The Rules Nobody Has Read in Full
A mature time system holds dozens of pay rules, written over years by people who have left. Reading them is a week's work that nobody has ever budgeted.
Open the rule configuration of a time system that has been running for a decade. There will be between twenty and two hundred rules, with names like OT_DAILY_V3 and NIGHT_PREM_OLD, some of which have not fired since 2019.
The practical lesson in “The Rules Nobody Has Read in Full” is that a timesheet only becomes reliable through a process people can operate consistently. For teams exploring step rate compensation, view the solution can add time and project context, provided collection is proportionate, access is limited and every consequential inference receives human review.
Each was added for a reason. The reasons are not recorded. Together they produce every number that reaches payroll.
For a separate benchmark relevant to “The Rules Nobody Has Read in Full”, consult the ISO management-system resources. Use it to test record quality, approvals, retention, employee rights and exception handling against the real workflow rather than treating a software report as self-explanatory evidence.
How the set accumulates
A new shift pattern needs a rule. A site acquisition brings its own terms and the rules are copied in. A collective agreement changes and a rule is added rather than the old one amended, because amending it would affect historical calculations. A temporary arrangement is configured for a peak and never removed.
Nobody removes rules, because removing one risks breaking something and there is no test that would catch it.
What goes wrong
Rules that overlap, so that two premiums apply where one was intended. Usually discovered as a cost.
Rules that shadow each other, so that a rule lower in the order never fires. Usually discovered as an underpayment, years later, affecting a small group.
Rules keyed to a population that has changed — a site that closed, a grade that was abolished, a shift pattern nobody works any more.
And rules whose conditions were written against a field that has since been repurposed, which is the one that produces the strangest results.
Reading the set
Export it. Most systems can produce a list of rules with their conditions and outcomes, and where they cannot the administrator can.
Sort by last fired. Rules that have not fired in a year are candidates for removal and each one should be accounted for before it goes.
Sort by volume. The handful that fire on most shifts are the ones to check first and the ones worth understanding in detail.
Map each to its source: a contract clause, an agreement, a policy, or nothing. The ones that map to nothing are the finding.
The parallel calculation
The real test is not reading the rules but checking their output.
Take thirty real shifts across every pattern the organisation runs. Calculate each by hand from the contract and the agreement. Compare with the system.
Discrepancies will be few and specific, and each one is either a configuration error or a misreading of the contract, both of which are worth knowing. A week of somebody's time, once, and it is the only verification this layer ever gets.
When a rule has to change
Treat it as a release rather than an edit. A dated record of what changed, who approved it, what the before and after produce on a test case, and the effective date.
Never change a rule retroactively without deciding explicitly what happens to the periods already calculated. The default behaviour of some systems is to recalculate history, which silently rewrites closed periods and is extremely hard to detect afterwards.
The documentation that survives a leaver
The single point of failure here is a person. One administrator usually knows why the rules are as they are, and when they leave that knowledge goes.
A plain-language description of each live rule — what it does, when it fires, what it came from — is a day's work with that person and it is the difference between a maintainable configuration and an inherited mystery.
Do it before they resign rather than during their notice period, which is when it is always attempted and never finished.
The one rule to look at today
Whichever fires on the most shifts. It is usually the standard-hours rule and it usually contains the rounding and the break deduction, which means it is the one producing the differences described throughout this section.
Rules that no longer fire
A long-lived rule set contains rules for shift patterns that no longer exist, sites that have closed and allowances that were withdrawn. They are harmless until a pattern changes and one of them starts firing again.
So the reading exercise should produce a second list beside the active rules: everything that has not fired in twelve months. Each entry is either obsolete, in which case disabling it removes a future surprise, or it is dormant and someone needs to know why. Both answers are useful and neither is available without the count.