Understand waiting time, workload pressure and staffing trade-offs.
Model an eight-hour service shift using reproducible arrivals and service times. Compare staffing choices before making real operational decisions.
Simulated alternatives at the selected SLA target
| Agents | SLA met | Avg. wait | p90 wait | Labor / shift | Status |
|---|
Stress-test a queue with common-seed comparisons across staffing levels.
Scenario presets alter demand intensity and average handling time.
Adjust staffing economics and service constraints
Results represent stochastic simulations, not measured service performance.
| Agents | SLA met | Mean wait | p90 wait | Utilization* | Cost / shift | Decision |
|---|
Build and compare fictional eight-hour service environments.
Backup your current library or restore the original samples.
JSON import validates its structure and limits. Imported text is treated as data, never interpreted as code. Backups only contain your local application settings and workloads.
A readable explanation of what is simulated, assumed and excluded.
QueuePilot is a transparent educational capacity planner. All starter profiles are synthetic and every recommendation depends on simplified model assumptions.
Each hour has a configured mean arrival rate. Requests are generated by an exponential interarrival-time process within that hour, equivalent to a nonhomogeneous, piecewise-constant Poisson arrival model. Back-to-back hours retain queueing delays.
A fixed number of parallel agents serve requests first-come, first-served. Service duration is exponentially distributed around the configured mean. Requests arriving before shift end remain eligible for service even when completion takes place after the shift.
SLA attainment is the percentage of arriving requests whose wait to begin service is no longer than the configured threshold. Mean wait and p90 are calculated from individual wait times, and aggregated across seeded simulation runs. Shift labor cost = agents × 8 hours × hourly wage; it excludes overhead, breaks, overtime premiums and idle inefficiencies. Utilization* means expected workload minutes divided by scheduled agent-minutes; this is a nominal capacity-load ratio, not measured actual busy time or a bounded utilization percentage.
Capacity Lab tests 1 through 16 agents by default, or up to 30 when more are already assigned. Every staffing candidate uses the same seed sequence, enabling more comparable stochastic outcomes. The recommended count is the lowest candidate meeting the configured SLA attainment target and a nominal load ratio no greater than 85%. If none meets both constraints, the application reports that no feasible staffing level was found in the tested range.
No abandonment, scheduled breaks, heterogeneous agents, priority routing, skill constraints, interruptions, shift handoffs, or observed operational inputs are modeled. Because there is no abandonment, waiting times can extend beyond shift end under overload. Performance results are stochastic estimates, not predictions. Validate with real staffing and arrival data before making operational decisions.
All processing occurs locally in the browser. Data persists via localStorage where available; your browser may clear it. Use JSON backup to protect customized work. The app has no external fonts, APIs, analytics, trackers or external libraries. It is released under the MIT license.