The State of
Industrial
Estimation 2026
The benchmark report for specialty mechanical & EPC estimating teams - how U.S. contractors win and lose work before the first weld.
Inside this report
This is a diagnostic instrument, not a think-piece. Every metric in these pages is a line you can measure yourself against. As you read, hold one question: are we above or below the line? Part 6 consolidates every figure into a scorecard you can grade your own desk against in ten minutes - including an interactive Estimating Capacity Index calculator, a single number you can track year over year.
The sample. 312 estimating professionals at U.S. industrial contractors - specialty mechanical, process piping, structural steel, electrical & instrumentation, insulation, and multi-discipline EPC. Fielded January–February 2026 through the RFP Pros community, at firms from under $25M to over $1B in revenue.
Two data layers. Figures marked RFP PROS SURVEY 2026 are original practitioner data. Third-party figures are cited inline (BLS, ABC, AGC/NCCER, Flyvbjerg, CII, FMI). The two layers are never blended in a single number.
Industrial contractors are not short of work. They are short of the capacity to bid it.
Across 312 estimating teams, the constraint on growth is no longer demand, price, or even skilled field labor. It is the estimating desk. The firms that have widened it are pulling away from the firms that have not, and the distance between them is the subject of this report.
of industrial estimating teams turned down qualified work in the last 12 months - for lack of capacity to bid it, not lack of pipeline. The top-quartile firms in this study have largely designed that decision out.
What share of your qualified pipeline did you actually bid last year? Under 45% and you are below the median in this study. Over 68% and you are top quartile. Hold that number as you read. Everything that follows is an explanation of it.
Demand is outrunning capacity. Bid volume per team is up 24% over five years; estimating headcount is up just 4%. The gap is the story.
The capacity no-bid separates the field. 63% of teams declined qualified work they wanted, because no one was free to estimate it in time. The teams pulling ahead have already stopped.
Half the pipeline never gets priced. Median bid coverage is 45%. Firms pursue fewer than half the qualified opportunities they identify. Top-quartile firms pursue 68%.
Experience is the scarce input. 35% of industrial estimators are 55 or older, and 38% of chief estimators expect to retire within five years. The bench behind them is thin.
The knowledge is not written down. 64% say their most complex trades depend on one or two individuals whose methods aren't documented.
Ramp time is brutal. It takes a median of 20 months to bring a new estimator to the point of bidding a complex package solo.
Piping is the heaviest desk. Process piping and mechanical packages take the most estimator-hours of any trade, which makes it the natural first bottleneck.
Accuracy is a coin toss past 10%. 41% of teams say their typical estimate-to-actual variance exceeds 10%, squarely in margin-erosion territory.
A quarter of the desk is re-work. Estimators spend roughly 24% of their time re-checking, chasing missing information, and redoing takeoffs.
The top quartile is pulling away. Leaders bid nearly 2× the coverage, win at 38% against 19%, and are 5× more likely to run AI-assisted takeoff.
The bottom line. Every firm in this study is bound by the same wall: how much qualified work its estimators can turn into credible, on-time bids. The firms that widen that wall - through people, process, and increasingly automation - grow. The firms that leave it fixed give up revenue one no-bid at a time.
The estimating function today
Industrial estimating remains a craft desk running on general-purpose tools. The median team is small, senior, and stretched. The software stack underneath it has hardly moved toward automation.
The typical U.S. industrial estimating team in this study runs four to six estimators against a bid pipeline that would have been staffed by twice as many a decade ago. The work is concentrated in a handful of senior people: across respondents, the median firm carries roughly one estimator for every $18–22M of annual bid volume, and the most complex trades sit with the most experienced heads. This is a desk that rewards judgment. It also punishes any interruption to it.
The stack hasn't caught up to the workload
Spreadsheets and PDF markup still run the industry. 89% of teams estimate in Excel and 71% mark up drawings in Bluebeam or a PDF tool. On-screen takeoff and dedicated estimating software are common but far from universal, and purpose-built AI takeoff sits at just 14%. As Part 8 shows, that number is about to move.
A field still mostly upstream of automation
Sorted by how work actually gets done, only a small minority of teams have reached a connected or AI-assisted workflow. Three-quarters still sit in the manual-to-digitized band: quantities produced by hand, one sheet at a time, then keyed into a spreadsheet. This is not a failure of ambition. It is the reality of a market whose tools were built for commercial buildings rather than process piping and multi-discipline industrial scope. Note that Fig. 1 counts tools a team has access to, while Fig. 2 counts how the work actually gets done. A team can own AI takeoff software and still run a manual desk.
