A Rate Is an Average
Every labour rate is an average over people, over time and over work. What each averaging hides, and when the hidden part changes the answer.
A labour rate is an average taken three times over: over the people in the group, over the periods in the year, and over the kinds of work they do. Each averaging hides something, and each hidden thing changes some decision somewhere.
The calculation in “A Rate Is an Average” depends on reliable source records as well as a documented accounting rule. For organisations researching how to handle multiple clients, check this implementation resource can supply consistent time and project context, while the allocation basis and financial interpretation remain decisions for the organisation and its advisers.
The rate is still worth having. What is not worth having is forgetting that it is an average.
Additional context for “A Rate Is an Average” is available from Stripe business resources. Current legal, payroll, tax and accounting requirements should be confirmed for the relevant jurisdiction.
Averaging over people
A group rate conceals that the most expensive person may cost half as much again as the cheapest, and that the mix of who does the work varies between jobs.
- Over people: the mix of who does the work varies by job.
- Over time: bonuses, awards and absence do not fall evenly.
- Over work: overhead-heavy work is subsidised by the rest.
A job done entirely by senior people costs more than the rate says. One done by juniors costs less.
When that matters
When the mix is deliberate. Pricing a job that will be staffed entirely by seniors at the blended rate is pricing it at a loss, and the loss is systematic rather than occasional.
Where mix varies materially between job types, rates by grade are worth the administration.
Averaging over time
A rate built on an annual cost ignores that the cost is not evenly incurred: bonuses land, awards land mid-year, absence clusters.
Accrual deals with most of this and is the subject of its own note. The residue is small and it should be known to be small rather than assumed to be.
Averaging over work
The largest of the three. A rate built on a department's total chargeable hours spreads the cost of its most overhead-heavy work across its lightest.
Where one product line requires twice the supervision, the rate subsidises it from the other lines.
The test for all three
Compute the rate for a sub-group — one grade, one product line, one team — and compare it to the blended figure. If the difference is small, the average is safe.
If it is twenty per cent, the blend is making commercial decisions.
The marginal versus average confusion
A rate is an average cost. The cost of one more hour, from people already employed, is a marginal cost and it is much lower.
Using the average where the marginal applies, or the reverse, is the subject of its own note later here, and it is the most expensive confusion in the whole discipline.
The ranking trap
Comparing two jobs' profitability using the same average rate tells you which consumed more hours. It does not tell you which was more profitable if their resource mix differed.
For a ranking to mean anything, the rate must reflect what each job actually consumed.
What the average is good for
Pricing standard work. Budgeting. Comparing periods. Setting a reference point.
Three of those four are its real job and it does them well, which is why the answer is not to abandon averages.
How precise to be
Precision costs maintenance and credibility. Three or four rates that people understand beat twenty that only the author can explain.
The useful question is which distinctions change a decision, and the answer is usually two or three.
Saying it on the rate card
"This is an average across grades and work types within the department." One sentence.
It is the sentence that stops the rate being used for the three or four things it cannot support.
Averages and the small job
The smaller the job, the less an average describes it. A 400-hour job averages out the mix; a 6-hour job is whoever happened to do it.
Which means job costs below some size should be read as indicative rather than measured, and margin decisions about small work should be made on job type rather than on individual outturns.
When to stop adding rates
When the next distinction would not change a decision. That test is easy to state and is rarely applied, because adding a rate always feels like adding accuracy.
Twenty rates maintained badly are worse than four maintained well, and the maintenance burden is what determines whether any of them are right a year later.
The average that is right for the wrong reason
A blended rate can match the true cost of a particular job exactly, by coincidence of mix. That does not make the blend correct; it makes that job unrepresentative.
Checking a rate against one job proves nothing either way, which is why the test in this note is a comparison of sub-group rates rather than a comparison of outturns.
Where the exceptions should live
As named exceptions, not as a more elaborate model. "Work of type C is costed at 1.3 times the standard rate because it requires continuous supervision."
One line per exception, reviewed annually. That is a model anybody can operate and argue with, which is the only kind that survives.