With AI and estimation, winning more work isn’t everything. AI can help construction companies capture tribal knowledge by documenting how experienced people recognize risk, evaluate options, make decisions, and handle exceptions. The goal is not to replace those experts. It is to turn what they know into a reusable company asset that helps more people make better decisions.
I can usually find the most valuable undocumented asset in a construction company in less than an hour.
I ask the estimator who they call when a number does not feel right.
I ask the project managers who they call when a job starts slipping.
I ask the superintendents who they call when the drawings do not match what is in the field.
Very often, everybody gives me the same name.
That person may be a senior estimator, an operations manager, a general superintendent, or the owner. They have been around for 20 or 30 years. They remember the job that looked exactly like this one. They know which customer requires extra documentation, which scope gap will become a change-order fight, which sequence looks good on paper but falls apart in the field, and which subcontractor needs to be managed differently from everybody else.
The company depends on that person constantly.
It just does not own what they know.
Their knowledge is not in the project-management system. It is not in the operations manual. It is not attached to the estimate. It is sitting behind one set of eyes, walking around the office with a cell phone.
Then one day that person retires, takes another job, gets sick, or simply stops answering every question.
Everybody acts surprised.
But here is the brutal truth: if critical knowledge can leave the company in one person’s truck, it was never institutional knowledge. It was borrowed expertise.
That is the biggest opportunity I see for AI in construction.
Not writing faster emails.
Not producing better meeting summaries.
Capturing the knowledge your company already has, standardizing how it gets applied, and making it available to the next person before the expert has to rescue the job.
Does AI-assisted estimating actually make a construction company better?
AI-assisted estimating makes a construction company faster, not automatically better. Pushing more bids through the same win rate, the same customer mix, and the same margins increases volume and workload without improving the quality of the business.
Faster isn’t the same as better.
For years, most contractors have judged their estimating department by a simple scoreboard:
- How many bids did we get out this month?
- How fast can we turn one around?
- Do we have enough capacity to keep up with what’s coming through the door?
Most track win percentage, too — and that’s exactly where the trap springs. Once you know you’re winning ten percent, the reflex is to treat that number as a fact of nature and just bid more of everything. Win rate stops being something you go to work on and becomes the very thing that pushes you into the volume game. Those aren’t bad metrics. But they’re incomplete — and they nudge you toward an arithmetic answer instead of a strategic one. Treat your win rate as a fixed fact, and growth turns into pure multiplication: push more bids through the same formula and the revenue line goes up. That’s not strategy. It’s just a volume play with the math done for you.
Here’s the part that gets skipped: win rate isn’t fixed at all. The most profitable contractors in the country aren’t stuck winning ten percent of their work — they’re often winning closer to a third. Not because they’re faster at takeoffs, but because they’ve earned their way onto better invitation lists. An owner or developer looking for a general contractor usually isn’t collecting ten bids. They’re getting three. General contractors run their trade bid lists the same way: three subs invited per trade, which means each one starts with something closer to a one-in-three shot before anybody even opens a number. That’s not luck. That’s a relationship, built over years, that gets you onto the short list before the pricing ever starts.
Let me put some numbers on it, and keep them round so the math is easy to follow. Say your company bids $400 million worth of work a year and wins about ten percent of it. That’s a $40 million business. Now AI comes along and lets your estimators push out thirty percent more bids without adding a single person. On paper, that’s a slam dunk. You’re bidding $520 million with the same team.
But then what?
If you’re still winning at the same ten percent, you didn’t build a better company, you built a busier one. You’re winning thirty percent more jobs at the same margins, from the same mix of customers you already have, and now you have to manage every one of them. More work for your PMs. More for accounting. More for your field crews. More for you. You didn’t fix the business. You just made it faster at doing the exact same thing. I’ve watched contractors do this and call it growth, right up until they realize you can drown in revenue.
So the question I put to that executive wasn’t “how do we bid more work?” It was, “how do we win more of the work that’s actually worth winning?” Those are two entirely different strategies. One says growth comes from volume. The other says it comes from better decisions. I’ll take the second one every time.
Are you building a bigger estimating department, or a better business?
Are you certain that you are chasing the work worth winning?
Ascent Consulting can walk your team through your own bid history (ustomer by customer, bid margin against realized margin) so you can see which relationships are actually building the company before you sign for a faster estimating tool.
