Monday looks worse because Friday never finished

Somebody looks at the weekly figures and asks why Monday is always the worst day. The usual answers are about people: a slow start, the weekend, a meeting. Occasionally that is true.

Far more often, Monday is worse because of arithmetic that happened while nobody was there.

The gap is not idle, it is filling

A step staffed one shift a day processes work for eight hours. Work arrives for considerably longer than eight hours, and in most operations it arrives for twenty four: orders come in overnight, other time zones are working, upstream steps on different rosters keep releasing, and anything automated never stops at all.

---

	Arrivals across the full day	96
	Arriving inside the staffed shift	44
	Arriving outside it, into the queue	52
	Capacity of the shift	96
	Waiting at the start of the next shift	52

Illustrative figures. Daily capacity exactly matches daily demand, so on paper this step is perfectly sized and its utilization reads 100 percent with nothing left over. It still starts every morning with 52 items already waiting, because more than half the day's arrivals landed while it was closed.

The step is not under-resourced. Over a day it processes exactly what it receives. But the queue it starts with is the arithmetic of the gap, and no amount of working harder inside the shift removes it, because the gap is where it came from.

Now make the gap a weekend. Two full days of arrivals with nobody processing, and Monday morning starts with roughly three times the backlog of a normal morning. Monday is not worse because of Monday.

Why the daily average hides it completely

The figure that gets reported is usually a daily or weekly average, and at that resolution this effect is invisible. Daily capacity matches daily demand, so the number looks correct, and it is correct. It is simply answering a question nobody asked.

What a reader needs is the queue depth through the day, not the total across it. The same twenty four hours that average out to a balanced step contains a queue climbing steadily from six in the evening to eight the next morning, and being worked back down through the day. That shape is the finding. The average is the shape thrown away, for the reasons in the piece on averages.

The three fixes, in order of cost

Move the shift, do not lengthen it. If most arrivals land in the evening, starting the shift later costs nothing and removes a chunk of the overnight accumulation. This is free and it is almost never tried first.

Stagger the roster against the arrival pattern. Two people from eight, two more from eleven, matched to when work actually arrives rather than to when the building opens. Same headcount, different queue.

And only then, extend the cover. Adding hours costs real money, and it is the right answer only once the first two are exhausted, because they are free and it is not.

What to do with it

Plot arrivals by hour of day, for a fortnight. Not by day. Most operations have never looked at this chart and it usually explains the complaint on its own.

Then plot the same for the roster, on the same axis. The gap between the two curves is the queue, and where they cross is where the backlog peaks.

And check whether the shift boundary sits on the constraint. A gap in front of a step with spare capacity is absorbed by lunchtime and costs nothing. The same gap in front of the constraint is never recovered, because there is no spare capacity anywhere to recover it with.

Monday is a symptom. The roster is the cause, and it is visible in an hourly chart that takes twenty minutes to draw.