Smarter break modelling — lower Risk Index
From October 2026, Skillo describes the breaks on a shift to the Fatigue & Risk Index algorithm in a smarter way.
The algorithm itself has not changed, the shifts have not changed, and the break time your workers get has not changed. What changed is how the existing break allowance is spread across the shift when we hand it to the algorithm — and that alone brings the Risk Index down by roughly 8–10% on the most common shift shape (a shift with a 1-hour break, which covers around 90% of the shifts our customers schedule).
The short version
✅ Same algorithm — we have not touched the HSE Fatigue & Risk Index itself (see Fatigue & Risk Index).
✅ Same shifts, same total break time — nobody's break is shortened, moved, or added.
🔄 The break allowance is now described to the algorithm as several shorter breaks spread evenly across the shift, as long as every piece stays at least 20 minutes. Previously it was described as one single block in the middle of the shift.
📉 The Risk Index decreases on shifts with a recorded break of 40 minutes or more.
⚠️ The Fatigue Index may read slightly higher on some of those shifts — one part of the fatigue score used to be cancelled out by a single long block. The Risk Index, which drives the safety exceedances, improves by more.
How the algorithm sees a break
The HSE algorithm walks through a shift in 15-minute blocks. Risk builds up the longer someone works without a pause, and every break pulls it back down. The important detail is that the recovery a break gives depends strongly on its length:
| Break length | Built-up risk recovered by the break |
|---|---|
| 10 min | ~20% |
| 13 min | ~50% |
| 15 min | ~65% |
| 20 min | ~93% |
| 30 min or more | ~100% |
The recovery curve falls off a cliff right around 13 minutes. In other words: one long break gives one strong recovery point, but several shorter breaks (20 minutes or more each) give several of them — and no single stretch of continuous work gets long enough for risk to pile up.
Example: a 10-hour shift with a 1-hour break
The typical Skillo shift. Before, we told the algorithm:
work 4h 30 → break 60 min → work 4h 30
Now we describe the same 60 minutes of break as:
work 2h 15 → break 20 min → work 2h 15 → break 20 min → work 2h 15 → break 20 min → work 2h 15
Same total break time — but three recovery points instead of one, and no work stretch longer than 2h 15. The algorithm rewards that rhythm: the breaks component of the Risk Index drops from 1.045 to 0.958, so the whole Risk Index for such a shift comes out ~8% lower. A candidate sitting at a Risk Index of 70 on such a shift now sits at about 64.
More examples
| Shift | Recorded break | Before | Now | Risk Index change |
|---|---|---|---|---|
| 8 hours | 1 hour | 1 × 60 min | 3 × 20 min | ≈ −10% |
| 9 hours | 1 hour | 1 × 60 min | 3 × 20 min | ≈ −8% |
| 10 hours | 1 hour | 1 × 60 min | 3 × 20 min | ≈ −8% |
| 12 hours | 90 min | 1 × 90 min | 4 × ~22 min | ≈ −12% |
| 10 hours | 45 min | 1 × 45 min | 2 × ~22 min | ≈ −5% |
| any | under 40 min | 1 × break | 1 × break (unchanged) | no change |
Shorter breaks stay as a single block on purpose: splitting a 30-minute break into two 15-minute halves would push every break below the recovery cliff and actually score worse, not better.
Why 20 minutes?
Two reasons, and they agree:
- The law. 20 minutes is the shortest uninterrupted rest break that counts as a rest break under the Working Time Regulations. It is the smallest break that is meaningful in practice.
- The algorithm. The HSE recovery curve gives back almost nothing below ~15 minutes (see the table above). 20 minutes keeps each break firmly on the effective side of the curve while allowing as many recovery points as the break allowance supports.
What does not change
- The total break time on the shift — only how it is presented to the algorithm.
- Shifts with a recorded break under 40 minutes — these are modelled exactly as before.
- Shifts without a break set — these keep using the standard default parameters.
- All compliance checks and exceedance thresholds (Fatigue Reduction triggers) — unchanged.
- The HSE algorithm itself — untouched, as always.
What you may notice
Existing Risk Index values update as each candidate's fatigue is recalculated (for example when a shift is assigned, edited, or removed), so numbers will shift gradually rather than all at once. On shifts with a break of 40 minutes or more you should see fewer risk exceedances. On some of the same shifts the Fatigue Index may read slightly higher than before — that is the expected trade-off of no longer assuming the whole break lands in one block, and it does not affect the safety-side Risk Index, which improves.
For the full technical description of how the Fatigue and Risk Index are calculated, see Fatigue & Risk Index.