Two competent teams estimating the same scope will land twenty percent apart more often than not. The gap is almost never one thing. It is a dozen small differences in basis that were never compared, because nobody had a structure to compare them in.
The scene is always the same. The contractor says 240 million, the owner’s estimate says 190, and a meeting is called about “the 50”. Without a method, that meeting is two teams defending totals, and it ends with positions hardened and the gap intact. With a method, the 50 stops being an accusation and becomes an address list: here is where each piece of it lives, and here is what each piece is made of.
Why competent estimates disagree
Neither team is lying, and usually neither is wrong. An estimate is a stack of choices: which drawings, which pricing date, which construction method, which risks, whose margin, what confidence. Each choice is defensible on its own, and two teams making a dozen defensible choices independently will diverge by exactly the kind of margin that starts these meetings. The gap is information about the choices, not about the competence. Treating it that way is the whole trick.
Line up the bases first
Before a single dollar is compared, six alignment questions get answered in writing, side by side.
- Pricing date and escalation treatment: same date, and escalation to the same point in time, shown the same way?
- Scope boundaries and exclusions, line by line: the fastest source of phantom gaps is one team pricing something the other listed as excluded.
- Quantities: same drawings, same revision? A take-off from revision D against a take-off from revision F is not a disagreement; it is two different projects.
- Method: same access, sequence and durations? A contractor’s plan and an owner’s assumption routinely differ, and the difference is money, not error.
- Indirects and margin: included where, and at what level? Owner estimates often carry no margin and thin indirects; bids carry both, distributed differently.
- Contingency: included, and at what confidence? A P80 owner number against a bid with risk in the rates is an apples-to-orchards comparison.
Until the bases align, comparing totals is noise. After they align, half the gap has usually already explained itself.
Then map to a common WBS
Both estimates are re-cut into one structure, usually the owner’s, so that every dollar on each side has an address in the same system. The variance report then shows the gap by element and by cause: scope, quantity, rate, method, indirects, risk. Most of the 50 turns out to be three or four elements, and the elements come with names, which means they come with conversations. Why a common structure makes this possible at all is its own subject: see A WBS that survives construction.
Walking the fifty
Here is an illustrative reconciliation of that 240-against-190. Escalation first: the bid is priced to the midpoint of construction, the owner’s estimate to its pricing date; aligning them moves, call it, 8 million. Scope boundary: the contractor priced the temporary trestle and the owner’s estimate listed temporary marine works as excluded; that is 12 million that was never a disagreement, just two documents that had not met. Quantities: the bid took rebar off revision F, the owner’s take-off is from revision D, and the design grew between them: 6 million. Method: the contractor assumed two crane spreads and a longer tidal programme where the owner’s estimate assumed one spread on a causeway: 15 million, and it is a genuine difference of construction plan, the kind worth an actual engineering conversation. Indirects and margin: the bid carries a site organisation and a margin the owner’s estimate never pretended to include: 5 million. Contingency treatment: 4 million. Total located: 50 of 50.
Notice what just happened. Not one dollar of the gap was anybody’s mistake. Three items were bookkeeping alignments, one was a design-growth fact, and two were real commercial and engineering differences that the owner and contractor can now discuss specifically: is the trestle necessary, and is two spreads the right plan? That meeting has an agenda instead of an argument.
The gap is a list. Treat it like one.
Where this goes wrong
Arguing the total. The meeting compares 240 with 190 as numbers, each side explains why theirs is right, and the session produces heat. Totals cannot be negotiated because totals do not contain reasons. Only the elements do.
Splitting the difference. Somebody proposes 215 and everyone goes home. The new number has no basis at all: it inherits neither estimate’s quantities, neither’s method, and both sets of errors. Six months later nobody can defend it, because nobody ever could.
Assuming bad faith. The gap is read as padding on one side or naivety on the other, and the reconciliation becomes an audit. In practice the gap decomposes into legitimate choices almost every time, and starting from suspicion just delays the decomposition.
The spreadsheet merge. The two estimates are lined up element by element without aligning the bases first. Every element difference is then contaminated: the “concrete variance” is really the escalation difference plus the revision difference wearing a concrete costume, and the report misleads with precision.
Reconciling after the decision. The award is made, the budget is set, and the reconciliation happens later as an autopsy. Run before the decision, the same work changes the decision: scope gets clarified, the method question gets answered, and the number that gets approved has been through the wash.
What comes out
Sometimes the owner’s estimate was low, and the reconciliation is the document that lets the project ask for the right budget with reasons attached. Sometimes the bid carried a method the owner did not intend to pay for, and the conversation redesigns the approach or the risk allocation. Often it is both, in parts. Either way the negotiation is now about identified items with a basis, which is the only kind of negotiation that ends. The same discipline applies inside a project across time: a Class 3 reconciled to the Class 2 that follows it, element by element, is how an estimate keeps its memory, and it is one of the eleven questions any independent review will ask: see What a peer review actually checks.
A reconciliation typically takes days, not weeks, because it is structured comparison rather than re-estimating. Against a fifty-million-dollar question, it is the cheapest analysis either party will buy that year.
Emerald Group reconciles estimates that disagree, for owners facing a bid above budget and for teams whose estimates must survive a gate: bases aligned, both numbers re-cut to one structure, and the gap reported by element and cause. If two numbers in your project are fifty apart and the meeting is booked, get in touch first.