Is all construction work worth winning?
No. Two customers can award identical contract values and leave a contractor in completely different places at closeout. One pays on schedule and releases retainage cleanly. The other disputes change orders, stretches payment, and negotiates one final concession before releasing money: turning a fifteen-point job into a five-point job.
Not all work is worth winning.
Here’s the piece the volume math quietly ignores: not every job is worth the same to you, and some aren’t worth much at all. Two customers can hand you the exact same contract value and leave you in two completely different places by the time the job closes out.
One of them is a pleasure. The scope is clear, they don’t fight you on every legitimate change, they pay on schedule, and when the work wraps they release your retainage without turning it into a negotiation. You make your margin. You’d happily do ten more just like it.
The other one grinds you the whole way. They nickel-and-dime every change order, argue the extras you’re actually owed, stretch out payment, and then at the very end (when all you want is to close the job and get your retainage released) they come back looking for one more concession. So you cut your final number just to get paid and move on. On paper, you “won” that job. In reality, you bid it at fifteen points and closed it at five, and it tied up your people the whole way through.
Now go back to the growth push. When you crank up bid volume and win more work at the same rate, you’re not quietly cherry-picking the good customers — you’re winning more of both. More of the profitable, easy jobs and more of the ones that bleed you at closeout. You didn’t get more selective. You got more of everything, including the everything you should have walked away from.
That’s why the real question was never “how do we win more?” It’s “who are we winning from, and what does it actually cost us to keep them?” Some of your customers are a five-point grind with a fight at the end. Others are a steady fifteen points and a handshake. If you actually knew —customer by customer, job by job — which was which, would you really keep spending equal estimating effort chasing both?
Here’s the part most contractors won’t say out loud: the relationship isn’t worth anything if it isn’t making you money. You wouldn’t work for free, so why keep working for almost free, for a customer who treats your margin like an opening offer? The smarter move runs the other direction. Take your most profitable partners and figure out how to do more with them. You can even give a point or two back to lock in the volume and still finish further ahead than you would grinding it out with the one who fights you to the last dollar.
I saw this play out firsthand years ago, at a trade contractor I worked for early in my career. Instead of chasing every general contractor’s bid list, we created a strategic business plan around building real relationships with the developers who actually controlled the work — and they started guaranteeing us a set number of projects every year. That relationship alone ended up being worth a third of our revenue. We didn’t win more bids. We won fewer, bigger, better ones, from people who’d already decided we were their contractor.
And the small, commoditized work that never really pays? Set a floor and let it go. A ten-thousand-dollar job can take about as much time to set up, submit, buy out, bill, and close as a thirty-thousand-dollar one. The little ones don’t sink you because they lose money on paper. They sink you because they eat the same hours as the good ones.
I know how hard that is, because I lived it back when I had my own contracting company. We had a service line that had underperformed on profit for years. We turned every dial we could reach, and the truth was simple: it was too commoditized to charge enough for. So, after three-plus years, I made the call to sunset it. That decision carried a real human cost. Someone on our team had built their role around that work.
That wasn’t a spreadsheet decision. It was a business decision and an HR reality, and you have to sit in both at once. But I made it, because holding onto work that doesn’t pay isn’t loyalty. It’s fear wearing a nicer hat.
And when you finally run the math honestly, it tends to surprise people. Say you’re running a $40 million company at a ten percent net: that’s $4 million in your pocket. Now picture cutting the worst of the commodity work and doing $32 million instead, but at twenty percent net: that’s $6.4 million.
Less on the top line, more in your pocket. Shown that on paper, almost every owner picks the smaller, more profitable business. But in reality, almost every owner does the opposite because walking away from revenue feels like losing, even when it’s the very thing that would make them more profitable.
What does a construction estimate actually cost you?
Every estimate costs estimator time, company overhead, management attention, and the opportunity that went unchased because the team chased this one instead. Estimating capacity is a finite investment, but most contractors manage it by keeping people busy rather than by measuring return.
Every estimate is an investment
Here’s a way of thinking about it that most owners never do. Every estimate you produce is an investment. It is your estimator’s time, your overhead, your management’s attention, and the other opportunity you didn’t chase because you chased this one. Every hour spent estimating one project is an hour you cannot spend on another. And yet almost nobody runs estimating capacity like an investment portfolio. The default goal is simply to keep everybody busy.
