Executive summary
Centralising governance does not mean merging local queues.
A multi-location organisation needs a common standard, but it must not assume one common capacity. A staff member in one branch cannot automatically serve a customer in another. Locations differ in opening hours, service offer, staff on duty, service duration and local demand.
The branch is therefore the basic operating boundary of the queue. A ticket is issued only after location selection. Availability, estimate and admission decisions must use the state of that branch rather than an organisation-wide average.
The central layer has a different job: define authority, maintain comparable rules and provide oversight. The local layer delivers service and owns queue accuracy. A reliable system connects the two without merging their responsibilities.
Research basis
An organisation average is not the operating truth of a branch.
Little's Law applies to the system boundary that is actually defined. If a branch is a separate arrival and service system, its people in queue, throughput and waiting time must be analysed locally. The organisation total is useful for management, but it cannot explain the wait at one location.
Kingman's heavy-traffic result further explains why two branches with equal daily volume can produce very different waits. Arrival pattern, service-time variation and moments of actual staff availability matter more than the total number of completed services alone.
Resource-pooling research shows that pooling can improve performance when demand and servers are sufficiently interchangeable. Benjaafar also identifies limits and conditions under which pooling can lose its benefit or degrade performance. For physically separate salons, capacity may be represented as shared only when a customer can actually be redirected to a compatible service, not merely because both branches have the same owner.
Wang, Liu and Wan show that walk-in arrivals introduce distinct uncertainty and cannot be treated as a simple filler for scheduled capacity. Their model comes from healthcare, but the operating lesson transfers: local walk-in demand must be an explicit planning input, not residual work absorbed after every other decision.
Operating flow
From organisation policy to local execution without losing ownership.
1. Configuration ownership
The organisation defines who may edit branches, services, staff, channels and notifications. Every change remains bound to the exact tenant, location and authority.
2. Location selection
The customer selects a physical branch before joining. The public surface exposes only that branch's services, opening hours, channel price and queue state.
3. Local admission
The branch decides whether to admit a new ticket from its own load and projected completion. Spare capacity elsewhere does not change that decision without a real redirect.
4. Local execution
Staff, kiosk and TV are bound to a branch. The local team runs its own queue and cannot change the lifecycle of a ticket at another location.
5. Organisation oversight
The portal combines comparable measures after local records are complete. Central reporting must not rewrite operating history to improve the result.
Control model
Data, authority and responsibility boundaries.
- Capacity boundary
- The branch, its opening hours, active staff and locally available services.
- Data boundary
- Every ticket, service, staff member and device belongs to an exact organisation and branch.
- Authority boundary
- Organisation role grants access; each local action remains restricted to assigned scope.
- Device boundary
- After activation, Staff, kiosk and TV operate only on the branch to which the device is paired.
- Reporting boundary
- Organisation measures aggregate local records without losing original branch identity.
Measurement
Prove local state first; compare the organisation second.
- Local waiting
- Median and 90th percentile by branch, day, hour and service type.
- Local throughput
- Completed services per available hour and staff member, with actual service mix.
- Abandonment
- Cancelled, expired and no-show tickets by branch and admission channel.
- Configuration accuracy
- Share of time opening hours, staff presence and service availability match reality.
- Comparability
- Compare branches only after separating service mix, opening hours and local demand.
- Organisation view
- Totals and trends that retain a path back to the source branch record rather than an anonymous average.
Common model failures
Central visibility must not erase local reality.
One global queue
Physically separate locations appear interchangeable even though customer and staff cannot move at the point of service.
Organisation-wide ETA
Average waiting across all locations does not describe the branch the customer selected.
Global catalogue without local availability
A service may exist in the organisation while one branch does not offer it or has no qualified staff that day.
Device without a hard location
A kiosk or Staff surface that can ambiguously switch branch can write to and mutate the wrong queue.
Ranking raw averages
A branch with shorter services looks more efficient even when the comparison ignores work mix and demand.
Evidence boundary
What research justifies and what an organisation must prove with its own data.
This study connects Rowomat's organisation model with research on flow, heavy traffic, walk-in uncertainty and resource pooling. The literature supports clear system boundaries and caution in pooling; it does not establish a result for a named salon network using Rowomat.
Physical salons differ from call centres or digital networks because customer, device and staff have a location. Theoretical pooling benefit does not transfer automatically. A redirect requires a compatible service, acceptable travel and genuinely spare capacity at the destination branch.
A customer case study requires data from at least two locations before and after adoption, stable metric definitions and permission to publish. Without that evidence this document remains an operating study, not a claim of measured commercial effect.
References
Primary work behind the organisation analysis.
Sources are provided for verification. Interpretation for physically separate walk-in salons is original to this study.
- A Proof for the Queuing Formula: L = λWJohn D. C. Little · Operations Research 9(3) · 1961
Defines the relationship between units, throughput and time inside a properly chosen system boundary.
- On Queues in Heavy TrafficJ. F. C. Kingman · Journal of the Royal Statistical Society, Series B 24(2) · 1962
Shows waiting sensitivity when local demand brings a branch close to full utilisation.
- Performance Bounds for the Effectiveness of Pooling in Multi-processing SystemsSaif Benjaafar · European Journal of Operational Research 87(2) · 1995
Examines benefits, limits and conditions under which resource pooling can improve or degrade performance.
- Managing Appointment-Based Services in the Presence of Walk-in CustomersShan Wang, Nan Liu and Guohua Wan · Management Science 66(2) · 2020
Shows that walk-in demand creates distinct uncertainty that planning must model directly.