Price Per Ton Isn't a Takeoff: Why Connection Counts Decide Steel Margins

Two steel jobs with the same tonnage can carry very different labor. Why connection and piece counts, not a per-ton rate, decide whether a steel bid makes money.

Published . Last updated .

By Eva Shivers

Two jobs land on your desk the same week. Both come in around 180 tons. One is a single-story warehouse with long-span joists and simple bolted shear connections. The other is a three-story medical office building with moment frames, a mezzanine, and a lot of field welding.

If you price both at the same rate per ton, you'll probably win the second one. That's the problem.

Per-ton pricing is a useful sanity check. It's a bad way to build a bid. The weight tells you what the mill charges. It doesn't tell you what your shop and your erection crew will actually spend making and standing that steel.

Where the Labor Actually Lives: Connections

A beam is a cut, maybe some holes, and a coat of paint. A connection is layout, drilling, fit-up, welding, and inspection packed into a small piece of steel. That's where the hours go.

SMA Estimating puts numbers on it: connections make up less than 5% of total steel tonnage, but they can make up 30% to 50% of fabrication costs. They also note that moment connections run 2.5 to 4 times the cost of a simple shear connection at the same location.

The same source gives rough shop labor ranges by job type:

Type of work Shop hours per ton

Simple bolted framing 15 to 20

Moderate, with some moment connections 20 to 30

Heavy welded work 30 to 45

Run that against our two 180-ton jobs. The warehouse sits near the bottom of the range. The medical office building could land at double the shop hours or more. Same weight, very different job. A single blended rate per ton quietly underprices the complex one and overprices the simple one, which means you lose the jobs you'd make money on and win the ones you won't.

Why Estimators Fall Back on Per-Ton Pricing

Nobody thinks per-ton is precise. Estimators use it because the alternative takes time they don't have.

A real piece-level takeoff means counting every beam, column, and plate, reading every connection detail, and tying each one to a member mark and section size. On a mid-size set, that's a full day or more. A full manual steel takeoff can run 4 to 40 hours depending on the package. When there are three bids due this week and one senior estimator, something gives, and what usually gives is detail.

So the shop estimates tonnage from the framing plans, applies a rate that worked on the last few jobs, and adds a little contingency. It's fast. It's also how margin disappears on connection-heavy work, and how the errors we covered in the real cost of a manual takeoff creep in.

What a Piece-Level Takeoff Gives You

The answer isn't to ban per-ton pricing. It's to make the detailed count fast enough that you never have to guess.

RenderDraw TakeOff reads structural drawings directly from PDF, CAD, or BIM. On a steel set it counts beams, columns, plates, and connections, measures member lengths, and identifies member marks, section sizes, and schedule references. That's the raw material for both numbers you need: weight from section sizes and lengths, and labor from how many connections of each type the job actually has.

The estimator's job shifts from counting to judging. They review the extracted quantities, check anything flagged, and approve. Shops using this approach take steel takeoffs from hours to under an hour, with a 90%+ AI-extracted match rate. For a closer look at what comes off a structural set, see member counts, weights, and the handoff to CPQ.

Then the approved quantities map to your catalog and go straight into your CRM or CPQ, where your connection and labor pricing already lives. The quote reflects the job you're actually bidding, not the average of the last five.

How to Use It on Your Next Bid

You don't need to change your whole process to start. Try this on the next job that feels connection-heavy:

  1. Price it your usual way, per ton, and write the number down.

  2. Run a piece-level takeoff and break connections out by type: shear, moment, seated, braced.

  3. Apply your shop hours per connection type instead of one blended rate.

  4. Compare the two numbers. If they're far apart, you've found where margin was leaking.

  5. Before you trust any automated count, spot-check it. Our post on auditing AI beam counts before bid day walks through how.

Do that on three or four jobs and you'll know whether your per-ton rate is helping you or hurting you. Keep per-ton as the gut check at the end, not the method at the start.

The Bottom Line

Tonnage tells you what the steel costs. Connections tell you what the job costs. When the detailed count takes a day, estimators fall back on per-ton pricing and hope the average holds. On connection-heavy work, it usually doesn't.

Make the piece-level takeoff fast and the tradeoff goes away. Your estimator reviews real counts instead of guessing, the quote goes into your CRM or CPQ without re-keying, and the same team can take on more bids because each one takes less time. Shops working this way report going from around 10 bids a month to 20 or 30 with the same people.

Send us a recent job you priced per ton, and we'll show you the piece and connection counts TakeOff pulls from the same drawings.

Let's talk -> https://renderdraw.com/contact