The 5 Types of Construction Estimates

The five estimate types at a glance

Each type is defined by its inputs, not by how much effort goes into it. More hours on a schematic set do not produce a bid estimate.

What is an order-of-magnitude estimate?

An order-of-magnitude estimate prices a project that has not been drawn. The inputs are the program — gross area, unit count, story count, structure type, quality tier — plus location and a target start date. The method is parametric: apply a historical cost per square foot, per apartment, per bed or per parking stall, normalized for what the comp actually included — sitework, utility extensions, FF&E and soft costs are commonly outside a $/SF figure — then adjust for location index, market conditions and escalation to the assumed midpoint of construction.

Its value is speed and its honest output is a range with a written basis. That basis should name the analog projects used, the assumed quality tier, and what was assumed about the site. A ROM delivered as a single number, with no basis and no range, is a ROM being misused.

Two things quietly wreck ROM numbers. The first is a bad analog: pricing a surgical suite off medical office comps is not estimating. The second is that cost per square foot hides what actually moves the number at this stage: net-to-gross efficiency determines how much gross area the program requires in the first place, while height and the structure it forces, below-grade and site conditions, floor-to-floor, skin-to-floor ratio and quality tier drive the rate itself. A tall, narrow, articulated building and a simple box of the same square footage are not the same project, and a per-square-foot rate cannot see the difference.

What is a conceptual or schematic estimate?

At schematic design there are drawings: a site plan, floor plates, elevations, and usually a narrative describing structural and MEP intent. Systems are named but not sized. The estimate follows that logic and prices assemblies rather than components — a structural frame as a system, an exterior wall by facade area with an assumed makeup, HVAC by area or by installed tonnage. UniFormat is the usual organizing structure because it maps to the decisions the design team is actually making.

This is the estimate that earns its keep by supporting comparisons. Slab-on-grade versus a structured podium, a punched-window facade versus a curtain wall, VAV versus VRF: those are decisions that are still cheap to change at this stage, and they set cost that later value engineering can only work around.

Its limitation is real and worth saying out loud to whoever receives it. Quantities are derived from areas and ratios, not measured from components. Below-grade conditions, utility capacity and existing-conditions surprises are assumptions, not findings. A schematic estimate should carry a design contingency that is large and clearly labeled, and that contingency should shrink measurably at the next milestone. If it does not shrink, the design is not developing — or nobody is rereading the estimate.

What is a design development estimate?

Design development documents give an estimator sized systems, equipment schedules, and specification sections in outline form. That changes the method. Major architectural and structural scope gets a real quantity takeoff — concrete, structural steel, envelope, roofing, interior partitions. MEP is priced off the equipment schedules and single-line drawings by connected load, tonnage and fixture count, normally with budget pricing from mechanical and electrical subcontractors, since MEP pricing carries real uncertainty at DD and tends to move between DD and CDs. Unselected finishes, specialties and low-voltage stay as allowances. The estimate typically migrates from UniFormat assemblies to CSI MasterFormat divisions here, because the output starts being something you can send to subcontractors.

This is the last milestone where a budget problem is cheap to solve. After this, fixing an overrun means redesign, which means resubmittal and schedule.

Two practices pay for themselves at DD. First, reconcile against a second, independently prepared estimate, line by line. Differences sort into quantity, scope assumption, unit pricing and productivity, or indirects and markup — and each category gets resolved on its own terms. You want all of them surfaced now rather than at bid. Second, hunt the scope gaps — blocking, firestopping, patching, roof curbs, equipment pads, temporary protection, final cleaning. These are the items that sit between trades, that every division assumes another division carries, and that quietly reappear as change orders.

What is a definitive or bid-ready estimate?

A definitive estimate is built from a complete, coordinated construction document set — issued-for-construction or bid-issue drawings, full specifications, and all addenda. A permit set is not the same thing: it is filed for code review and commonly lacks final coordination, equipment selections and connection details, so pricing one as bid-ready means carrying scope you never measured. Everything measurable is measured by trade. Unit prices are built from crew composition, production rate, and fully burdened labor rates — base wage plus payroll taxes, insurance and fringes, not the base wage alone — with crew equipment, small tools and consumables carried in the crew cost rather than pulled whole from a book. Bought scope is priced from subcontractor and vendor quotes that have been scope-checked against the specification, not just compared on price. General conditions and general requirements are built from the schedule and the site logistics plan rather than applied as a percentage: time-dependent items (supervision, trailers, temporary power and heat, hoisting, clean-up labor) priced per week of duration, and one-time or quantity-driven items (mobilization, permits and fees, temporary utility connections, fencing, surveys, final clean, closeout) priced on their own basis. Escalation is carried to the actual buyout and installation dates. On top of the cost of work sit sales and use tax on materials and equipment, insurances (GL, builder's risk, subguard where used), permits and fees if not owner-carried, fee or markup, and then bond, which is rated on the resulting contract price.

The deliverable is not only a number. The qualification list — inclusions, exclusions, clarifications, allowances, alternates and unit prices — is the part that determines whether you get paid for what you did not carry.

The honest limitation: a bid-ready estimate is precise about the documents, not about the job. If the drawings conflict with the specifications, or a detail is missing, you get a precise price for the wrong scope. Whether that ambiguity becomes a change order or becomes your loss is decided by the contract, not by the estimate. Bid-ready estimates also age. Quotes expire, and volatile commodities and equipment lead times move; a number carried for months without a recheck is no longer bid-ready.

