Ask a steel erector how productive their crane was last week and you'll get an answer. Ask them how that compares to similar projects, and the conversation stops. Construction has never had reliable productivity benchmarks because the data to create them didn't exist. Every project tracked production differently, manually, and with enough inconsistency that comparing across projects was meaningless.
That's changing. With over 12 million crane picks captured across hundreds of projects, patterns are emerging that give erectors, GCs, and owners something they've never had: an objective baseline for what "good" looks like on a steel erection project.
Other industries have benchmarks for everything. Manufacturing measures units per hour. Logistics measures delivery time per route. Even agriculture has yield-per-acre data going back decades. Construction has been the outlier, and the reason is simple: the work happens in the field, every project is different, and nobody had a consistent way to measure production across sites.
Manual tracking made benchmarking impossible for three reasons.
Inconsistent definitions. What counts as a "pick"? Does shake-out count? Does repositioning a piece count? When one project counts shake-out and unloading in their erection numbers and another doesn't, their picks-per-hour figures aren't comparable. Estimating teams rely on picks-per-hour as a core metric, and when the definition varies, the metric becomes unreliable.
Subjective counting. When the foreman reports 23 pieces set today, that's an estimate based on memory at the end of the shift. The actual number might be 21 or 26. Over a week, the margin of error compounds. Over a project, it's large enough that any benchmark derived from it would be meaningless.
No cross-project data. Even if one project tracked perfectly, comparing it to another required someone to normalize the data manually: adjusting for crane type, building height, piece weight, crew size, and dozens of other variables. Nobody had time for that analysis, so nobody did it.
When production data is captured automatically from the crane hook, measured consistently across every project, the benchmarks start to tell a story. Here's what Versatile data reveals across hundreds of steel erection projects.
Pick cycle times vary more than most teams realize. The average pick cycle time across projects falls in a range, but the spread within that range is significant. Top-performing crews consistently run shorter cycle times not because they move faster, but because they have less dead time between picks. The gap between an average crew and a high-performing crew is primarily coordination efficiency, not speed.
Crane utilization rarely matches assumptions. Most teams estimate their crane utilization at 70 to 80 percent. The data consistently shows actual utilization running 15 to 25 points lower than what teams assume. That gap isn't laziness or incompetence. It's the accumulation of 30 to 60 minutes of micro-delays per crane, per day: coordination gaps, material staging issues, trade transitions, and weather holds that add up invisibly.
Morning production outpaces afternoon production. Across the dataset, the first four hours of a shift consistently produce more picks than the last four. The dropoff isn't dramatic on any single day, but it's consistent enough to be a real pattern. Contributing factors include fatigue, afternoon deliveries that disrupt staging, and coordination meetings that break momentum.
Weather impacts are larger than most teams budget for. Wind holds, lightning delays, and temperature-related slowdowns account for more lost production than most project schedules accommodate. The data shows the actual impact, which gives erectors and GCs a realistic number for weather contingency instead of a guess.
"Having benchmarks from other projects gave us a realistic target instead of just hoping for the best." (VP of Operations, regional steel erector)
Benchmarks are only valuable if they change decisions. Here's where steel erectors are applying production benchmarks to protect margins and improve performance.
Estimating and bidding. When the estimating team builds a bid, they need a production rate to calculate labor cost. Historically, that rate came from the last project's foreman, adjusted by gut feel for the new project's complexity. With benchmark data, the estimating team can see how similar projects actually performed: similar building type, similar piece weight, similar crane configuration. The bid becomes more accurate because the production assumption is grounded in data, not memory.
Crew performance conversations. Telling a foreman "you need to be more productive" is vague. Telling a foreman "your pick cycle time is averaging 14 minutes when similar projects run 11" is specific and actionable. The benchmark gives the foreman a target that's based on what other crews have actually achieved, not what the PM wishes they would achieve. The conversation shifts from pressure to problem-solving: what's different about this project that's adding three minutes per pick?
Project planning and scheduling. When the PM builds the erection schedule, the production rate assumption drives everything: how many days per zone, how many shifts per week, when the crane needs to move. If that assumption is off by 15 percent, the schedule is off by 15 percent. Benchmarks from comparable projects give the PM a realistic starting point instead of an optimistic one.
Portfolio performance tracking. For erectors running multiple projects simultaneously, benchmarks enable a portfolio view. Which projects are outperforming the baseline? Which ones are underperforming? Where should management attention go? Without a benchmark, every project exists in isolation. With one, the portfolio tells a story.
Benchmarks are built on data, and data is only useful if the field trusts it. This is a real challenge. When a number looks wrong, even once, the team's confidence in everything the system produces drops. One erector's field team flagged that the piece count in the Control Center seemed low. The system was counting loads, not unique piecemarks, so when a fabricator bolted several pieces into one lift, the count understated the crew's actual output. The crew felt the data was working against them instead of for them.
This is why data accuracy matters more than data volume when building benchmarks. A benchmark built on verified, model-validated picks is trustworthy. A benchmark built on manual counts with inconsistent definitions is noise. The crane intelligence system validates every pick against the IFC model, which means the data feeding the benchmarks has been checked against the design, not estimated from the ground.
Trust also comes from transparency. When the foreman can see the same data the PM sees, and when that data matches what the foreman observed on the pad, the system earns credibility one shift at a time. Benchmarks that the field doesn't trust will never drive improvement, no matter how statistically valid they are.
Steel erection is a margin business. The erectors who consistently protect 2 to 6 margin points more than their competitors aren't necessarily running faster crews or using better equipment. They're making better decisions because they have better data.
Construction productivity benchmarks are part of that data advantage. They turn estimating from guesswork into analysis. They turn crew performance conversations from subjective to specific. They turn schedule assumptions from optimistic to realistic. And they give erectors something to show owners and GCs that most competitors can't: objective evidence that they measure, manage, and improve their production.
The industry has operated without productivity benchmarks for its entire history. Not because the benchmarks weren't needed, but because the data to build them didn't exist. With crane intelligence capturing every pick on every project, that data now exists. The erectors who use it to benchmark, plan, and improve will have an advantage that compounds with every project they complete.