How we measure
Scheduling results depend on the group being scheduled. So we measure on one fixed group, the same every time, and report every month we run, not the best one.
The reference group
An entirely fictional emergency medicine group. No customer's data is in it.
- Six emergency departments with day, mid, evening and night physician shifts and day and night APP shifts.
- 92 clinicians on contracts of 60 to 160 hours a month, with contracted minimums; about one in ten per diem.
- A realistic mix: nights only, no nights, one night block a month, a weekday someone never works, pairs who never work the same day, some who work at a second site.
- Time off every month: a few days for most people, a week away for about one in eight.
- About half the clinicians have personal preferences: how many days in a row they like, extra weekends they are willing to work, more or fewer nights.
- Staffed "just enough": everyone's maximum hours add up to what the shifts need, so not every shift can be filled.
The rules
The defaults a new SahVir site starts with, as hard limits: at most 5 days in a row, at least 3, 2 days off between runs, 10 hours' rest between shifts, at most 2 weekends a month (more only for someone who opted in), and nobody over their maximum hours. A shift that can only be filled by breaking one of these is left open and listed.
Results: SahVir Optimize, all six sites together
| Month | Shifts | Filled | Left open | Rules broken |
|---|---|---|---|---|
| November 2026 | 1,008 | 97.9% | 21 | 0 |
| December 2026 | 1,052 | 97.9% | 22 | 0 |
| January 2027 | 1,036 | 97.4% | 27 | 0 |
| February 2027 | 944 | 99.6% | 4 | 0 |
Optimize as it runs in production: the strict version (no extra hours, no extra weekend), 90 seconds per month. Every schedule is scored by one independent rule check. Measured October 2026.
Optimize and instant generation
Every plan includes instant generation, which builds a month in seconds. Optimize, on the Professional plan, builds every site together and takes a few minutes. On this group, Optimize left fewer shifts open than instant generation in every month, with no rule broken by either. How many fewer depends on the group and the month, so in a pilot we build your own month both ways.
What this does not show
There are no human-built schedules for a fictional group, so this cannot show how SahVir compares with your scheduler. That is what the free audit and a pilot are for: the audit scores your own last month on the same checks, and in a pilot your scheduler and SahVir build the same month side by side.