What is a control estimate, and why does it often get skipped?

A control estimate is the accepted price rebuilt as a management tool. You take the winning estimate, reorganize it into your own cost codes, and reconcile it to the cost baseline — contract sum less fee and any contingency or allowance you do not control — so that what you forecast against is cost, not revenue. From there you fold in buyout savings and overruns as they land, and phase the result against the schedule so you have a time-based budget rather than a flat total.

It exists because a bid estimate and a project budget serve different purposes. The bid estimate was structured to win the work and to be readable by the owner. The control estimate has to be structured so field costs coming out of payroll and accounts payable land against something comparable, which is almost never the same breakdown.

Without one, three things break. You cannot produce a defensible cost-to-complete, so you find out whether the job made money only when it is over. Change orders can still be priced off the bid detail and any contract unit prices, but without a control estimate you cannot show what the change does to the cost of the remaining work — so impact, inefficiency and schedule-driven costs become opinion against opinion. And there is no baseline for earned value, so schedule progress and cost progress get discussed separately, which is how a job that is behind and over looks fine on both reports.

This is also the one type that is a living document. It gets updated at buyout and at every forecast cycle, and it belongs to the project team.

How do you know which estimate a drawing set can support?

Work from what you actually have. The estimate type is a consequence of the documents, and no amount of estimating effort promotes a set to the next class.

What goes wrong when you ask for the wrong type?

The harm usually comes from mismatch between the format of the estimate and the quality of its inputs. A detailed-looking document implies a level of investigation that may not exist.

Bid-ready format on schematic information. When the documents will not support a takeoff, an estimator has two options: assume, or pad. Both produce a number, and the line-item format makes either one look rigorous. That document then goes to a lender, an owner or a board, and it is read as the price. If you ask for detail the drawings cannot support, ask instead for the assumptions list that generated it — and expect it to be long.

An order-of-magnitude number treated as a commitment. The first number anyone hears becomes the budget of record, regardless of what it was labeled. Every later estimate is then read as an overrun rather than as the first honest look. The countermeasure is procedural: publish a range, publish the basis, and re-issue the number at each milestone so the budget of record moves with the design.

Skipping the design development estimate. Milestone estimates cost money and take time, so the middle one gets cut. The result is that the first rigorous check on the design happens at bid, when the only levers left are redesign, descoping, or accepting the number. Redesign at that point can trigger permit revision or resubmittal and schedule impact, which often costs more than the milestone estimate would have.

Never producing a control estimate. Teams assume the bid estimate is the budget. It is not, because it is organized to explain a price rather than to track cost. Converting it after award is a bounded, one-time exercise that determines whether your monthly cost reports mean anything.

A number with no basis attached. Any estimate delivered as a single figure, without assumptions, exclusions, contingency breakdown and a pricing date, will be remembered as a promise. This is true at every level from ROM to bid. The basis of estimate is not paperwork; it is the part that defines what you agreed to.

Contingency, allowance and escalation are not interchangeable

A recurring source of estimate disputes is one line called "contingency" that is quietly doing several unrelated jobs, so nobody can tell what has been consumed or by whom.

Design contingency. Covers the drawings that have not been drawn yet — the details, connections and coordination that will appear as the design develops. It belongs in early estimates, it should be largest at concept, and it should be drawn down deliberately at each milestone.

Estimating contingency. Covers uncertainty in your own pricing and productivity assumptions on scope that is already defined. It is a different risk from incomplete design and should be carried on its own line.

Construction or owner contingency. Covers changes after award. Whether it sits inside the contract sum or outside it, and who controls it, is set by your contract — read the specific agreement rather than assuming. Treating it as available to absorb estimating misses is how budgets disappear before the slab is poured.

Allowance. A priced placeholder for scope that is defined but not selected — a flooring package, a lighting package, a door hardware set. It sits inside the contract sum and reconciles against actual cost. An allowance is not contingency, and a list of vague allowances is a sign that the estimate outran the design.

Escalation. A function of time, not of design completeness: the gap between the pricing date and when material is bought and installed. It is calculated per scope, because commodities, equipment and labor do not move together. Escalation is not a substitute for contingency, and a schedule delay usually increases it rather than releasing it.

Which of these should you outsource?

Two of the five usually should not leave your office. If you build the same project type repeatedly, your own closed job costs are usually a more relevant basis for an order-of-magnitude estimate than a generic database, and the ROM is quick to produce. The control estimate should also stay in-house, because it lives in your cost codes and accounting system and has to be maintained by the team running the job. An outside firm cannot forecast a job it is not on.

The middle three are where outsourced takeoff and estimating services are genuinely useful: quantity takeoff at DD and bid stage, overflow bid volume, trades outside your normal scope, and an independent second estimate for a reconciliation. Measuring is labor-intensive and eminently delegable work.

Be clear about what that does not buy you. An outside estimator does not have your labor rates, your crew productivity, your equipment costs or your subcontractor relationships — those are what make an estimate yours rather than generic. Any estimator who prices your labor without asking whose rates and productivity to use is giving you a database output with your logo on it. Ask how they handle that question before you ask what it costs.

The other limit is upstream. Outsourcing cannot promote an incomplete drawing set to a higher estimate class. If the documents support a schematic estimate, that is what comes back, and any firm that promises bid-level detail from concept drawings is telling you something useful about how they work.