But busy and productive aren’t the same thing. I’ve sat with contractors who could tell me exactly how many estimates they produced over the last few years — but couldn’t tell me which customers actually made them the most money. They knew their backlog cold. They had no idea whether their estimating hours were flowing toward the relationships and clients that feed the company or the ones quietly bleeding it. If your estimators burn hundreds of hours a year chasing work you have almost no shot at winning, that time didn’t disappear. It got spent. It just didn’t buy you anything.
What can AI tell a contractor about their own bid history?
AI can connect a contractor’s technology stack (estimating software, ERP, project management platform, CRM, spreadsheets, accounting system, etc.) to answer questions the company could never afford to answer manually: which customers produce the best realized margins, which general contractors actually award work, which estimators win profitable work, and which project sizes return the most on estimating effort.
Your data already knows the answer
This is where AI actually gets me excited, and it has almost nothing to do with takeoffs. It’s the analysis. Most contractors are already sitting on a mountain of data — years of it, sometimes decades. It’s buried in the estimating software, the ERP, the project management platform, the CRM, a stack of spreadsheets, and the accounting system. The problem was never collecting it. The problem was that connecting it used to mean somebody exporting from four systems, cleaning it up in Excel, and building pivot tables for a week — so most companies never did, and ran on gut instead. AI collapses that week into an afternoon.
Which means, for the first time, the average contractor can actually ask the questions that matter. Which general contractors consistently award us work, and which ones just use our number to shop the job? Which customers produce our best margins? Which estimators aren’t just winning work, but winning profitable work? Which project sizes give us the best return on estimating effort? Where are we consistently handing margin back after the award?
Those aren’t estimating questions. Those are executive questions. And the same tool everybody’s buying to speed up takeoffs is the one that can finally answer them.
The contractors who pull ahead over the next decade won’t be the ones producing the most estimates. They’ll be the ones who get disciplined about where they spend the effort — who know their ideal customer, their ideal project size, and the kind of work that actually builds the company instead of just occupying it. AI doesn’t hand you that discipline. It just finally gives you the visibility to build it with confidence.
Should contractors buy an AI estimating tool?
Yes. Go evaluate the estimating platforms. Pilot them. Learn them. The productivity is real, the low-hanging fruit is right there for the picking, and the smart move is to run both tracks at once: chase the speed and do the analysis. They’re not a year apart. The same AI that reads your plans can read your history.
Please, don’t hear “strategy first” as “technology never.” That’s not what I’m saying at all.
Just don’t confuse the two. Productivity gets you more bids. Strategy gets you a stronger business. One makes the department faster. The other makes the company worth more. It was never really an estimating problem. It never is. It’s a strategy problem wearing an estimating uniform.
What should a contractor do before buying an AI estimating tool?
Pull the last eighteen to twenty-four months of bids into a single sheet — every job, won or lost — capturing bid amount, final contract amount, bid margin, and realized margin, then tagging each by customer, estimator, project manager, region, and project size. Running win rate and realized margin side by side against those tags shows which relationships are worth more estimating effort and which should be dropped.
Here's what to actually do
So, before you sign a contract for an estimating tool, do this right now, this week: Pull your last eighteen to twenty-four months of bids into one sheet. Every job, win or lose. For each one, capture the bid amount and the final contract amount, the margin you bid and the margin you actually closed at, and tag it by customer, by estimator, by project manager, by region, and by project size.
Win rate isn’t one number, it’s a calculation, and it looks different depending on which of those tags you run it through. A customer you win forty percent of the time might be your best account or your worst, depending on what your margin looks like at the end of each job. An estimator with the highest win rate might also be the one giving away the most margin to get there. Run win rate and realized margin side by side, sliced the same way, and the picture gets uncomfortable fast.
Find the handful of customers you’d happily do more work with, name the ones quietly costing you money, and decide out loud who you’re going to stop chasing. Then (and only then) go point that fast new tool at the work you actually want to win.
Technology can help you estimate faster, but it cannot decide which opportunities deserve your company’s attention. That decision has to be made strategically: by looking hard at the customers, relationships, project types, and margins that make the business stronger, then pointing the entire organization toward more of that work. That is where I suspect the real opportunity has been hiding the whole time.
Winning more work has never been the goal.
Winning better work is.
