Why a day level comparison lies to you
Invoices cluster on fixed days. That makes this year against last year at day level often simply wrong. Work on closed months.
It is the fifteenth of the month. Your dashboard says your cumulative revenue is twelve percent below last year. Phones get picked up, meetings get called, course gets corrected.
And at the end of the month it turns out to be four percent above last year.
Nothing was wrong. Only your comparison was no good.
What is going on
Invoices do not arrive evenly across a month. They cluster, and they cluster per channel at different moments.
In the model I worked with, that was pronounced. Business invoicing sat entirely in the first twelve days of the month. The webshop invoiced at four fixed moments spread over the month. And store sales came in as one collected amount at the end.
That pattern is not universal, but its shape is. Almost every company has moments when invoicing happens and days when nothing happens, and those moments sit differently per channel.
Now compare day fifteen of this year with day fifteen of last year, and you are comparing two arbitrary points in two different rhythms. If retail invoicing fell on the twenty-seventh last year and on the twenty-eighth this year, your comparison on day fifteen misses nothing, but on day twenty-seven it is off by an entire month of store sales.
Why weekdays make it worse
On top of that comes something else. The same date falls on a different weekday every year.
If your invoicing happens on working days, the fifteenth of this month may have ten working days behind it and last year eleven. That is a ten percent difference in processing capacity, purely because of the calendar.
Public holidays do the same, and in Belgium those shift from year to year as well.
What to do instead
Work with closed months. Compare the year up to and including the last fully closed month, not up to today.
That feels slow, and that is exactly the intention. Your year comparison is a strategic number, not an operational one. To answer whether your year is on track, you do not need to know what happened yesterday.
If you do want something inside the running month, use a different instrument: compare the number of elapsed working days instead of calendar days, or compare the running month with the average of the past three instead of with last year.
And if you absolutely want this year against last year at day level, do it per channel and not on the total. Then you see straight away that the deviation sits with one channel and that it is about timing.
How to put it in your report
Two numbers, clearly kept apart.
One that says how your year is running, on closed months, with a label saying how far it runs. That is your steering number.
One that says how the running month is progressing, against the normal course of a month at your company. That is your operational number.
What you do not do is throw those two into one chart, because then someone compares them with each other and you are back where you started.
The claim
Daily numbers are useful for seeing whether something is breaking. They are unsuitable for seeing whether your year is going well. Mix those two up and you steer on noise while calling it data driven work.