Six Person-Months Is a Bill, Not a Date
Proposals are sized in person-months because that is the unit an estimator can defend in a meeting: three people, two months, six person-months. The moment the estimate reaches procurement it stops being a schedule and becomes a quantity to be priced, and rate cards are written per hour. Somebody then has to say how many hours are hiding inside "six person-months" — and the honest answer depends entirely on which of several unstated conventions the estimator had in mind.
Where the Ambiguity Comes From
One word doing two jobs
Elapsed hours set the ceiling
Money is denominated per hour
Two bids, two definitions
From a Sized Estimate to a Priced One
The sequence below turns an internal sizing exercise into a number a finance team will accept, without quietly changing what was estimated.
Enter the calendar span first
Put the elapsed duration — 1, 6, 18 — in the left field and read the hours it contains. This is the outer boundary of the engagement, the figure that tells you a "6-month, 4 000-hour" proposal is claiming almost round-the-clock occupancy from one person.
Pick the labour convention you are actually quoting
Take the hours-per-person-month figure from the table below rather than the calendar number, and write it into the proposal. An estimate whose basis is stated cannot be renegotiated by reinterpretation later.
Reverse it when the hours arrive first
Timesheet exports and rate-card ceilings come in as hours. The swap button (↔) turns the page into h → mo, so 2 400 logged hours can be reported back as the fraction of a calendar month or year it represents.
Copy the bare number into the pricing sheet
The copy control on each field hands over digits with no unit and no separators, which is what a spreadsheet cell needs before it multiplies by a rate. Ctrl + C inside the field does the same.
What One Person-Month Buys Under Each Convention
Six conventions in circulation, each defensible, each producing a different price for the identical piece of work. The right-hand column is the same six-person-month estimate restated under each one.
| Convention | What it assumes | Hours per person-month | A 6-person-month estimate |
|---|---|---|---|
| Calendar month (this converter) | Elapsed wall-clock time, 30.436875 days | 730.485 h | 4 382.91 h |
| 261 working days a year | 21.75 days a month at 8 h, no leave deducted | 174 h | 1 044 h |
| 2 080-hour year | 40 h × 52 weeks, divided by twelve | 173.33 h | 1 040 h |
| COCOMO person-month | 19 days at 8 h, holidays and sick days removed | 152 h | 912 h |
| 1 720-hour year | Statutory leave and public holidays deducted | 143.33 h | 860 h |
| Billable at 75 % utilisation | 2 080-hour year minus internal and bench time | 130 h | 780 h |
Ignore the first row for pricing — it is there to show the ceiling — and the remaining five still span 780 to 1 044 h, a third of difference on identical scope. That spread is why a proposal should carry its hours-per-person-month assumption in writing rather than leaving the reader to supply one.
What This Pair Does While an Estimate Is Being Argued
Sanity-check a quote mid-call
Both fields update as you type, so an hours figure that has just been read out loud can be set against the elapsed hours a month actually contains before the call moves on.
Timesheets read back the other way
Swapping the direction turns logged hours from an actuals export into months of elapsed engagement, the form a steering-committee slide wants.
Quarters and years without leaving the page
The searchable dropdowns hold every time unit in the app, including a quarter of three months, so a multi-year framework agreement can be sized in one place.
Five-figure totals stay legible
Thousands are spaced apart in the output, so 17 531.64 for a two-year programme is read correctly instead of being mistyped into a pricing model.
Questions That Decide Whether an Estimate Survives Review
What does a vendor actually mean by one person-month?
Ask, because there is no standard. In practice it lands between 130 and 174 h depending on whether leave, public holidays and non-project time have already been deducted. The safest wording in a statement of work is the one that never uses the unit alone: "6 person-months, defined as 152 hours each, totalling 912 hours." Once the hour figure is on the page, a disagreement about scope stays a disagreement about scope instead of becoming an argument about vocabulary.
Should the estimate be built on paid hours or productive hours?
Estimate in productive hours and price in paid hours, keeping the two visible separately. A salaried person is paid for roughly 173 h a month but delivers project work in perhaps 110–130 of them once meetings, hiring, support rotations and training are taken out. Estimating in paid hours quietly assumes someone who does nothing but the estimated task; estimating in productive hours and then buying paid hours at the same number leaves the overhead unfunded. Carry the utilisation assumption as its own line so it can be challenged.
Where do public holidays and annual leave disappear to?
Into whichever party did not write the definition. A year with 25 days of leave and 9 public holidays loses 34 working days — about 272 h, or 13 % of a 2 080-hour year. If the estimate used 173.33 h a month and the delivery calendar contains that leave, the programme is short roughly six and a half weeks of capacity per person per year. It is also why identical headcount costs different amounts across countries: the statutory leave entitlement changes the divisor, not the salary.
How do I turn person-months into a rate-card price?
Multiply the person-months by the agreed hours per person-month, split the result across the roles the work needs, then apply each role's hourly rate. Six person-months at 152 h is 912 h; split 60/40 between a senior and a junior role it becomes 547 h and 365 h, and only then is the price meaningful. Skipping the role split and applying one blended rate hides the fact that a change of staffing mix changes the cost without changing the estimate.
If it is 12 person-months, can 12 people finish it in one month?
Almost never, and the reason is the oldest observation in software management: effort and duration are not interchangeable. Fred Brooks put it as "the man-month is a fallacious and dangerous myth" — people and months trade off cleanly only when a task can be partitioned with no communication between the parts. Add people and the communication paths grow roughly with the square of the headcount, while newcomers consume the time of those already productive. The arithmetic in the table converts effort into hours; it says nothing about how few weeks those hours can be compressed into, and treating a person-month figure as a deadline is how a late project gets later.
No comments yet. Be the first to comment!