Daily attendance is a distribution, not an average: the desk count comes from the busiest weekday and the service level you choose.
Set the desk-sharing ratio from the busiest weekday and the chance you are willing to run out, not from the average attendance. For an illustrative 200 people coming in three days a week, with Tuesday and Wednesday at 75 per cent, a desk for everyone on 95 per cent of peak days takes 160 desks, a ratio of 0.80, and leaves about 3.8 days a year short of desks. Sizing on the 60 per cent average gives 120 desks and a shortage on 138 days a year.
Worked in full in Office Real Estate by Julian R. Sterling, with every figure reproduced in a free workbook.See the book on Amazon →
The ratio is the single number that turns a hybrid policy into floor space, and every desk carries about 10,449 a year of rent, service charge, rates, fit-out and running costs on the companion office case. Too high and the occupier pays for empty desks; too low and the office fails on exactly the days the policy asks people to be there. The calculation that settles it is a probability, not an average.
| Input | Value |
|---|---|
| Headcount | 200 |
| Attendance probability, Monday | 50% |
| Tuesday / Wednesday | 75% / 75% |
| Thursday / Friday | 70% / 30% |
| Average attendance (3.0 days a week) | 60% |
| Office weeks a year | 46 |
| Target: everyone has a desk on the peak day | 95% of the time |
| Space per desk, all-in cost per m² | 11.08 m², 943.04 |
The cost per desk is 11.08 × 943.04 = 10,449 a year: 11.08 m² per desk including meeting, support and circulation space, from the space budget in how much office space you need per employee, and 943.04 per m² all-in from how much an office costs per desk per year. The weekday pattern is illustrative but a common shape: a busy middle of the week and quiet edges.
If each of the 200 people decides independently whether to come in on a Wednesday with probability 75 per cent, the number who turn up follows a binomial distribution: mean 200 × 0.75 = 150, standard deviation √(200 × 0.75 × 0.25) = 6.12. Some Wednesdays bring 144, some 156. A floor with 150 desks is short on 47.3 per cent of Wednesdays.
Desks = the smallest k with P(attendance ≤ k) ≥ 95%
Wednesday, n = 200, p = 0.75: k = 160
Excel: =BINOM.INV(200,0.75,0.95) returns 160.
Normal shortcut: 150 + 1.645 × 6.12, rounded up, gives 161; the exact binomial is one desk lower here.
Ratio = 160 ÷ 200 = 0.80, or 1.25 people per desk.
The peak day sets the answer. Monday needs 112 desks at the same service level, Thursday 151 and Friday 71, so any of the quieter days can be served by a floor sized for Wednesday.
The service level on one day is not the number a facilities manager hears about. The complaints come from the total of short days over the year: the probability of a shortage on each weekday, summed, times 46 weeks.
| Desks | Ratio | Wednesdays short | Short days a year | If teams cluster | Annual cost |
|---|---|---|---|---|---|
| 120 | 0.60 | 100.0% | 138.0 | 138.0 | 1,253,866 |
| 150 | 0.75 | 47.3% | 45.8 | 52.5 | 1,567,332 |
| 155 | 0.78 | 18.5% | 17.4 | 32.1 | 1,619,577 |
| 160 | 0.80 | 4.1% | 3.8 | 16.9 | 1,671,821 |
| 165 | 0.82 | 0.4% | 0.4 | 7.5 | 1,724,066 |
| 170 | 0.85 | 0.0% | 0.0 | 2.8 | 1,776,310 |
| 200 | 1.00 | 0.0% | 0.0 | 0.0 | 2,089,777 |
At 160 desks the office runs 417,955 a year below a desk each, 8,359 per employee against 10,449. On an average day it is 75.0 per cent occupied and on a Friday 37.5 per cent, which is what a 95 per cent service level on the busiest day costs.
The binomial assumes every person decides alone. They do not: teams agree anchor days and come in together, which widens the spread of daily attendance. A standard way to allow for it is a design effect, 1 + (team size − 1) × correlation. With teams of eight and a within-team correlation of 0.30, the effect is 3.10, and the Wednesday standard deviation rises from 6.12 to 10.78.
The same 95 per cent target then needs 168 desks, a ratio of 0.84 and eight more desks than the independent case. Kept at 160, the office is short on about 16.9 days a year instead of 3.8. The correlation is the input most worth measuring from badge data before the lease is signed: a few weeks of daily counts give both the mean and the spread.
Hold the average at 60 per cent but make every day equal and the peak-day requirement falls to 131 desks, 29 fewer than with a Tuesday-to-Wednesday peak, worth 303,018 a year at 10,449 a desk. That is the value of an attendance rota, and it is a saving that needs no change to the floor plate. The size of the floor around those desks follows from how much office space you need per employee.
Desks = BINOM.INV(headcount, peak-day probability, service level), checked against the clustering seen in badge data. Here 160 desks, a ratio of 0.80, with 168 if teams come in together. For a landlord the same arithmetic forecasts how much space a hybrid tenant hands back at expiry, which feeds the re-letting cycle in the free workbook for this case.
It depends on the busiest day, not the average. With 200 people and a 75 per cent Wednesday, 160 desks, a ratio of 0.80, give everyone a desk on 95 per cent of Wednesdays. The same policy spread evenly at 60 per cent every day needs only 131 desks. A ratio copied from another firm without its weekday pattern is not transferable.
Treat each day's attendance as a binomial draw and take the percentile you want to serve. In Excel, BINOM.INV(200, 0.75, 0.95) returns 160. Then allow for teams coming in together, which widens the spread: with teams of eight and a correlation of 0.30 the answer rises to 168 desks.
At an illustrative 10,449 a desk a year, 160 desks instead of 200 save 417,955 a year, from 10,449 to 8,359 per employee. Going further, to 150 desks, saves another 104,489 but leaves the office short on 47.3 per cent of Wednesdays, about 45.8 days a year.
This article is one calculation from Office Real Estate. The book takes the same case from first principles to the decision, chapter by chapter, and every figure it prints is a live formula in the free companion workbooks.
Get the book on Amazon →Free companion files
Also on Amazon UK · Amazon Germany · Amazon France · Amazon Canada
Reading guide: real estate investing, finance and fund management → · All 1682 articles →
If this book helped, or didn’t, a few lines on Amazon are worth more than they look: they are what the next reader goes on. Write a review. The workbook stays free either way.