Figures and sources

Where every number comes from.

Every product number on this site is sample data: one simulated pay period, checked by the same rules the software runs. No real student data appears anywhere. Here is where each figure comes from, and how it was worked out.

The sample pay period

Every figure starts from one simulated two-week pay period.

Pay period: Mar 1 – 14, 2026; two weeks
Students: 1,000; synthetic student employees
Time entries: 6,070
Work areas: 50
History: 5; earlier pay periods
Wage: $13; an hour: the department default wage, used for every dollar figure

HourFlag’s synthetic data package generated it, and HourFlag’s rules engine checked it: the same code the free check runs. Each time the website is built, every figure is regenerated from that code, and the build stops if any number was edited by hand.

Why this size

The size is an assumption, chosen to look like a large campus department: 1,000 student employees working about three shifts a week each, so about 6 time entries per student in a two-week pay period.

For comparison, 40 published counts for large campuses, gathered in October 2026, give these medians:

850 student dining employees at public campuses of 20,000 students or more
1,000 at public campuses of 40,000 students or more that run their own dining

Which entries are flagged, and how often, was written into the synthetic data to show every check working. It is not a measurement of any institution.

The dataset itself, one file per report, is on the sample data page. Open it in the free check and you get the same list and the same numbers.

The supervisors in the sample

The sample gives its 1,000 students 30 timesheet approvers over 50 work areas: three per location group.

Timesheet approvers: 30; for 1,000 students
The home page’s supervisor: 42 students; “262 entries, 13 to look at”

These are assumptions, not measurements. None of the universities searched publishes how many people approve student timesheets. So the count was set from what dining departments do publish about their managers and supervisors, in research done in October 2026.

What dining departments publish

Those published counts put managers and supervisors at about 25 to 60 per 1,000 student employees, with a middle near 40. The sample’s 30 sits in the lower half of that range.

46 administrators for 850 students
A public university’s dining operation, with five dining courts.
Purdue, 2024
10 managers or assistant managers for 258 students
A dining service with five dining centers.
Nebraska, staff page Its 2018 count
3 supervisors for 75 students
One food court.
Eastern Illinois, 2018

Public pages, read in October 2026. None of these institutions is a customer or partner of HourFlag.

Counting only each large unit’s general manager gives a lower range: 10 to 25.

A timekeeping system’s routing rules allow it too. Timekeeping systems commonly require at least one approver per work area and allow several, so 50 work areas for 30 approvers is within the rule.

The yearly estimate

The story on the home page ends on “$158,288 saved a year”: an estimate for 1,000 student employees.

It is an illustration from simulated sample data: the estimated potential over-payment over one academic year, if every flagged hour were corrected, at the printed wage, with a schedule file.

Nothing here is a result. Inside the product, the same hours read “needs an answer” until a supervisor answers them.

The figure is the engine’s own extrapolation: the rate measured on the sample pay period, times students, times pay periods.

Hours over: 608.8 h; measured in the sample pay period
Students: 1,000; student employees
Pay periods: 20; a year: two semesters of ten two-week pay periods
Wage: $13; an hour

The arithmetic

608.8 hover in one pay period

×20pay periods a year

=12,176 hover in a year

×$13an hour

=$158,288a year, estimated

The engine’s own notes

The engine prints these lines with the estimate:

Measured over 1 pay period: 6,070 shifts, 1,000 student pay periods.
Measured rate: 0.6 hours of estimated potential over-payment per active student per pay period.
Students: 1,000 (an input; replace it with your own count).
Pay periods a year: 20 (an input; busy pay periods only, if you prefer).
Wage: $13.00 an hour, the department default wage.
The checked pay periods are assumed typical. Items that still need an answer have no dollar figure and are not included. Hours over the cap are a policy figure and are not included.
Every figure is an estimate; replace with your own numbers. $158,288 a year is the estimated potential over-payment at the wage above, before approval.

The student count and the 20 pay periods are inputs, not measurements. The free check’s findings report prints the same table with your own counts.

How hours and dollars are counted

Hours come first, then dollars.

Hours are the recorded time minus the expected time. The expected time is the one a supervisor enters; without it, the schedule; without a schedule, the opening hours.
Dollars are those hours times the wage.

So every dollar figure is an estimated potential amount to review, never a loss.

What reaches a supervisor’s list

Hours paid over the schedule are put in front of supervisors.
Hours under are counted in the findings report and its headline (88.8 h on 53 entries in the sample), never put on a supervisor’s list. If a supervisor’s answer shows hours are owed, the fix still goes through.
Hours over the weekly cap are a policy figure, never dollars.
An item whose right time isn’t known yet has no dollar figure until someone answers it.

Why there are two totals

Each check counts its own minutes, so a minute that two checks describe appears in both checks’ figures. Totals count each minute once.

The supervisors’ lists leave out small one-offs that the findings report still counts. The yearly estimate uses the findings report’s figure.

On the supervisors’ lists: 555.6 h over; 210 timesheets to review
In the findings report: 608.8 h over; small one-offs included: the figure the yearly estimate uses

The checks figure

“102 of 263 flags appear in no single report your timekeeping system makes”

For each flag on the list, the engine records everything it took to find it, then sorts the flag to one of two sides.

The timekeeping system’s side: 161; One of its own reports alone, or a sort of the base report, would have listed it.
HourFlag’s side: 102; Only when everything it needed was the schedule file, the opening hours, two reports read together or memory of earlier pay periods.

What the 102 on HourFlag’s side needed (a flag can need two):

71 needed the schedule
21 needed the opening hours
51 needed earlier pay periods remembered
None needed two reports read together

The rule is the cautious one: if any of the system’s own reports would have listed a flag, it counts on the system’s side.

The sample data page prints the count by source.

“What gets measured gets improved.”

A common management saying. The home page gives it without a name, because it has no reliable author.

It is often credited to Peter Drucker, but an article published by the Drucker Institute in 2013 (Paul Zak, “Measurement Myopia”) says that is wrong.

Timekeeping systems

The free check reads the standard reports a timekeeping system can export: time entries, automatic clock-outs, missed punches, corrections and who approves what. Today it knows one system’s report layouts, and the first meeting checks that it reads yours. On the timekeeping system’s reports alone, the repeated-pattern rules only see forgotten clock-outs, missed punches and removed meal deductions; repeated early or late clock times need a schedule file.

The part that reads the files is separate from the rules, so another system’s reports can be added without changing a single rule.