26 30 25 B
More weighted organization demand (1, 1, 3, 1) 37 26 22 A
More recent support (1, 1, 0, 5) 26 22 33 C

For example, B under the breadth policy is 8 + 5×4 + 0×4 + 1×2 = 30. C under the recency policy is 5 + 1×3 + 0×3 + 5×5 = 33.

A different winner is not proof that one formula is more accurate. It demonstrates that these policies answer different questions. Ask the decision owner which trade-off is intended before touching the live configuration.

Run four sensitivity checks

1. Remove the signal whose importance you are debating

Set its coefficient to zero in the offline comparison while leaving everything else fixed. If removing W reverses the shortlist, weighted organization demand is doing material work. That deserves an explicit policy explanation, not a vague statement that the new score is “better.”

A weight change has no effect on S when c = 0. It can still change the displayed W. This is a useful diagnostic when an administrator changes weights and expects the ranking to move.

2. Check whether breadth and weight duplicate your intention

If every represented organization has weight 1, then W = O. The two terms become b×O + c×O = (b+c)×O. That may be deliberate, but it is not two independent kinds of evidence.

Similarly, R30 is a subset of V, not a separate population of voters. A recent active vote contributes through both terms when their coefficients are positive. Explain that intentional extra emphasis to reviewers.

3. Test a modest boundary change

Repeat the comparison with one additional active vote, one additional represented organization, or one vote aging out of R30. Change one assumption at a time. A request that changes position under a tiny adjustment should not be described as decisively ahead.

For a tied result, use a secondary review criterion such as assessed effort or uncertainty. Do not add arbitrary decimals solely to manufacture a unique winner.

4. Check scaling and degenerate settings

Multiplying every coefficient by the same positive factor preserves the order when the same inputs are used. A doubled score is not doubled customer demand. All-zero coefficients produce no useful differentiation even if the settings are syntactically accepted.

Wezy accepts coefficients and organization weights from 0 to 10 with at most two decimal places. An individual organization override takes precedence over its segment; a zero weight does not erase V or O. Verify the actual saved values in Advanced settings.

Distinguish arithmetic validation from policy validation

The downloadable fixture lets you recompute every displayed total and rank reversal without a Jira account. Passing that arithmetic check establishes consistency of the example, not that the Wezy implementation or your installation has been tested.

The three cases were deliberately chosen to disagree under different policies. They are not a representative customer sample. Before activation, add authorized examples of the actual decision boundaries your team faces, including tied requests, unattributed votes and a zero-weight organization. Do not manufacture customer data or tune coefficients until one favored feature always wins.

Move from the worksheet to a controlled activation

With the required Advanced edition and project administration permissions, save the intended configuration, use Preview saved ranking, and inspect known requests. The preview may include only the first accessible requests; it is not proof that every request in the project was considered.

Compare the contribution breakdown with your worksheet before selecting Activate Advanced formula. Reread Current scoring afterward. A license upgrade and saved settings alone do not activate the Advanced formula.

Keep a short decision record: policy purpose, coefficients, weight ownership, representative requests checked, known reversals, approver, activation time and review date. Return to Standard score is the documented way to restore the fixed formula; record that as another policy change rather than rewriting history.

Wezy Advanced on Atlassian Marketplace is useful when your team can own this policy explicitly. The output remains a demand signal alongside feasibility and obligations. No calibration can turn votes into a validated revenue forecast or an automatic delivery commitment.

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.

Runs on Atlassian Read the full documentation

The site works without optional cookies. Your choice is saved on this device for 180 days.

Read more about data and Google