Construction & Contracting// definition

What a unit cost library needs to carry before it can price a second job

In short

A unit cost library is a set of records that pair a rate with the conditions under which it was true: crew composition, production basis, region, date, job type and a stated inclusion boundary. The rate itself is the least transferable part of the record. Strip the qualifiers away and what remains is a number somebody once achieved on a job you cannot identify.

Key takeaways

  • A unit rate is derived from a labour rate and a production rate. Store the inputs or you cannot re-derive it when wages move.
  • Crew-hour and man-hour are different denominators, and confusing them produces errors of 3x that survive review.
  • The inclusion boundary is the field that decides whether two rates can be compared at all.
  • A rate with no date and no source job is not evidence; an estimator's memory is more honest about its own uncertainty.
  • Record the spread across jobs alongside the rate, so the estimator knows which numbers to lean on.

A unit cost library is a set of records that pair a rate with the conditions under which that rate was true. The conditions are the reusable part. A contractor who keeps 400 rates and no context has an archive; a contractor who keeps 60 rates with crew, production basis, boundary, market and date has something that can price a job it has never seen.

Two objects sit in the library. A cost item is the smallest priced thing — one unit of one operation, with one unit of measure. An assembly is a bundle of cost items that are always bought together, priced per unit of the thing being built: a partition assembly holds studs, track, board, tape, insulation and fixings, and produces one rate per square metre of wall. Assemblies save time and hide detail, which is the trade they exist to make.

Three rates people call the same thing

RateWhat it measuresUnitWhere it comes from
Labour rateThe all-in cost of an hour of a trade, including burdenPer hourPayroll and the agreement in force
Production rateHow much a defined crew installs in an hourUnits per crew-hourYour own closed jobs, or a published reference
Unit rateThe installed cost of one unit, all inputs combinedPer unit of measureDerived from the two above plus material and plant
Labour rate, production rate and unit rate are different objects with different lifespans

The unit rate is derived, and libraries that store only the derived number age badly. When a wage agreement moves 6 percent, a library holding labour rates and production rates separately reprices itself; a library holding composite unit rates has to be rebuilt by someone guessing at the labour share of each one.

The context that makes a rate transferable

  • Crew composition. The rate assumes a specific gang. A 4-person crew rate applied to a 2-person crew is a different production rate wearing the same number.
  • Production basis and its denominator. Units per crew-hour, stated as such, with the crew defined on the same record.
  • Region and market conditions. Labour availability and travel change the rate more than material does on most self-performed scope.
  • Date, and what it is a date of. The month the work happened, not the month someone typed the record in, because escalation is applied from the former.
  • Job type and size band. A rate from a 1,200 m² fit-out does not transfer to a 40 m² one; mobilisation and learning curve dominate at small quantities.
  • Access and working conditions. Height, occupied premises, night work and restricted lay-down are the qualifiers most often left out and most often decisive.
  • Inclusion boundary. What the rate covers — labour, material, plant, consumables — and explicitly what it does not: taxes, overhead and profit, scaffold, temporary works, waste removal.
  • Source and confidence. Job reference, cost code, the actual quantity the rate was derived over, and how many jobs support it.

One cost item, written out in full

FieldValue
Code and description03 30 53 · slab on grade, place and power-float finish
Unit of measurem² of finished slab area, no deduction under 1.0 m²
Crew1 foreman, 4 finishers, 1 pump operator
Production basis38 m² per crew-hour, gang basis
InclusionsPlacing, finishing, curing compound, power float and trowel
ExclusionsConcrete supply, reinforcement, formwork, pump hire, testing
SourceJob 2411-B, cost code 03300, derived over 1,240 m² actual
Period and marketSecond quarter 2026, metro market, day shift
Confidence3 jobs, spread of 12 percent between highest and lowest
A single cost item record with its qualifying fields

The confidence row is the one contractors skip and estimators want most. A rate supported by 3 jobs with a 12 percent spread and a rate supported by 1 job are used differently by anyone competent, and a library that cannot tell them apart forces the estimator to distrust all of it equally.

A rate with no date, no boundary and no source job is not data. It is a number that once happened, and it will be wrong in a direction nobody can predict.

Why most contractor libraries stop being used

Almost never because the numbers were wrong. Three failures do the damage, and all three are missing-context failures. A rate lifted from a job with unusual access gets applied to a normal one. A rate whose inclusion boundary differs from the estimate's structure gets added to a line that already covers part of it, and the scope is priced twice. A rate with no date gets escalated by an estimator who assumes it is old, or not escalated by one who assumes it is fresh.

The boundary problem is the same one that runs down the subcontractor side of the job, where two quotes cannot be compared until somebody writes down what each was asked to include. That is the work a scope sheet does before bids arrive, and a cost library is the internal version of the same discipline.

Scope you cannot price at all is not a library problem and should not be forced into one. It becomes an allowance with an owner and a reconciliation date rather than a rate with a guess attached. And a library earns its keep fastest on jobs with alternates, where the same rates get applied across parallel scope branches — pricing a deduct alternate is nearly free when the rates carry their own boundaries, and a day of work when they do not.

One last caution about where accuracy actually comes from. Quantity precision and rate precision are independent, and improving one does nothing for the other: a flawless count that meets a rate with the wrong inclusion boundary produces a confidently wrong number. That is worth remembering when an extraction tool is being evaluated on counting alone, as in the automated count that finds devices on some sheets and none on others. We usually build the library side as a small internal system with an owner and a review cycle rather than a spreadsheet, which is the shape of work described in MVP and product builds. The rest of the silo sits under preconstruction, takeoff and estimating, within our wider construction and contracting work.

Frequently asked questions

Short answers to the follow-ups this page tends to raise.

What is a unit cost database in construction estimating?

It is a store of rates for units of work — one square metre of slab, one metre of conduit, one door installed — with the context each rate depends on recorded alongside it. It differs from a price list in that the record is designed to be interrogated later: which crew, which production basis, which market, which date, and what the rate did and did not include.

What is the difference between a cost item and an assembly?

A cost item prices one operation in one unit; an assembly bundles several cost items that are always bought together and prices them per unit of the finished element. Assemblies make estimating faster and make variance analysis harder, because a single assembly rate hides which of its components moved. Most contractors keep both and use assemblies for conceptual estimates, items for hard bids.

How many past jobs does a rate need before it is trustworthy?

Three comparable jobs is a reasonable working threshold, and the spread between them matters more than the count. Two jobs 4 percent apart tell you more than five jobs ranging over 40 percent, because a wide spread means the record is missing a qualifier that explains the variation — usually access, crew size or quantity band. Store the spread so the estimator can see it.

Should overhead and profit sit inside the unit rate?

No — keep them out of the rate and apply them as a separate, visible step. Rates that carry embedded markup cannot be compared against a subcontractor quote, cannot be escalated cleanly, and quietly compound when an assembly built from marked-up items is itself marked up. The exception is a genuinely all-in rate used for conceptual estimating, which should be named as such in the record so it never gets mixed into a hard bid.

  • unit rates
  • cost library
  • estimating
  • preconstruction
// shipped work

The work behind this page

Builds from our portfolio that this page draws on.

Working on something in this space?

Tell us where you are in a sentence or two. We'll tell you honestly whether we're the right team, and what a sensible first slice of the work looks like.

Start the conversation