Excel + PDF/paper
on-screen takeoff
integrated platform
automated takeoff
The workload
The demand side is rising on every axis at once: more bids, denser documents, tighter deadlines, and more trades packed into a single package. Capacity has not kept pace.
The clearest signal in the data is a widening gap between how much work lands on the desk and how many people are there to do it. Bid volume per team has climbed steadily, up 24% over five years, while estimating headcount has crept up just 4%. Every point of separation between those two lines is work that has to be absorbed by the same people, in the same hours, at the same deadline.
The macro backdrop reinforces it. U.S. construction spending continues to tilt toward the most estimating-intensive work in the market - data centers, semiconductor fabs, grid and clean-energy buildouts. And the labor gap behind it is now a retirement story more than a growth story. ABC's 2026 model puts the need at 349,000 net new workers this year, rising to 456,000 in 2027, and attributes most of that demand to people leaving rather than new work arriving (Associated Builders and Contractors, 2026 Workforce Shortage Analysis, January 2026). More scope, older desks, and no slack in the system that prices the work.
Every package is now a multi-trade package
Scope complexity is compounding the volume. 71% of industrial bid packages now require quantities across three or more trades - piping, structural steel, electrical and instrumentation, and insulation, all reconciled against one another on a single job. Each added trade multiplies the reconciliation burden. Each is a place where a missed interface becomes a missed cost.
The capacity
The people who can price industrial work are the most valuable asset in this study, and the scarcest. The risk is not that they are experienced. It is that almost nobody has written down what they know.
More than a third of industrial estimators are 55 or older, and 38% of chief estimators in this study expect to retire within five years. This mirrors the wider trade: roughly one in five U.S. construction workers is now over 55, and the National Center for Construction Education and Research projects that about 41% of the construction workforce will retire by 2031 (NCCER). The most capable estimators in the industry are also the ones with the least time left at the desk, and most firms have made no plan for that.
A single point of failure on the hardest work
Because the expertise is concentrated and undocumented, so is the risk. 64% of teams say their most complex trades depend on one or two individuals whose estimating methods live in their heads rather than in a system. When that person is on vacation, out sick, or retired, the firm's ability to bid that scope leaves with them. That is a failure of the firm's record-keeping, not of the estimator.
One respondent described it plainly. The only estimator at the firm who can price a compressor circuit is 58. He went in for surgery the week three bids were due. All three were no-bid. Nothing about his work was written down, so there was nobody to hand it to.
is the median time to bring a new estimator to the point of bidding a complex industrial package on their own.
Hiring does not solve it quickly. 74% of respondents call it difficult or very difficult to hire experienced industrial estimators, consistent with the 92% of construction firms nationally reporting trouble filling open roles (AGC/NCCER, 2025), and the 57% who say candidates lack the necessary skills. Even a strong hire is nearly two years from full productivity. Capacity, in other words, cannot simply be bought when it's needed.
The squeeze
This is where rising demand and constrained capacity collide. The result is measurable. Work takes too long to price, so most of the pipeline never gets bid, and growth is capped by the desk rather than by the market.
Where the hours go
The unit of the problem is estimator-hours per package. On a mid-size industrial job, process piping and mechanical is the heaviest desk of any trade, followed by electrical and instrumentation. These are the high-value, high-complexity scopes that win or lose the job. They are also the ones that consume the capacity a team has least of.
The capacity funnel
Follow a firm's pipeline through the desk and the leakage is stark. For every 100 qualified opportunities a team identifies, a median of only 45 are actually bid, and about 10 are won. The first drop, from identified to bid, is the capacity drop. It is not a sales problem or a pricing problem. No amount of effort could have produced a credible number in the few hours that were left. The top quartile turns roughly 68 of those 100 into bids. The gap between 45 and 68 is capacity, and capacity can be built.
This is the mechanism behind the headline finding. When capacity is the ceiling, teams triage, and triage means declining the work you would win. 63% of firms did exactly that in the past year. Every one of those no-bids is a growth decision made by the calendar rather than on the merits. The firms that widened the desk stopped making that trade.
“Builders will have to create capacity through efficiency, not headcount.”
The cost of getting it wrong
Speed pressure and thin capacity don't just cost bids you never submit. They degrade the quality of the ones you do - and in industrial work, an estimate that's wrong is more expensive than one that's late.
