What a Fixed-Price Machine Proposal Should Contain — and the Five Things That Cause Overruns

A fixed-price proposal is a promise about the future made with incomplete information. That's not a flaw — it's the whole point. You're paying a builder to absorb the uncertainty so you don't have to. But the only way that promise holds is if both sides agree, in writing, on what's being promised. Most proposals don't, and that's where overruns come from.

We know because we've written proposals that didn't, and paid for it. What follows is what we now put in every one, and what we'd tell you to demand from anyone.

What "done" means, in numbers

The single most important section, and the one most often missing. A proposal that says the machine will "meet production requirements" has promised nothing. A proposal that says the machine will run 35 jars per minute sustained for 30 continuous minutes, rate logged from the PLC counter, with removal torque within ±10% of setpoint across a 30-jar sample has promised something you can test.

Insist on: the rate and how it's measured; the duration of the acceptance run; the quality metric and its tolerance; the product(s) it applies to; and the conditions — at the builder's shop (FAT) and at your site (SAT). If a builder resists writing this down, they either don't know whether they can hit it or don't want to be held to it. Both are answers.

What's excluded, explicitly

Every machine touches things around it: utilities, upstream and downstream equipment, the building, your product, your operators. A good proposal names what the builder is not responsible for — power to the disconnect, compressed air at a stated pressure and flow, floor flatness, product supplied to spec, the performance of equipment they didn't build. Exclusions aren't the builder dodging; they're the map of where the risk actually sits. A proposal with no exclusions has hidden them inside the price, and they'll surface as change orders.

The design basis — what the machine is sized for

Rate, product dimensions and variants, changeover expectations, run hours per day, sanitation regime, ambient conditions. This is the fine print that prevents the most common argument: "It doesn't run this product." If the design basis says three SKUs and you have five, you've found a scope conversation before it became a dispute. We once quoted a client two machines that were wrong — one too slow, one too big — before the third was sized to their measured data instead of their ambition. The design basis is where that gets fixed.

Milestones tied to evidence, not dates

Payments should follow proof: design review complete, long-lead materials ordered, FAT passed, SAT passed. A proposal that bills 50% up front and 50% "on delivery" has removed your leverage at the exact moment you need it. Holdback tied to site acceptance — 10–15% — is normal and healthy. It keeps the builder in the building until it works.

What triggers a change order, and how it's priced

Scope moves. It always moves. The question is whether the mechanism for handling it is agreed in advance or invented in an argument. Look for: what counts as a change (new product, new rate, new interface, client-caused delay); how it's communicated (in writing, before the work); and how it's priced (rate card or negotiated per change). The proposals that go badly are the ones where the builder absorbed $5K of "small stuff" out of goodwill, then $15K, then sent a surprise invoice for $40K. Written thresholds prevent that. Ours is $5,000 — nothing unbilled above it without a written notice first.

The five things that actually cause overruns

From our own closeout reports and from the industry, in rough order of damage:

• No numeric acceptance criteria. "Done" becomes a negotiation. The builder keeps tuning, the client keeps finding issues, hours pile up on both sides, and nobody can say when to stop. This is the root cause of most of the rest.
• Design basis built on assumptions instead of measurements. The client said 40 per minute; the line actually runs 25. Or the product samples were from the good batch. Or the "standard" jar has four variants. Every one of these is a machine that's wrong on arrival.
• Scope creep absorbed silently. The builder wants to be helpful. The client doesn't realize the small asks are adding up. The relationship is great right up until the reconciliation.
• Interfaces nobody owned. The machine is fine. The conveyor feeding it isn't. The compressed air is at 70 psi, not 90. The PLC needs to talk to an MES nobody documented. Every gap between the builder's scope and yours is a place the project can stall, and it stalls at the end, when it's most expensive.
• Underestimated hours on novel work. If nobody's built it before, the estimate is a guess. Honest builders say so and structure the work in phases — prove the hard part first, then quote the rest with real information. Builders who fixed-price a first-of-kind machine without a proving phase are either very experienced with that exact mechanism or about to lose money on you.

A quick test for any proposal

Read it and try to answer these without calling the builder: How will I know it's done? What am I responsible for supplying? What happens if I need a change? What do I pay, and when, for what proof? What's it sized for? If any of those takes more than a minute to find, the proposal is incomplete — and the missing answer is where the overrun lives.