The Cheapest Modular Building Quote Is Almost Never the Cheapest Project

My position: stop comparing quotes, start comparing totals

I run procurement for a 340-person general contractor. I've owned our temporary facilities and site services budget — roughly $610,000 a year across modular offices, storage containers, and steel structures — for seven years. I've negotiated with 40-plus vendors and logged every invoice in our cost tracking system since 2018.

Here's my position, stated plainly: the lowest quoted price on a modular building is a weak signal, and in about a third of the deals I've personally reviewed, it pointed to the most expensive outcome.

Not because vendors are dishonest. Because the quote covers maybe 60% of what you actually pay. The rest — hauling, blocking and leveling, utility tie-ins, permit-driven modifications, demobilization, damage thresholds in the lease — shows up later, on someone else's cost code, usually yours.

I'm going to argue this with three cases from my own spreadsheets, then deal with the obvious objection: that this is just a big-company luxury and small buyers should chase price. I don't buy that, and I'll explain why.

Argument 1: On leased modular offices, the invoice is the smallest number

In 2023 I ran a comparison for a six-month project office package — eight 12x60 units, delivered and set. Three quotes came in at $41,200, $38,900, and $44,600 for the term.

We went with the $44,600 quote. Our project controls lead thought I'd lost it.

Then I rebuilt the comparison as total cost of occupancy:

• Transport and re-haul. Two of the quotes priced inbound delivery only. Outbound re-haul on eight units, at the mileage rate in their own tariff: $6,800 and $7,400. The third quote bundled both directions.
• Blocking and leveling. One vendor required a licensed site contractor to set blocking to their spec. We got that scoped at $3,900.
• Utility connections. Power drop, water, and sewer tie-in ran $5,200 regardless of vendor — but only two of the three quotes told us it was excluded.
• Damage and cleaning thresholds. One lease defined "normal wear" so narrowly that repainting was billable at $1,150 per unit.

Add it up and the $38,900 quote lands near $58,000 in real occupancy cost. The $44,600 quote landed at $49,300. That's a 15% swing hidden entirely in scope language, not in rate.

Why does this happen? Because modular space is a logistics product wearing a real estate costume. Every unit moves twice, gets set on ground that isn't yours, and connects to utilities you don't control. The vendor who quotes all-in looks expensive until you price the pieces they left out.

Argument 2: Imported folding container houses — the FOB price is the least interesting number

This is where buyers get hurt worst, and it's the counterintuitive one.

We evaluated folding container houses from Chinese suppliers in Q1 2024 for a workforce housing pilot. The FOB quotes were startling — in a few cases less than half what a domestic modular unit costs per square foot.

Then I built the landed cost model. It was not half.

What goes on top of FOB: ocean freight and the current bunker adjustment, port drayage, customs brokerage, Section 301 duties on the applicable HTS classification, and — the one everybody forgets — inland trucking from the port to a site that is usually not near a port. On our pilot, landed cost ran roughly 1.6 to 1.9 times FOB before a single unit was set on blocks.

Then the part you can't put in a spreadsheet cell: conformance risk. A folding container structure sold into the US market needs to satisfy the building code adopted by the local jurisdiction, and if it's used as occupiable space, that generally means demonstrating structural performance against ASCE 7 load criteria for the site — ground snow, roof live load, wind — plus egress, and often state-level factory-built or modular building program approval.

Which brings me to the spec language that shows up in these listings: 1.5 kPa roof load, 50mm wall thickness.

1.5 kPa is about 31 psf. ASCE 7 sets minimum roof live load for ordinary flat, pitched, and curved roofs at 0.96 kPa (20 psf), so 1.5 kPa is a real margin above the floor — but it is a minimum-class number, not a snow load number. If the site's ground snow load is 40 psf, the roof requirement is driven by ASCE 7 Chapter 7, not by the catalog. And 50mm walls tell you the panel thickness, not the thermal performance, the cladding gauge, or the fire rating — all of which are separate questions with separate code answers.

I'm not saying imported prefab is a bad buy. I'm saying the catalog spec sheet is a starting point for engineering review, not a substitute for it. Budget the review, or budget the retrofit.

This was accurate as of Q1 2025 on the duty and freight side. Tariff schedules and ocean rates move fast — verify current numbers before you build a budget on them.

Argument 3: Lease-versus-purchase is a duration question, not a rate question

Everyone asks which is cheaper. Wrong question. The question is: how long will you actually need it?

Our rule of thumb, built from 40-plus transactions: under about 18 months of need, leasing a modular office almost always wins once you price residual value, storage, redeployment, and disposal honestly. Past roughly 36 months, purchase usually wins — but only if you have a realistic plan for the asset afterward. Between 18 and 36 months, it depends on whether you have another project to redeploy into.

I want to say our crossover point was 22 months in the 2022 analysis, but don't quote me on that — it shifted with rates and it'll shift again.

The failure mode I see most: buying because the monthly lease rate looked high, then paying storage for two years because nobody lined up the next job. Storage is the silent killer. We were paying $340 a month per unit cluster at one point for units sitting in a yard doing nothing.

I knew I should have written the redeployment assumption into the capital request. I thought, what are the odds the next project gets delayed? The odds caught up with me in 2021 — nine months of storage on twelve units, roughly $30,000 that bought us absolutely nothing.

The objection: isn't TCO analysis just for big buyers?

Fair challenge. If you're renting two units for a four-month job, you're not building a 40-line cost model.

But you don't need one. You need four questions on the phone before you sign:

1. Is outbound transport included in this number?
2. Who sets blocking, and to whose spec?
3. What exactly counts as damage I'm paying for?
4. What's the notice period, and what does early termination cost?

Those four questions are the difference between a $38,900 project and a $58,000 project. That's not enterprise procurement. That's ten minutes.

The "TCO is for the big guys" thinking comes from an era when buyers got one quote and signed it because comparing was genuinely hard. Today the comparison is a spreadsheet and four phone calls. The barrier is gone; the habit isn't.

And to be clear about my own bias: I don't work for a modular vendor, I don't get a referral fee, and I have no visibility into anyone's rate card. I'm the guy who has to explain a cost overrun to a CFO.

Where I land

Price is a data point. Total cost of ownership is the decision.

If you're comparing modular office leases, put transport both directions, site prep, utility tie-ins, damage thresholds, and termination terms into one column per vendor. If you're importing folding container structures, build landed cost through inland delivery and add a line for engineering and code review. If you're weighing lease against purchase, model the duration and the redeployment plan before you model the rate.

The cheapest quote I've ever received cost my company about $19,000 more than the second-cheapest. I still have the spreadsheet. It's the reason our procurement policy now requires at least three vendors quoted on identical scope, not identical headline price.

One caveat, and I mean it: these numbers are ours, from our projects, in our markets. Labor rates, freight, and local code enforcement vary enough that you should run your own model. Use mine as a template, not an answer.