How to set a carry number you can defend

A carry number is the amount you put in the estimate for a trade package. It is not necessarily the lowest submitted bid, and it is not always what you expect the award to be. It is the number you are prepared to defend in a review, build to in the field, and explain at buyout when the low subcontractor turns out not to be available.

Most estimating processes are careful about the leveling sheet and casual about the step that follows it. The comparison gets built properly, adjustments get priced, and then somebody picks a number. That last move is where margin is quietly decided, and it deserves the same discipline as everything upstream of it.

Three numbers, kept apart

The most useful thing a leveling sheet can do is refuse to collapse three different values into one.

Never overwrite the submitted value. Three weeks later, an estimator's provisional add-back reads exactly like a subcontractor commitment unless the sheet keeps them visibly separate, and somebody will act on it. Preserve the bidder's original language and the original number alongside your normalized view.

Getting to evaluated cost

Evaluated cost is a comparison device. It is not a rewrite of the bidder's proposal and it does not change what they offered.

Evaluated cost = submitted base bid + required scope additions − confirmed credits + allowance normalization + quantity-basis adjustment + schedule and logistics adjustment + commercial adjustment + approved risk adjustment

Each of those adjustments needs six things attached. A leveling sheet without them is just a second estimate with no backup.

Confidence deserves its own column rather than a footnote. An amount confirmed by the bidder in writing and an amount an estimator guessed from a comparable project are both dollars in the same cell, and only one of them will survive contact with buyout.

Unit rates and lump sums are not directly comparable

A unit-rate bid distributes quantity risk differently from a lump sum. To compare them honestly:

1. Establish a common, source-backed quantity.
2. Extend the unit-rate proposal on that quantity.
3. Add mobilization, minimum charges, premium time, waste, and testing that the unit rate does not cover.
4. Confirm whether the rate includes labor, material, equipment, taxes, freight, supervision, overhead, and profit.
5. Record who owns final quantity variance.
6. Keep the original unit rate visible for the commercial recommendation.

The failure to avoid is converting a unit-rate bid into a "lump sum" in the leveling sheet and then forgetting the quantity exposure that conversion assumed away. The extension is for comparison. The bidder's commercial offer is still a rate.

Allowances hide more than they normalize

Normalizing every bidder to a common allowance amount makes the columns comparable and tells you almost nothing about the risk. The review still has to establish what the allowance covers, what it does not, whether labor and equipment and taxes and freight and markup are inside it, how unused allowance is treated, what happens when actual cost exceeds it, and whether it is owner-controlled, contractor-controlled, or trade-controlled.

Two bidders carrying the same allowance value under different definitions are not level. They just look level.

Keep alternates out of the base

An alternate should never be moved into or out of the base bid to make a bidder look lower. Confirm the bidder's intended structure and align to the owner's instructions, in that order.

From evaluated cost to the carry

Carry number = selected executable bid + confirmed scope adjustments + unresolved expected exposure + approved market or schedule reserve − verified credits

The load-bearing phrase is selected executable bid. The bidder with the lowest evaluated price may still be the wrong basis for the carry because of capacity, schedule, safety record, bonding, prior performance, commercial qualifications, or an inability to confirm required scope. Evaluated cost ranks the proposals. It does not select one.

A worked example

Bidder A submitted less than Bidder B and evaluates higher. Bidder C is the apparent low bidder and the most expensive outcome. That inversion is the entire reason the exercise exists: price is evaluated after scope and risk, not before.

Pricing the risk you decide to carry

For discrete, identifiable exposures, an expected-value calculation makes the treatment consistent across estimators instead of a matter of temperament.

Expected risk cost = probability of occurrence × cost impact

Expected value is the right tool for a portfolio of repeated decisions. It is a weak tool for a single catastrophic one. A 5% chance of a $4 million exposure prices out at $200,000, and no estimator should sleep on that number. Low-probability, high-severity events usually need a larger reserve, contractual transfer, an alternate, or an escalation to management, not an expected value quietly folded into a trade line.

AACE's risk guidance supports linking risk drivers to cost and schedule outcomes and recognizes expected value as one quantitative method for contingency analysis. It also warns specifically against double-counting when methods are combined, which is the failure below.

The duplicate-contingency trap

The most common way a bid becomes uncompetitive for no reason is carrying the same uncertainty four times:

• The subcontractor prices a risk premium into their number.
• The estimator adds a leveling adjustment for the same condition.
• The project estimate carries a trade contingency.
• Management adds general contingency on top.

Each step is individually defensible. Together they price one risk at four times its value, and nobody can see it because the four amounts live in four different documents. The fix is unglamorous: the issue log records where each risk is carried, and a risk that appears twice gets removed from one of them.

The mirror-image failure is just as common. An undocumented round number in a trade line is not a risk treatment, it is a plug that no one can review, adjust, or release at buyout. If you cannot say what the money is for, you cannot defend keeping it or justify spending it.

What the leveling sheet has to record

A workbook that supports a defensible carry keeps source requirements, bidder responses, pricing, clarification status, and carry treatment in separate fields.

Show at least four figures per bidder: submitted bid, confirmed revised bid, evaluated cost, and remaining unpriced exposure. Collapsing them into one "leveled total" makes the analysis impossible to audit, which means it is impossible to defend six weeks later when someone asks why the number moved.

The recommendation should fit on one page

Not this:

Award Bidder B because it is lowest after leveling.

This:

Carry Bidder B at $2,028,000, subject to written confirmation of perimeter fire-containment responsibility. Bidder B is $37,000 below the next evaluated proposal after normalizing roof-curb scope, winter conditions, and the manufacturer inspection allowance. Capacity and schedule are confirmed, with no unresolved long-lead exceptions. Carry includes $8,000 of expected risk for the open firestopping interface, which is not duplicated in the trade contingency. Final subcontract scope must incorporate Clarifications 03, 05, and 07.

The second version can be checked by someone who was not in the room. That is the whole standard.

Where Piper fits

The carry decision is judgment, and it should stay that way. What makes it hard is not the judgment, it is reconstructing enough evidence, under time pressure, to make the judgment defensible.

Piper keeps submitted values, adjustments, sources, and confidence separate all the way through, so the carry and its basis never collapse into a single number nobody can explain later. The exclusions and silent items behind an adjustment come from the same reading of the drawings, specifications, and company standards that produced the scope, which is why the backup is already assembled when the estimator arrives at the decision. The decision stays theirs. For the full pre-submission process this sits inside, see the final bid review playbook.