The Honest Guide to Structural Steel Estimating

Book a Demo

Back to blogThe Honest Guide to Structural Steel Estimating<br>JJosh Ford

Jun 10, 2026 · 9 min read

tldr: Most guides to structural steel estimating teach you the steps. The steps were never the problem. Bidding the wrong jobs, slow takeoffs, and revision churn are where fabricators actually lose time and margin, and where the right software earns its keep.

Search "how to estimate structural steel" and you'll get a dozen guides that walk you through the same steps. Read the drawings. Do your takeoff. Price the material, add labor, submit the bid. All technically correct. None of it explains why good shops with experienced estimators still lose money on bids.

Here's the part nobody writes about. In structural steel, estimating isn't where you win work. It's where time and margin quietly disappear. The steps were never the hard part. The volume, the revisions, and the wrong jobs are.

We build estimating software for steel fabricators. Over the last several months we've sat in on dozens of demos and onboarding calls with shops of every size, from two-person fab shops to some of the biggest fabricators in North America. This is what we actually heard, not the textbook version.

What structural steel estimating actually is

At its core, structural steel estimating turns a set of drawings into a number you can bid. Three moving parts:

The takeoff. Counting and measuring everything you have to fabricate and erect: columns, beams, connections, and miscellaneous steel. This is where your tonnage and piece counts come from, calculated off the standard section weights published by bodies like AISC in the US and CISC in Canada.

Pricing. Material plus labor plus everything that rides along: connections, coatings, freight, erection.

The bid. A go or no-go call, and a number you're willing to stand behind.

Most of the industry still runs the takeoff in a tool like Bluebeam, exports to Excel or Tekla, and stitches the rest together by hand. That works. It's also exactly where the trouble starts.

Where steel estimates actually break

If you've estimated steel for any length of time, none of this will surprise you. It's worth naming anyway, because almost every guide skips it.

You're bidding the wrong jobs. Estimators tell us they get dozens of invitations to bid every week. Try to bid them all and you land where most shops do, a 10 to 15% win rate, with a team working past midnight. The most expensive estimate is the one you spend a week on and lose. The second most expensive is the one you win and then can't fabricate on schedule. Good structural steel estimating starts before the takeoff, with an honest go or no-go.

The takeoff is a time sink. Columns and beams on a single job routinely take days. One chief estimator described spending the better part of a week just to pull columns and beams off a single floor. Now multiply that across every ITB on the board. That's the real cost of a manual steel takeoff. Not the difficulty, the hours.

Revisions quietly drain your margin. Steel jobs don't sit still. You bid off a 30% or 60% set, then a fresh design set lands and you're re-estimating from scratch. Addenda pile up. If the same beam gets counted twice across three revisions, your tonnage is wrong and so is your number. One estimator summed it up: "with each addendum, I lost a day." Structural steel cost estimation lives or dies on catching what actually changed between sets.

Columns are the hard part. Beams are relatively easy to take off. Columns, especially with no column schedule, are where it falls apart for people and software alike. Any structural steel estimator will tell you the same.

The honest truth about AI steel estimating software

This is the section the polished guides won't write, because most of them are selling you something. In 2026, AI is only now reaching structural steel, a trade that's been slow to modernize, and even industry publications like Modern Steel Construction are catching up to the shift.

Every tool in this space says it does columns. One estimator described the trial experience perfectly. They all say "we do columns, we do columns, we do columns," and then you run it and it turns out to be "yeah, we do columns… kind of."

So if you're evaluating structural steel estimating software or steel takeoff software, here's how to cut through the demo.

Test columns, not beams. Anyone can detect beams. Columns without a column schedule are the real test, so make that the first thing you check.

Run your own job, live and unseen. Demo projects are rehearsed. Hand them a set they haven't seen and watch it run. If it looks too good to be true on their project but you never tested yours, you don't actually know anything yet.

Push hard on revisions. Can it compare two sets and tell you what changed without double-counting? That's where most tools quietly break.

Watch for tools built for general contractors. A lot of AI...