RSMeans Cost Data Explained for Contractors

What is RSMeans cost data?

RSMeans began as the R.S. Means Company and is published today by Gordian under the RSMeans data name. The core of it is a large set of unit cost line items covering most common building construction scopes, organized by CSI MasterFormat. Each line carries a material cost, a labor cost, an equipment cost, the crew assumed to do the work, and how much that crew installs in a day.

It ships two ways. The annual cost books are split by scope — building construction, mechanical, electrical, plumbing, site work, facilities, residential, repair and remodeling, assemblies, square foot costs. The subscription database carries the same data, updates more often than the print editions, and connects to some takeoff and estimating platforms so line items can be pulled straight against a takeoff.

The framing that matters: RSMeans is a cost model, not a price list. It does not report what anything was actually bought for on a specific job. It reports what a modeled crew, buying at a modeled price, working at a modeled productivity, would cost. Knowing the assumptions is the entire difference between using it well and being quietly wrong on every line.

How is an RSMeans unit cost line structured?

Every unit cost line uses the same columns. Reading them correctly is most of the skill.

What do crew codes and daily output actually mean?

The crew code defines who is on the wall, not just a rate. The crew table in the reference section gives each crew's exact composition — trades, headcount, foreman, specific equipment — along with its bare hourly cost and its cost per labor-hour. If the crew you would actually send is smaller, mixed-trade, or working off different equipment, the labor cost on that line is describing someone else's operation. Check the crew before you accept the number, particularly on equipment-heavy work where one machine changes the whole line.

Daily output is the assumption most worth arguing with. Daily output assumes a normal day: reasonable access, average job size, standard hours, material staged nearby, a crew that has done this work before. Occupied buildings, phased work, restricted laydown, night or shift work, high-rise material handling, security screening and small scattered quantities all cut output, and none of it is in the line. Halving daily output doubles the labor cost of that item — and the equipment cost with it, because equipment is priced by the crew-day and spread across the same output. On equipment-heavy lines that second effect is the larger of the two. On labor-driven lines, nothing else in the line has that kind of leverage, which makes it the first thing to adjust and the first thing a reviewer should ask about.

Labor-hours is the bridge to your own numbers. If you track job cost at all, you already know your real labor-hours per unit for the work you self-perform. Substituting your figure while keeping the rest of the line's structure is one of the most valuable adjustments available, and it compounds — after a few completed jobs you have a trade-by-trade correction factor that makes the whole database behave more like your company.

Bare cost versus Total Incl O&P is where double-counting happens. If you apply your own overhead and profit at the estimate level, build the estimate on bare costs — but do not stop there. RSMeans bare labor is base wage plus fringes only; payroll taxes, unemployment insurance, workers' compensation and public liability are bundled into the labor O&P markup along with overhead and profit. Those burdens are direct costs, not markup. Strip the line back to bare, add your own labor burden by trade, then apply your overhead and profit once on top. Pulling Total Incl O&P and adding markup over it prices you out of work; pulling bare and applying only a company markup leaves the burden out entirely and loses you money on the jobs you win. Both are quiet errors because no single line looks absurd — the estimate is just consistently off in one direction, and you either lose bids or win them at a loss without ever finding out why.

How do city cost indexes localize the numbers?

The published costs are national averages. RSMeans also publishes location factors — city cost indexes — for a large number of US and Canadian locations, scaled so that the national average equals 100. The current edition's location factor table is the only authority on which locations are covered. You multiply the national cost by the index divided by 100. An index of 112 means costs in that location run twelve percent above the national average; an index of 91 means nine percent below.

Each location gets separate material and installation indexes as well as a weighted total, and the indexes are also broken out by MasterFormat division. The split matters because material and labor do not vary the same way across the country. Bulk commodities such as gypsum board and structural steel tend to vary less by region than labor does, but freight, local supply and fabrication capacity still move them, sometimes sharply. The wage rate for a sheet metal worker in one metro versus another does not behave that way at all.

The practical consequence is that you should use the division-level index that matches the work in front of you, not the weighted total, whenever the estimate is not a whole building. The weighted total is weighted for a typical mixed building. An electrical-only estimate or a sitework package is not a typical mixed building, and running it through the total index imports the wrong labor market.

