Lateness Rules That Produce the Behaviour
How a lateness rule with a sharp edge produces badge-passing, why the compliance figures look excellent, and the three ways to soften it without losing the point.
One lateness rule, and what the punch distribution shows about it
The rule achieves near-perfect compliance on its own measure and the distribution shows how. This is one employer's own data, not a statement of what any rule requires.
A lateness rule with a sharp edge and a real consequence produces avoidance, and the avoidance looks like excellent compliance. Ninety-seven per cent of punches before the hour, three recorded instances of lateness in a month — and forty-one per cent of all punches in the final ninety seconds, which is not what natural arrival looks like.
The control described in “Lateness Rules That Produce the Behaviour” should be matched by transparent operating rules. Organisations considering inspect the published feature set for how employees cheat time trackers can make that use more credible by publishing the purpose, selecting only necessary settings, limiting manager access and fixing a review date before the first record is collected.
The rule is working on its own measure. What it is measuring is how well people avoid it.
The Google Workspace security guidance offers another lens on the issue raised in “Lateness Rules That Produce the Behaviour”. Compare its principles with the actual record, ownership model and review route rather than importing a generic checklist unchanged.
Why a sharp edge produces this
Because the cost of being one minute late is identical to the cost of being twenty minutes late, which makes the minute before the hour enormously valuable and the minute after it worthless.
Any rule with that shape concentrates behaviour at the boundary. The response is not dishonesty; it is what the rule asks for.
Reading your own distribution
Plot punches by minute for a month around shift starts. The shape answers the question immediately.
A natural distribution spreads over several minutes either side with a gentle peak. A rule-driven one has a cliff: a dense cluster before the boundary and almost nothing after it.
The three softenings
A small grace that does not deduct pay — forgiving lateness without removing minutes, which is a distinction most systems implement as one setting and which matters.
A cumulative measure instead of an event one: minutes late over a month rather than instances, which removes the cliff entirely.
A staggered start, so that the boundary is not the same moment for everybody and the terminal is not the constraint.
Keeping the point of the rule
None of this is an argument against expecting people to be on time. The operation needs people at their stations, and a rule that nobody can fail achieves nothing.
The distinction is between a rule that produces attendance and one that produces punches before the hour. A cumulative measure does the first; an event-based one with a sharp consequence does the second.
Consequences that people can see
Where lateness has a consequence, people should know what it is and be able to see where they stand.
A policy whose consequences are discretionary and invisible produces anxiety and avoidance out of proportion to the actual risk, which is the mechanism by which a mild rule produces the behaviour described here.
The rule and the queue together
Neither alone produces much. A queue with no lateness rule produces grumbling. A lateness rule with no queue produces people arriving at five to.
It is the combination — a boundary that matters and a device that cannot process everybody before it — that produces badge-passing, and fixing either side removes it.
Testing a change
Change the rule or the start time for one shift for a month and plot the distribution again. The shape will tell you whether it worked within weeks.
That test is cheap, it produces evidence rather than opinion, and it is the thing to do before anybody buys a control to address the behaviour the rule is producing.
Reading the shape
The distribution is produced by one query and read by eye, and the shape says which rule you have.
- A gentle peak spread over several minutes: natural arrival.
- A dense cluster in the final ninety seconds, almost nothing after: a cliff.
- Two clusters, one before the hour and one well after: a queue plus a cliff.
- A flat spread with a tail after the hour: no effective rule.
Four shapes. The second and third are the ones that produce the behaviour described here, and both are fixed by changing the rule rather than the people.
The rule nobody wrote down
In several operations the formal policy is mild and what is actually enforced is not: a supervisor who comments, a pattern of being given the worse jobs, a conversation that everybody has heard about.
That unwritten rule produces the same distribution as a written one, and it is invisible to anybody reading the policy. Asking people what actually happens if they are late is the only way to find it.
The rule, written down with its distribution
The rule, what it measures, what the consequence is, and the shape of the distribution it produces, reviewed annually.
The last field is the one that makes the rest meaningful. A rule producing a cliff is a rule that is being avoided rather than met, and the distribution is the only place that is visible.