A rough order of magnitude number and a guaranteed maximum price aren’t the same instrument, and treating them as interchangeable is how sponsors end up revising their pro forma mid-construction. A ROM vs GMP cost estimate comparison isn’t just a matter of decimal points. It reflects how much design information actually exists at the moment the number was generated. If you’re underwriting a hotel deal based on a number that was never meant to carry that weight, your IRR model is standing on sand.

Why doesn’t my early estimate match the final number?

Early estimates and final estimates diverge because they’re built from different inputs at different points in the design process. A rough order of magnitude figure is generated with minimal drawings, while a guaranteed maximum price is locked against complete construction documents. The gap between them isn’t an error, it’s a function of how much is actually known.

Sponsors who get burned usually aren’t victims of a bad contractor. They’re victims of a misread number. A ROM pulled together from a napkin sketch and a flag brand’s prototype plans gets treated like a budget commitment, then gets locked into a capital stack conversation with lenders and equity partners who expect precision that doesn’t exist yet. The estimate was honest. The expectation around it wasn’t.

This is also where the lowest bid becomes a risk signal rather than a win. A number that looks tight at the ROM stage, with no scope detail behind it, tells you more about what’s missing than what’s controlled.

What’s the difference between a ROM and a GMP cost estimate?

A ROM estimate is a directional number built on conceptual design, typically used for early feasibility and go/no-go decisions, while a GMP is a contractual ceiling price built on completed drawings and specifications. The accuracy range between them is wide by design, not by accident. Understanding where you sit on that ladder tells you how much confidence your pro forma should actually have.

The typical progression looks like this:

  • ROM (Rough Order of Magnitude): Built on program and massing, often before a full architect is engaged. Useful for site selection and initial lender conversations, not for locking a capital stack.
  • Schematic Design estimate: Based on early floor plans and building sections. Scope is clearer but systems and finishes are still assumptions.
  • Design Development estimate: Structural, mechanical, and electrical systems are defined. This is where real trade pricing starts to inform the number.
  • Construction Documents estimate: Drawings are near complete. Quantities can be taken off directly instead of estimated by ratio.
  • GMP: The contractor commits to a ceiling price against finished documents, with defined allowances and contingency.

Each rung on that ladder should tighten the accuracy band. If your contractor is giving you GMP-level confidence language on a ROM-level number, that’s a conversation worth having before you sign anything.

Why do cost estimates get more accurate over time?

Estimates get more accurate as design documents get more complete, because pricing shifts from industry benchmarks and square-footage ratios to actual quantities, actual systems, and actual subcontractor pricing. Early in design, an estimator is pricing a concept. Late in design, an estimator is pricing a building. Accuracy is a function of information, not effort.

This is also why budget overruns start in design long before the first trade shows up on site. A ROM built without structural engagement, without a real MEP concept, without site-specific geotech data, is carrying hidden risk that doesn’t show up as a line item. It shows up later, as a change order.

The firms that protect accuracy early bring cost input into the design process itself rather than waiting for drawings to be declared finished before pricing them. That’s the entire argument behind early GC involvement protecting the capital stack: constructability and pricing feedback happening in parallel with design, not after it.

How do you hold your team accountable for estimate accuracy?

You hold a development team accountable by requiring every estimate to disclose its basis, its contingency assumptions, and what hasn’t been finalized yet. A number without a stated confidence range isn’t a budget, it’s a guess dressed up as one. Sponsors who ask the right questions at each stage catch drift before it compounds.

At minimum, every estimate handed to you should answer:

  • What design documents was this number built from, and what percentage complete were they?
  • What contingency is carried, and does it shrink as design matures or stay flat?
  • What allowances are placeholder numbers, and what’s actually been priced by a subcontractor?
  • What’s changed in scope or market conditions since the last estimate was issued?

If the answer to any of those is vague, you don’t have an estimate problem, you have an accountability problem. A contractor who can’t articulate the basis of a number shouldn’t be trusted with the number itself. This also ties directly into schedule, because every re-estimate cycle costs time, and schedule drift erodes IRR independent of whether the budget itself holds.

Frequently Asked Questions

What accuracy range should I expect from a ROM estimate?

A ROM estimate typically carries a wide accuracy range, often plus or minus 20 to 30 percent, because it’s built on conceptual information rather than completed drawings. It’s useful for feasibility screening and early site decisions, not for locking debt or equity commitments. Treat it as a directional signal, not a budget.

When should a GMP be locked in relative to construction start?

A GMP should be locked once construction documents are substantially complete and major subcontractor pricing has been secured, generally shortly before mobilization. Locking it too early, against incomplete drawings, defeats its purpose as a ceiling price and reintroduces the same risk a ROM carries.

What causes an estimate to move backward in accuracy?

Scope changes, delayed design decisions, and market volatility in materials or labor can all push an estimate’s accuracy backward even late in the process. A GMP built on finished drawings can still drift if the owner introduces changes after it’s locked, which is why change order discipline matters as much as the original estimate.

How does contingency change between estimate stages?

Contingency should shrink as design matures, reflecting the fact that less is unknown at each successive stage. A ROM might carry 15 to 20 percent contingency, while a GMP typically holds closer to 3 to 5 percent, because most scope and pricing uncertainty has already been resolved by that point.

Treat every estimate as a confidence statement, not a fixed number

The distance between a ROM and a GMP isn’t a flaw in the process, it’s the process working correctly. The real risk isn’t that early numbers move. It’s underwriting a deal as if they won’t.

If you want to know where your project actually sits on that accuracy curve before you commit more capital, a pro forma and feasibility review will tell you.