Location factors go down to three-digit ZIP prefixes, but a ZIP prefix is not a jobsite. A project well outside the indexed metro draws labor differently, costs more to deliver to, may sit in a different bargaining jurisdiction, and may have a completely different subcontractor pool. The index is a starting adjustment, not a finished one.

Separately, a historical cost index lets you escalate or de-escalate between editions. If you are working from a book that is not current, escalate it rather than using it as printed. And whatever you do, record the edition, the release date and the location factor applied on the estimate itself — a year later, when someone asks why a number was what it was, that record is the only thing that answers them.

Unit costs, assemblies, or square foot models — which do you need?

RSMeans publishes the same underlying data at three levels of aggregation. Picking the wrong level is a common and expensive mistake.

Square foot models — for when there are no drawings. Model buildings by type, each with a cost per square foot and a breakdown of where that cost sits, adjustable for building area, perimeter and story height. This is order-of-magnitude budgeting for feasibility work. The perimeter adjustment is the easiest one to skip and it can move the number sharply — a long thin footprint and a square one of the same area do not carry the same enclosure cost per square foot. Cost per square foot is not linear with size, which is why the models carry a size adjustment — a small warehouse and a large one do not cost the same per foot, and ignoring that is how early budgets go wrong before design even starts.

Assemblies — for design development. Assemblies are organized by UNIFORMAT — substructure, shell, interiors, services, equipment, sitework — rather than by trade. An assembly bundles the components of a system into one price per unit of that system: a wall assembly priced per square foot of wall, covering studs, sheathing, insulation, board and finish. You can price a schematic set with assemblies before anyone has drawn a single detail, and you can compare two design options in an hour.

Unit costs — for a real estimate off a real set. The line-item data. It prices quantities and nothing else, which means it requires an actual quantity takeoff first. This is where RSMeans does the work you would otherwise pay an estimator or a trade partner for — and it is also the level where every assumption discussed above bites hardest.

Match the data level to the design stage. Unit costs applied to a schematic set produce false precision: you have priced a level of detail the drawings do not contain, and the number reads as far more certain than it is. Square foot models applied to a permit set throw away all the information the drawings are carrying. The discipline is to price at the level the design actually supports and say so.

Where is RSMeans genuinely the right tool?

There are situations where a national cost database is not a compromise — it is the correct answer.

• Conceptual and budget estimates produced before there is enough design to take off
• Independent, citable budgets for owners, lenders and public agencies that need a basis they did not produce themselves
• Sanity-checking a subcontractor bid that looks too high or too low, before you build a number around it
• Pricing scope that nobody quoted — the trade that came back with a hole in it, two days before bid
• Change order and claim pricing on conventional contracts, where an independent published basis settles arguments faster than two sets of internal rates — and separately, job order contracting, where the owner issues a unit price catalog and bidders bid a coefficient against it, which may be a customized Gordian task catalog rather than the published book
• Trades you rarely self-perform and therefore have no cost history for
• Facilities maintenance and repair work, where line items are small, repetitive and not worth chasing a quote for individually — priced from the facilities maintenance and repair data or the repair and remodeling data, never from the new-construction lines, which assume open access and repetitive runs that a renovation does not have

Where does RSMeans break down?

Anyone selling you on RSMeans without this section is selling you something. These are real limitations, and every one of them shows up on real projects.

Material costs are a national average purchase price, not your price. What you actually pay depends on your account, your volume, your buying group, freight to your site, and whether the item is stock or a special order. On volatile commodities — lumber, copper, steel, PVC, gypsum — the gap between published data and a live quote can be wide and can move in a matter of weeks. Annual print editions are the worst case. The online data updates more often, but it still lags a number you could get today by picking up the phone.

Labor reflects a wage and productivity model that may not be yours. The standard cost books are built on a union wage model with fringes included — RSMeans also publishes open-shop data, so check which basis the edition in front of you uses before you assume either. The installation index adjusts the overall level for your market, but the crew compositions and productivity assumptions still describe that model. Open-shop contractors can find one trade priced above what they actually pay and another below it inside the same estimate, in which case the total can look reasonable while both components are wrong.