The estimate is the single highest-leverage document in a project's life, and the research is unambiguous about what happens when it's off. In the Flyvbjerg megaproject database of more than 16,000 projects, nine out of ten run over budget (Flyvbjerg, Oxford), and the causes that predict overruns are front-end failures - absent or incomplete estimate detail, thin scope definition, inexperienced pricing - not weather or markets. The number that wins the job is the number that governs its margin.
Against that backdrop, 41% of industrial teams report a typical estimate-to-actual variance above 10% - wide enough, on a thin-margin job, to erase the profit entirely. And the errors are self-inflicted as often as not. Manual takeoffs carry a well-documented 5 to 10% fatigue error rate, and 47% of respondents attribute at least one margin-eroding job in the past year to a takeoff or scope miss.
The rework tax
Getting it wrong also consumes the very capacity that's already scarce. Estimators spend roughly 24% of their time on re-work: re-checking numbers, chasing missing information, and redoing takeoffs when a drawing or assumption changes. That pattern echoes the field, where the Construction Industry Institute puts direct rework at an average of 5% of project cost, and where poor data and miscommunication were tied to more than $31 billion in annual U.S. rework (FMI / CII). A quarter of a scarce estimator's week spent correcting avoidable error is capacity the firm can least afford to lose.
The benchmark
Here is the line. Score your own desk against the scorecard below, then work out your Estimating Capacity Index with the interactive dashboard that follows. Ten minutes here is the most useful thing you can do with this report.
| THE METRIC | LEADING top 25% | MEDIAN | LAGGING bottom 25% |
|---|---|---|---|
| Bids submitted per estimator, per year | 41 | 27 | 18 |
| Hours per mid-size package (takeoff + price) ↓ LOWER IS BETTER | 19 | 32 | 46 |
| Bid coverage, % of qualified pipeline pursued | 68% | 45% | 31% |
| Bid win rate | 38% | 22% | 14% |
| Estimate-to-actual variance ↓ LOWER IS BETTER | ±6% | ±10% | ±16% |
| Rework as a share of estimating time ↓ LOWER IS BETTER | 14% | 24% | 34% |
| Months to ramp a new estimator to solo bids ↓ LOWER IS BETTER | 13 | 20 | 26 |
| Estimating Capacity Index (0–100) | 71 | 54 | 41 |
Find your own number in each row, then see which column it sits closest to. On the shaded metrics, lower is better (hours, variance, rework, ramp). On the rest, higher is better. Leading is top-quartile performance.
The Estimating Capacity
Index calculator
Set each slider to your desk's real number from last year. The index updates live, scored exactly as published in the methodology note: each metric interpolated against the study's bands, rolled into four weighted pillars.
In this study the median firm scores 54, and capacity climbs with size, as larger firms invest in process, specialization, and tooling that smaller teams cannot yet justify. A desk sitting exactly on the median in all seven rows scores 55; all-lagging scores 40, all-leading scores 70. The published band values are 41, 54, and 71 because no real firm is uniformly lagging or leading across all seven rows, and the mix is what pulls the numbers apart. Full coefficients are in the methodology note.
What top-quartile firms do differently
The spread between the best estimating desks and the rest is not marginal. It is a different operating model, and the gap is widening fastest where automation is entering.
Rank the sample by Estimating Capacity Index and the top quartile separates on every axis that matters. They bid nearly twice the share of their qualified pipeline. They win at almost double the rate. Each estimator turns out far more bids per year, and they are roughly five times as likely to run AI-assisted takeoff. None of this is an accident of size. It is the compounding result of removing the manual bottleneck from the desk.
rate
coverage
per estimator/yr
takeoff
The pattern behind the numbers
Top-quartile desks share four habits. They industrialize the count, pushing repetitive quantity takeoff off senior estimators so judgment time goes to scope, risk, and price. They reconcile across trades systematically rather than trade-by-trade, which is where multi-discipline industrial bids are won or lost. They qualify harder, using capacity deliberately on the work they are positioned to win instead of spreading it thin - a discipline that independent analyses find lifts win rates by 8 to 12 points on its own. And they measure the desk, treating capacity as a number that moves rather than a fact of life.
The through-line is simple. Leaders treat estimating capacity as an asset to be engineered rather than a fixed headcount to be rationed. Capacity is a lever they pull, and that is what the next section is about.
Where industrial estimation is heading
The direction of travel is set. Estimating is moving from a manual, senior-only craft to an AI-assisted, human-in-the-loop discipline. The firms building that muscle now will define the next decade of the category, and the gap is closeable from where you are today.
