Methodology

Unit-Rate Estimating, Markup and Contingency

Why a 20% markup is not a 20% margin, why an estimate without a range is the wrong shape of answer, and why contingency should shrink as a design develops rather than staying at ten per cent forever.
  • 7Sections
  • 1Equations
  • 17Calculators

Quantity times rate, and all the risk is in the rate

Every cost calculator here reduces to the same expression: measure a quantity, multiply by a rate, add the rates that do not depend on quantity. The measurement can usually be done to within a few per cent. The RATE cannot, and that is where the whole of the uncertainty lives.

A unit rate is a bundle. Inside a single figure for a square metre of blockwork sit the blocks, the mortar, the labour to lay them, the waste allowance, the scaffold, the plant, a share of the site overhead and a share of the head office. Two rates that look comparable can include different things — and a comparison between a rate that carries preliminaries and one that does not is not a comparison at all.

This is why the first question about any published rate is what is in it and what is excluded, and why estimating standards specify the composition of a rate rather than only its value. A calculator that asks for a rate is asking the user to supply the hardest number in the problem, and every page here says so.

P=C⁢(1+m),margin=m1+m,m=margin1−margin
Markup is applied to COST; margin is measured against PRICE. They are different numbers and each converts into the other.
C
total cost of the work
P
price charged
m
markup, as a fraction of cost
margin
profit as a fraction of the PRICE — always smaller than the markup

Markup is not margin, and the gap is money

This is the most consequential arithmetic on the page, and it is the one most often got wrong in practice. MARKUP is a percentage added to cost. MARGIN is profit as a percentage of the price. They are never equal, and the markup is always the larger number.

Work an example. Cost one hundred, add twenty per cent markup, price one hundred and twenty, profit twenty — which is 16.7 per cent of the price, not twenty. A contractor who applies a twenty per cent markup believing they are earning a twenty per cent margin is short by more than three points, on every job, permanently.

Running it the other way: to EARN a twenty per cent margin the markup has to be twenty-five per cent. For a thirty per cent margin it is about forty-three per cent. The two diverge faster the higher the target, which is why the error grows precisely in the specialist trades that need the highest margins.

The cure is to decide which one the business is managed on and convert at the point of pricing rather than hoping they are close. The calculators here state which they are returning, because a figure labelled only "20%" is ambiguous in a way that costs real money.

An estimate without a range is the wrong shape of answer

A cost per square metre applied to a floor area produces a single number, and a single number implies a precision the method does not have. Estimating standards handle this by CLASSIFYING estimates: an order-of-magnitude figure produced from an area and a rate carries an expected accuracy of something like minus thirty to plus fifty per cent, and a definitive estimate built from a measured bill with priced quotations something like minus five to plus ten.

What moves an estimate between those classes is not effort — it is DEFINITION. An estimator cannot narrow the range by working harder on an undefined scope; the range narrows when the design decides what is being built, what the ground is, and what the specification says. This is why a client pressing for accuracy early is asking for the wrong thing, and why the honest answer to "what will it cost?" at concept stage is a band with the assumptions written next to it.

It also explains why two estimates for the same project can differ substantially and both be correct. They are different scope assumptions expressed as money, and reconciling them is a scope exercise rather than an arithmetic one.

Contingency is for the unknown parts of a known scope

Contingency exists to cover things that will happen within the defined scope but cannot yet be identified individually — the detail that turns out harder, the quantity that comes in over, the coordination nobody has drawn yet. It is not an allowance for work that has not been decided, and it is not a discount fund.

Because it covers definition risk, contingency should FALL as the design develops. A figure that is appropriate at concept is excessive at tender, and carrying the same ten per cent from first sketch to final account means it was wrong at one end or the other. A project that never spends its contingency did not manage risk well; it priced badly.

Two distinctions keep it honest. Contingency belongs to the estimate and is expected to be spent; a MANAGEMENT RESERVE sits above the estimate for scope changes and is not. And an allowance for a specific known item with an unknown cost — a provisional sum for a drainage connection, say — is a line in the estimate rather than part of the contingency, because it is a known unknown with a name.

Adding a percentage to an already conservative build-up is the common way to get this wrong. If every rate carries its own padding and a contingency is then applied on top, the project is priced twice for the same risk and loses the tender.

Labour is the volatile term, and the wage is not the cost

Material prices are visible, quoted and comparable. Labour is neither, and it is where estimates diverge. The quantity that matters is an OUTPUT RATE — how much of the work one person completes in an hour — and it moves with access, repetition, weather, crew composition, and how many times the work has been done before.

Repetition in particular is systematic rather than random. The first bay, the first bathroom, the first floor of a repeating layout take materially longer than the fifth, and a rate derived from a finished project already contains that average. Applying it to a small job with no repetition under-prices the labour, which is the usual reason a small contract loses money on the same rates a large one made money on.

The cost of an hour is also not the wage. Employer taxes, insurance, holiday and sick pay, pension, training, tools, protective equipment and non-productive time are all real and are conventionally expressed as a BURDEN on the base wage — commonly adding somewhere between a quarter and a half. A labour calculator fed a bare hourly wage returns a number that no employer has ever paid.

Preliminaries scale with time, not with quantity

Site establishment, supervision, welfare, temporary services, scaffolding hire, plant standing, insurance and the site manager's salary are not proportional to the amount of work. Most of them are proportional to how LONG the site is open, and the rest are fixed at the start and end.

That split — time-related against fixed — is the whole reason preliminaries are estimated separately rather than being spread into the rates. It means a programme that extends by twenty per cent adds twenty per cent to the time-related portion even though the quantity of work has not changed by a single unit, and it is the mechanism by which a delay becomes a cost.

It also means that spreading preliminaries into unit rates, which is common in domestic pricing, makes the estimate wrong in a specific direction: a job whose scope is reduced keeps almost all of its preliminaries, so the price should not fall by the full proportion of the work removed. A client asking to "take out ten per cent of the work to save ten per cent" is asking for something the cost structure will not deliver.

A rate is for a place and a date

Published and historical rates are observations of a market, and markets differ by location and move over time. Both are handled by factors — a location index against a national base, and an escalation index from the date of the source to the date of the work — and both are approximations that are worth applying because the alternative is to assume they are one.

Escalation compounds, which makes the date of the source matter more than people expect on a long programme. A rate two years old on a project completing in three years is being asked to predict five years of movement, and over that span the index is doing more work than the estimator is.

Which sets the limit of everything on this page. These calculators produce order-of-magnitude and comparison figures: useful for testing whether a quote is plausible, for comparing options, and for setting a budget with a band around it. They are not a priced bill of quantities, they are not a quotation, and no result here is a commitment by anybody to do the work for that money.

Calculators that use this method

Basis

  • AACE International Recommended Practice 17R-97 and 18R-97, Cost Estimate Classification System — the five classes and the expected accuracy range of each.
  • RICS New Rules of Measurement: NRM1 for order of cost estimating and the treatment of contingency and risk allowances, NRM2 for detailed measurement and the composition of unit rates.
  • CIOB Code of Estimating Practice, for the build-up of a unit rate and the separation of preliminaries from measured work.
  • RSMeans Building Construction Cost Data and comparable published rate books, including their location factors and historical cost indices.
  • Wright, T.P. (1936), Factors Affecting the Cost of Airplanes — the origin of the learning curve applied here to repeated construction operations.
  • Bureau of Labor Statistics employer cost for employee compensation, and equivalent national series, for the burden applied to a base wage.
Cite this page