It cannot price a subcontractor's appetite. For bought-out trades — mechanical, electrical, fire protection, elevators, curtain wall — the number you actually pay is set by that sub's backlog, their read on the risk, and how much they want the job. RSMeans tells you what the work should cost. A busy sub in a tight market is not selling at what the work should cost, and a hungry one in a slow market is selling below it.

It does not know what is missing from your takeoff. Perfect unit costs applied to an incomplete takeoff produce a confidently wrong number. The quantities and the scope review are where accuracy actually lives. No cost database has ever caught a missed elevation, an unquantified soffit, or a detail that appears on the architectural set and nowhere else.

Project-level costs are not in the line items. General conditions, sales and use tax on materials, equipment mobilization and demobilization, escalation to the midpoint of construction, bonds, insurance, permits, contingency, and markup on subcontracted work all sit above the line-item layer. Reference tables offer allowances for some of it, but an allowance is not your number. A line-item estimate with no project layer built on top of it is one common reason an estimate comes in suspiciously low.

The data assumes typical quantities, and your job may not have them. Unit costs carry an implied quantity for the item. Small isolated quantities absorb mobilization, setup and crew inefficiency the line does not account for; very large repetitive quantities beat the published rate once the crew finds its rhythm. Both directions need a deliberate adjustment.

Normal conditions are doing a lot of work in that phrase. Winter conditions, occupied facilities, active hospitals, secure sites, tight urban logistics, restricted delivery windows, and multi-shift schedules are all outside the model. RSMeans provides guidance on productivity adjustment, but the judgment is yours and the data will not warn you that you needed to make one.

How do you use RSMeans without getting burned?

The contractors who get value out of it treat it as a structured starting point and adjust deliberately.

• Build on bare costs, add your own labor burden by trade (payroll taxes, unemployment, workers' comp, liability — these sit in the RSMeans O&P column, not in bare labor), then apply your own overhead and profit once, at the estimate level
• Get live quotes on the line items that carry most of the cost — sort your own estimate by extended value to find out which lines those are rather than assuming a rule of thumb, and give each of them a phone call
• Where you have no labor-hour history of your own, keep RSMeans daily output and labor-hours as the productivity skeleton, but price them with your own rates applied against the crew's actual makeup from the crew table — labor-hours are total crew hours covering foreman, journeymen, apprentices and operators, so a single composite rate laid across the whole figure misprices every mixed-trade crew
• Use the division-level city cost index that matches the work, not the weighted total, on any single-trade estimate
• Adjust daily output explicitly for site conditions and write the adjustment down, so a reviewer can see what you assumed and challenge it
• Escalate older data with the historical cost index rather than using a prior edition as printed
• Record the data edition, release date and location factor on the face of the estimate, so the basis is auditable later
• Back-check completed jobs against what the data predicted, and keep a running correction factor by trade

Do you actually need RSMeans at all?

If you self-perform a narrow range of work in one market and you track job costs honestly, your own history is better data than RSMeans for that work. It reflects your crews, your supply chain, your foremen and your site conditions. No national average can compete with that, and you should not pay for one to tell you what you already know about your own trade.

RSMeans earns its keep exactly where your history is thin: trades you sub out, building types you have not built before, markets you are moving into, and design stages that are too early for a takeoff. It also earns it where you need a basis somebody else will accept — an owner, a lender, a public agency or an opposing party in a claim is not going to take your internal unit costs on faith, and an independently published database is a basis they can check.

The subscription and the time it takes to learn the reference section are real costs. For a contractor bidding a handful of jobs a year in familiar work, that combination can be hard to justify, and the honest answer is often that you do not need it.

Where outsourced estimating fits into this is narrower than most providers admit. A quantity takeoff produces the quantities, and the quantities are the part that does not care which cost database gets applied afterward. Once they exist, they can be priced against RSMeans, against your own historical unit costs, or against both so the two can be compared. The question worth asking any estimating provider — in-house or outsourced — is what the pricing basis is and what date it reflects. An estimate that does not state its basis cannot be checked by anyone, including the person who wrote it.