Defining fairness for the operation
Two workers can have the same number of scheduled hours and very different weeks. One may receive most of the late changes, long journeys or unpopular shifts. A utilization average will not show that difference.
For this article, scheduling fairness means defining and reviewing how assignments affect workers, including workload, predictability, preferences and access to opportunities. No single measure establishes fairness across every workforce.
1. What the schedule optimizes
Optimization follows the objectives and constraints it receives. Review whether those settings account for the working conditions the organization intends to provide, as well as service coverage and cost.
"The technology isn't falling short. It's doing precisely what it was configured to do. The shortfall is upstream — in the success metrics leaders chose."
Mark Graham — Senior Product Manager, Skedulo
A schedule that fills every available minute may leave too little allowance for realistic travel, variable job durations and required breaks. Those allowances need to be represented in the model, with mandatory requirements protected.
Repeated reallocation can also create disruption even when each change reduces travel. Consider schedule stability alongside efficiency, and review who receives the changes over time.
"Add slack as a deliberate constant. Add predictability as a goal alongside utilization. Treat the schedule like something a human will live inside — not just a problem to solve."
Mark Graham — Skedulo
Agree the priorities with operations, workforce representatives and the teams responsible for policy. Define what must be protected and what can be traded off, then check whether the resulting schedules reflect that decision.
2. Measuring how work is distributed
Choose measures that fit the workforce rather than treating equal hours as a complete definition. Contracts, availability, qualifications and worker preferences affect how assignments should be compared.
Examine total hours, travel, late changes and the distribution of less-preferred work over an appropriate period. A weekend assignment may be welcome to one worker and difficult for another, so recorded preferences matter.
"Policy alone fails because it's a post-hoc constraint. By the time HR notices the imbalance, three months of weekends have been stacked on the same ten people, and the damage is done."
Mark Graham — Skedulo
UKG's 2025 study found that 50% of respondents had difficulty changing a shift for personal reasons. An allocation review should therefore consider the process for requesting a change as well as the original schedule. Have the responsible specialists confirm any local requirements before configuring policies.
3. Explaining and reviewing decisions
Workers need to understand the relevant allocation criteria and know where to raise a concern. An explanation should identify the information and rules used without disclosing another worker's private information.
"Fairness without transparency is just policy with extra steps. The tools needed are fairness objectives in the platform, and decision explanations on the surface."
Mark Graham — Skedulo
Skedulo's "Building Technology for Frontline Workers" research found associations between sufficient technology and reported autonomy, satisfaction and intentions to stay. The figures below do not show that a particular allocation rule causes retention, or establish that a schedule is fair.
- 42% — more likely to say their job is very autonomous
- 44% — more likely to be very satisfied with their role
- 35% — more likely to plan to stay for the next five years
Source: Skedulo, "Building Technology for Frontline Workers"
4. The role of AI assistance
Mastermind supports scheduling optimization against configured objectives and constraints. Tempo can help schedulers investigate issues using the data and tools available to it. Neither product name should be treated as a guarantee of fair outcomes.
Useful evaluation cases for AI assistance include:
- Investigating an assignment: Can the tool retrieve the relevant records and criteria, and distinguish evidence from an inferred explanation?
- Reviewing a proposed rule: Can the team assess its effect before approval, including workers whose preferences or circumstances differ?
- Handling a disruption: Can schedulers compare replacement options without repeatedly allocating the disruption to the same people?
Keep responsibility for policy and outcomes with named people. NIST's AI Risk Management Framework treats fairness, privacy, transparency and reliability as related considerations; an explanation alone does not establish that a decision is fair.
5. Testing difficult cases
Test the measures and rules with representative schedules before expanding automated allocation.
If an optimizer and an experienced scheduler disagree, examine the data, requirements and chosen objectives. Either may be missing relevant context. Compare the effect on workers and service commitments rather than assuming one is correct.
Include unavailable skills, conflicting preferences, required breaks, urgent work and a worker challenging an assignment. Define who can approve an exception, how it is recorded and how its effects will be reviewed.
6. Questions for enterprise buyers
Use your own workforce data to assess benefits and costs. A change in retention or engagement may have several causes, so avoid attributing the whole change to scheduling software.
Ask the vendor to demonstrate:
- Distribution of work: How can the team compare hours, travel, less-preferred assignments and changes over time?
- Decision context: Which inputs and rules are available when reviewing an assignment, and what can a worker question?
- Predictability: Can stability, preferences and appropriate time allowances be represented alongside utilization?
- Exceptions: Who approves departures from policy, how are they recorded, and how can their cumulative effects be reviewed?
7. Frequently asked questions
8. Sources
- UKG and Workplace Intelligence. More Perspectives from the Frontline Workforce: findings from the 2025 global study.
- UKG. AI and the Frontline Workforce: reported difficulty changing shifts for personal reasons.
- NIST. Artificial Intelligence Risk Management Framework 1.0.
- NIST. AI Risks and Trustworthiness: fairness, transparency, privacy and reliability.
- Skedulo. Building Technology for Frontline Workers (2022 research).
