Build the quarter out of rep-weeks.

Every quarter, the default move is one calculation. Take the number, divide by the number of reps, nudge each rep up or down by gut feel, and call it a plan. It looks like math. It is arithmetic performed on the wrong unit.

The division only works if two things are true: that reps are interchangeable, and that weeks are interchangeable. Neither holds in the fourth quarter specifically. A rep three months into ramp does not close at the rate of a rep in year four. And the calendar is not flat. Some of the sixty-six weekdays between October 1 and December 31 are Thanksgiving. Some are the week the building empties out and nobody signs anything.

A plan that treats every rep-week as worth the same as every other rep-week is wrong before the quarter starts, and it stays wrong quietly, until the wrongness shows up as a miss in December. The fix is not a better guess at the same calculation. It is pricing the quarter in the unit that actually produces revenue.

The calendar is smaller than it looks.

Q4 2026 runs from Thursday, October 1 through December 31. That is 66 weekdays, 13.2 nominal weeks if you divide by five and stop there. Stop there and you have already overstated the quarter.

Start subtracting. Company holidays take 4 days: Thanksgiving on November 26, the Friday after on November 27, Christmas Eve on December 24, Christmas Day on December 25.

Then the freeze at the back of the quarter. December 21 through 31, excluding the two holidays inside it, is 7 weekdays in which essentially no new business closes at a company with a December shutdown. Those 7 days go into the ledger at zero for new-business capacity. Be honest that you are doing it. They are not zero for renewals and they are not zero for building next quarter's pipeline, but for this quarter's number they are zero. Why the close weeks belong to the next quarter more than this one is the subject of a separate piece. Here the only thing that matters is the count.

Then Thanksgiving week itself. Monday through Wednesday, November 23 to 25, run at roughly half rate ahead of the holiday. Call it 1.5 days lost, not 3.

Add it up. 66 minus 4 minus 7 minus 1.5 is 53.5 weekdays, 10.7 usable weeks. The quarter you are planning against is 10.7 weeks, not 13.2. That is a 19 percent haircut before a single rep's PTO comes out of the number, and Q4 is exactly when people burn use-it-or-lose-it days. The per-rep number is smaller still.

The ledger, row by row.

Once you have the team's usable weeks, build one row per rep, four columns.

1. Usable weeks. Start at the team baseline, 10.7 weeks for Q4 2026, and subtract that rep's actual booked and expected PTO.
2. Output per usable week. Trailing four quarters of closed-won, divided by the usable weeks in those quarters, not the calendar weeks. Compare like to like or the number is nonsense.
3. Capacity. Column one times column two. This is the honest number.
4. Assigned. The quota they were actually given.

Column two is where every manager's first objection lands, so be explicit about the two rules that govern it. A rep with fewer than three quarters of history does not have a rate. Do not use their own number. Use the team median per-week output, discount it for where they sit in ramp, and mark the row as an estimate on the page. Writing down that it is an estimate is the point, not a hedge.

And if one deal is more than 40 percent of a rep's trailing total, compute the rate twice, once with that deal and once without, and carry the lower one. A rep whose year was built on one whale does not have a repeatable weekly rate. He has one whale.

Read the ledger, not the total.

Here is a team of six. Q4 number of $4.5M, split evenly at $750K a rep. Kelsey started in August, so hers is the estimated row. Priya has two weeks already booked off in December.

Kelsey's rate is team median discounted for ramp, marked as an estimate.
Priya's week count reflects two weeks booked off in December.

Team capacity comes to $4.216M against a $4.5M number. A gap of $284K, 6.3 percent. That reads as survivable. Every manager has closed a 6 percent gap.

But the total is hiding the shape. Surpluses across Dana, Tom and Andre come to $278K. Deficits across Marcus, Priya and Kelsey come to $562K. The 6.3 percent is one netted against the other, and that netting is only real if you actually move the accounts to make it real. Tom's $179K of headroom does not walk across the floor to Kelsey on its own.

Do nothing and you do not have a 6 percent problem. You have a $562K problem sitting on two reps, and a $278K surplus that goes unsold because nobody assigned it anywhere.

A bad row is a decision, not a stretch.

Name the threshold: any row assigned above 110 percent of capacity is a decision, not a stretch. Marcus at 103 is a stretch. That is what a quota is for. Priya at 125 and Kelsey at 210 are past it, and there are exactly three things a manager can do about a row like that.

1. Move accounts. Take territory from a surplus row and give some of it away. This is the only option that actually puts the $278K to work.
2. Move the number. Reallocate quota inside the team so the rows sum honestly instead of summing by accident.
3. Accept the miss out loud. Write down which rep is going to miss and by how much, before the quarter starts, so nobody is surprised in January and nobody's comp gets renegotiated after the fact.

Pick none of them and you have picked the third one anyway, without telling anyone. A rep who was handed an impossible number and finds out in December is a rep you lose in January.

What it is for in January.

The ledger's first job is the plan. Its second job comes in January. Put actual output per rep-week next to what the ledger predicted. If the per-week rate held and the team still missed, the calendar assumption was wrong and you overestimated usable weeks. If the calendar held and output per week came in low, the productivity assumption was wrong. Two different problems, two different fixes, and without the ledger you cannot tell them apart.

Run it every quarter and column two gets more accurate every time. That is the compounding part. The first ledger you build is a guess with structure. The fourth one is a measurement.

Timing is the rest of it. Run this in the first week of September and every option is still open. Run it on October 15 and you are not planning anymore. You are writing a report.

The number itself never lies. Divided by headcount, it just does not say much. Divided by usable rep-weeks, it tells you exactly where the quarter is going to break, in time to do something about it.