Which time period should each sales report use?

Every number here comes from a synthetic dataset modeled on a small biotech: one injectable brand, 60 territories with one rep each, distributor sales data in units and calls logged in the CRM (customer relationship management system). There are no real patients, prescribers or accounts in it. The data runs through Monday, October 12, 2026. To show what a team sees midweek, we read it as of Thursday, October 8, using only rows through Wednesday, October 7, and look past that date once, to see how the open week closed. One of the six distributors sends its sales in a weekly file that lands on Monday, so on that Thursday its rows ran only through Friday, October 2. Weeks run Saturday to Friday and are named by the Friday they end. Net units are sales minus returns. Calls count each account once per day, channel, and live visit or drop-off. In this data, 99% of net units and 98% of calls fall on weekdays.

The open week reads as a drop

A dashboard that puts this week next to last week shows units down 59% and calls down 35%. But the week is five days old, and only three of those days are weekdays, when nearly all orders ship and calls happen.

The fair comparison is like for like: Saturday to Wednesday this week against the same days last week. On that basis, units are 18% lower and calls 4% higher. Three weekdays are still a small sample, and one distributor's file for the week hadn't landed, so treat it as an early read, not a result.

The data runs past this read, so the example can show how the week ended. With all six distributors in, it held 1,372 units and 897 calls: 16% fewer units and 4% more calls than the week before. The like-for-like read pointed the right way on both and came within two points. The open-week read put the drop in units at 59% instead of 16%, and showed calls falling 35% in a week they rose.

Pace against goal: prorate to the last day of data

The example's Q4 2026 goal is 19,359 net units, the sum of its 55 territory goals (the five medical science liaison territories carry no sales goal). Of the 1,484 units shipped quarter to date, 1,394 are credited to a territory. The other 90 went to ship-tos not yet mapped to an account, which no territory goal covers, so they stay out of both sides.

Seven days into a 92-day quarter, 7.2% of the full goal says little. Sales and calls here land on weekdays, so spread the goal evenly over the quarter's 66 weekdays, Monday to Friday, and prorate it to the weekdays that have data: 5 of 66 (7.6%) gives 1,467 units, and the team is at 95% of pace. Prorate it to today instead and Thursday counts as a weekday with no sales. The goal to date grows to 1,760, and the same team reads 79%: well behind, when it is close to pace. Early in a quarter, one weekday without data moves pace by 16 points.

Why weekdays and not calendar days? A goal spread over calendar days keeps growing through Saturday and Sunday while sales and calls nearly stop, so pace sags every weekend and reads behind on Monday morning. This read covers one full weekend, so here the two nearly agree: 7 of 92 calendar days is also 7.6%, or 1,473 units and 95% of pace. Business days net of holidays would be more exact still: Q4's 66 weekdays less Thanksgiving and Christmas leave 64, and 5 of 64 is 7.8%. The example counts plain weekdays and leaves holidays in, so a holiday reads as a weekday with no sales, and pace dips in a holiday week.

Trend charts: closed buckets, with the open one marked

Over the last 13 closed weeks, weekly net units ran from 1,137 to 1,720. As a plain bar, the open week's 669 sits below all 13. Projected to a full week by calendar days (669 times 7, divided by 5), it becomes 937. Those five days hold only three weekdays: projected by weekdays (669 times 5, divided by 3), it becomes 1,115. Both fall short of the 1,372 the week closed at, the weekday projection by 19%.

A projection is only as good as its base, so don't let one stand in for the bar. Draw the open bucket lighter and label it "in progress", or leave it off and state the data-through date. Label any projection as an estimate. Months work the same way: on this Thursday, October's bar holds seven days.

Trailing windows: smooth, but they move

All three cover about the last quarter. Their totals are up to 1,485 units apart, mostly because the windows differ in length: per week, they sit within 3% of each other. When two windows differ in length, compare rates, not totals, and write the dates on every number, not just a label.

The trailing 12 weeks starts on a Thursday, ends on a Wednesday and includes the five open days. Tomorrow it drops July 16, adds October 8 and changes its total, whatever happened in the field. That suits smoothing and rolling questions such as "which accounts haven't ordered in 12 weeks?", but not a scoreboard, where nobody can reproduce yesterday's number. The last 13 closed weeks stay put until the data reaches Friday, and fit a weekly chart. The closed quarter doesn't move with the calendar at all.

End a trailing window on the last day of data. A trailing 7 days that ends today pulls in Thursday, which has no data yet, and reads 1,132 units, 24% lower than the 1,484 for October 1 to 7. And on this Thursday, the trailing 7 days and quarter to date cover the same days. Next Thursday they won't, so label which one a report uses.

Some 13-week reports count the current week as one of the 13 and leave it out until it closes, so for most of the week they hold 12 full weeks. Why do two sales reports show different numbers? shows how far that, and a trailing window with the same label, can move a total.