Adoption of AI-assisted takeoff has roughly doubled in two years and is accelerating. In the broader market, 61% of contractors now use AI or plan to increase investment (AGC/Sage, 2026), and adoption among the top-100 GCs has crossed 60%. The tools that estimators trust are the ones that show their work, where every quantity can be traced back to the line on the drawing it came from and checked in seconds. An estimator can accept a number, or overrule it, on the evidence in front of him. Among industrial estimating teams specifically, intent is running well ahead of today's usage.
(planned)
The estimator is not replaced. The bottleneck is
This is not automation of the estimator. The industry consensus, and the clear preference of experienced estimators, is a human-in-the-loop model. The machine produces the raw count and a first-pass number in minutes. A senior estimator owns scope, productivity assumptions, pricing, and final accountability. Roughly half of estimating tasks are automatable (industry research). The other half is judgment that no model infers from a drawing. Tools that hide the review step and hand back a locked, unexplainable number draw the most skepticism, and rightly so.
What changes when capacity stops being the constraint
Return to the funnel in Part 4. If the count no longer gates the desk, bid coverage rises toward the pipeline a firm can actually identify. More qualified work gets a credible, on-time number. The no-bid stops being a capacity decision and goes back to being a judgment call. That is the prize, and it is why capacity rather than demand is the number to watch.
of industrial estimating teams agree that estimating capacity, rather than demand, will constrain their growth over the next three years.
Industrial contractors are bounded by one wall: how much qualified work their estimators can turn into credible, on-time bids.
Widen that wall and you grow. Leave it fixed and you no-bid your way out of the market, one qualified opportunity at a time.
Every finding in this report points to the same lever. The workload is rising and will not stop. The experience on the desk cannot be rehired quickly, and almost none of it is written down. The cost of a rushed bid is measured in erased margin. But none of that is fixed. The firms already engineering their estimating capacity - through people, process, and increasingly AI-assisted takeoff - are pulling away from the field, and they started from the same place you are reading this from. The constraint is real, it is measurable, and it is the single most useful number in the business to move.
Get your desk's numbers.
Start with your own number. Work through the scorecard in Part 6, then see how your desk compares against firms your size. It takes ten minutes and costs nothing. If you want to go further, join a State of Estimation readout - a small working session on video where we walk a real capacity funnel line by line with about a dozen estimating leaders. No slides and no vendors in the room. Most people meet the room through the podcast first. Details at thetakeoff.ai.
The RFP Pros State of Industrial Estimation Survey was fielded January–February 2026 to 312 estimating professionals at U.S. industrial contractors, recruited through the RFP Pros community. Percentages are rounded. Where medians are reported, they are firm-level medians. Survey figures reflect self-reported practitioner data for this edition. Third-party figures are cited inline with the report title given in full, and are drawn from the U.S. Bureau of Labor Statistics, Associated Builders and Contractors (2026 Workforce Shortage Analysis), AGC of America and NCCER (2025 Workforce Survey), the Flyvbjerg Oxford megaproject database, the Construction Industry Institute, FMI, and published 2025 to 2026 industry adoption surveys (AGC/Sage 2026 Construction Hiring and Business Outlook, Bluebeam, Dodge/CMiC).
Estimating Capacity Index. Each scorecard row is scored 0 to 100 by piecewise linear interpolation on the study's own bands, anchored at 40 (lagging), 55 (median), and 70 (leading), capped at 0 and 100. Below and above the median respectively, the row scores are: bids per estimator, 40 + 1.67(B−18) and 55 + 1.07(B−27); hours per package, 40 + 1.07(46−H) and 55 + 1.15(32−H); bid coverage, 40 + 1.07(C−31) and 55 + 0.65(C−45); win rate, 40 + 1.88(W−14) and 55 + 0.94(W−22); variance, 40 + 2.50(16−V) and 55 + 3.75(10−V); rework, 40 + 1.50(34−R) and 55 + 1.50(24−R); ramp months, 40 + 2.50(26−M) and 55 + 2.14(20−M). Pillar scores are the mean of their rows, weighted 30% coverage, 25% throughput, 25% accuracy, 20% resilience. The index is calibrated to the benchmark bands rather than derived from raw responses.
RFP Pros is a community of roughly 14,000 estimating, preconstruction, and bidding professionals. ContraVault AI builds ATLAS (thetakeoff.ai), an AI-native estimation engine for industrial contractors. It automates quantity takeoff across piping, structural, E&I, insulation, and equipment, and every quantity traces back to the drawing it came from, so the estimator stays in control of scope, pricing, and final accountability. This benchmark is published to give the industry a shared, measurable picture of the estimating desk, and a line to measure against.
