Evidence

Five case studies, published by someone else

These are not testimonials we wrote. They are case studies printed inside a EUROCONTROL Network Manager guideline, in a section the guideline introduces with the line “The source of Use cases is SkyRoster.” You can download the document and check every figure below against it.

The source document

Air Traffic Services Operations staff/personnel Rostering Best Practices and Tools Guidelines

Reference: NMD/SAF/2653
Edition: Edition 1.0, 26 April 2023
Classification: Green
Published by: EUROCONTROL, Network Management Directorate

“The source of Use cases is SkyRoster.” , page 20

What else the guideline attributes to SkyRoster →

Case study 1 · page 21 · 75 ATCOs
10% workforce cost reduction
published
The problem, as the guideline states it
The ANSP wanted to optimize their rosters in one of their ATC units.
The result, as the guideline states it
The application managed to cover the same traffic demand and accommodated the same number of leave days during a 30-days cycle using 1 less ATCO.
Extrapolating the simulations to other units, the ANSP managed to cut annual costs with the workforces with 8-12% per ATC unit, translating in an economy of 750.000 EUR / year.
Measured, and extrapolated
Measured: one fewer controller covering the same demand and the same leave over a 30-day cycle, in a target unit that was scheduling 10 controllers on a fixed M-M-A-A-N-N-R-O-O-O pattern. Extrapolated: the 8 to 12 per cent annual cost figure and the 750,000 euro total, which the guideline itself describes as extrapolated to other units.
How we read it
A fixed rotation was replaced with flexible allocation against the same manpower requirement. The saving is one controller in one unit. Everything beyond that is the ANSP's own extrapolation across its other units.
Case study 2 · page 22 · 96 ATCOs in 12 teams of 8
1-month Tactical Roster Simulation with the baseline headcount
published
The problem, as the guideline states it
An ATC Unit with 96 ATCOs rostered in teams on a fixed shift pattern sees significant variations in terms of overtime distribution per individual staff and increased staff leakage during the night shifts.
The result, as the guideline states it
Compared to the previous simulation, in which 24 additional ATCOs were used to reduce overtime to under 300 hours per month, how is the workload distributed when deploying only the current headcount of 96 ATCOs?
Measured, and extrapolated
A one-month team-based roster on a fixed MANROO pattern, 1 to 31 May, morning 07:00 to 14:00, afternoon 14:00 to 21:00, night 21:00 to 07:00. Annual leave was distributed randomly per employee and averaged over the year; no duty leave was registered in this run.
How we read it
This is the control case, not a result. It establishes what the same unit looks like without extra staff, so that case study 5 has something to be compared against.
Case study 3 · page 34 · 55 ATCOs
94.3% less overtime
published
The problem, as the guideline states it
The ANSP is understaffed, scheduling overtime shifts to cover the demand. Due to overtime shifts, staff is overworked, resulting in alarming fatigue scores.
The result, as the guideline states it
The team designed 5 rostering simulations to model their staffing plans in different scenarios.
In the scenario with the best result, the application managed to decrease the number of overtime days from 247 to only 14 during a 35-day roster cycle.
With the new design shift roster and same amount of staff the number of overtime has been reduced but also the number of shifts and consequently the ATCOs have a better fatigue index scores.
Measured, and extrapolated
Measured within a single 35-day roster cycle at one unit, as the best of five modelled scenarios. The headcount did not change: the guideline states the reduction came from redesigning teams and shift distribution with the same amount of staff.
How we read it
Overtime days fell from 247 to 14 in one 35-day cycle. That is where the 94.3 per cent comes from. It is the best of five scenarios, at one understaffed unit, with no extra people hired.
Case study 4 · page 35 · 450 operational ATCOs
Automating 60.000 shift swaps to save a ~360 000 EUR/year
published
The problem, as the guideline states it
The customer was experiencing a high number of shift swap requests - 60.000 per year. For each shift swap request, the planning manager wasted 15 minutes of his time to fill in paper forms and update the roster in a spreadsheet. The customer was paying 5 full-time employees only to manage these shift swaps.
The result, as the guideline states it
By deploying the application inside their environment, the shift swapping process was reduced to 1.5 minutes on average, counting since the manager receives the swap request in the system.
Employees were also empowered to swap their shifts self-service, without involving their manager, the Rostering Engine taking care of the responsibility to maintain a safe, compliant roster with every shift swap approved.
Only by automating the entire shift swapping process, the customer was able to cut costs by 360.000 EUR/year.
Measured, and extrapolated
Measured: the handling time per swap request, from 15 minutes to 1.5 minutes on average. Derived: the annual saving, which follows from that handling time multiplied by the volume and the five full-time staff the customer had assigned to the task.
How we read it
The measured thing is handling time per swap. The euro figure is what that handling time is worth at this customer's volume and staffing. At a different volume it is a different number.
Case study 5 · page 36 · 120 ATCOs in 12 teams of 10
1-month Team-based Roster Simulation with 120 ATCOs
published
The problem, as the guideline states it
An ATC Unit with 96 ATCOs rostered in teams on a fixed shift pattern sees significant variations in terms of overtime distribution per individual staff, and increased staff leakage during the night shifts. This simulation focused to look at difference by adding 24 more ATCOs to arrive at 120 ATCOs in total.
The result, as the guideline states it
How can the roster balance overtime shifts between available employees for a one month period (April 202z) and minimise leakage?
Measured, and extrapolated
The same unit and the same fixed MANROO pattern as case study 2, over 1 to 30 April, with 24 additional controllers. Annual leave and duty days were input for the period and averaged over the year.
How we read it
This one models hiring, not software. Read together with case study 2 it answers a staffing question the ANSP had: what does 25 per cent more headcount buy us.

What this does not show

What five case studies can and cannot tell you

Read them for the mechanism, not as a forecast for your unit.

Different units, different numbers

A 55-controller unit that is understaffed and a 450-controller operation drowning in swap paperwork have almost nothing in common. Neither result predicts yours. What transfers is the method: model your own unit, compare scenarios, then decide.

Some of this is hiring, not software

Case studies 2 and 5 are a staffing question, modelled. Adding 24 controllers to a 96-controller unit changes the overtime picture whatever tool you use. We separate those from case study 3, where the headcount did not change at all.

No benchmark is claimed

You will not find a solve time or a controllers-per-second figure anywhere on this site. No production-scale benchmark exists, so rather than estimate one we leave it out. If a tender needs the number, we will run it against your unit shape and show the working.

Model your own unit instead of borrowing someone else's number.

Bring a month that fought you. We will set it up the way these case studies were set up, and you will see what the engine does with your demand, your pattern and your rules.

Book a working session Send a question instead