Researcher and manufacturing cost-estimation specialist Iryna Honcharuk explains why part of the delay starts before an order ever reaches the machine.
American manufacturing is showing clear signs of recovery. According to the U.S. Manufacturing PMI, the index climbed to 55.6 percent, its highest reading since May 2022, while the Backlog of Orders Index rose to 55 percent, up from 50.5 percent. Growth showed up in transportation equipment, the category that includes aerospace.
Backlogs like these are usually explained by a shortage of equipment, staff, materials, or production time. But there’s another stage between a customer’s request and the moment a machine starts running: the technical evaluation and cost estimate. For mass production, that stage can be almost invisible. For a shop that builds small batches of complex, high-precision parts, it can become an engineering problem in its own right.
Pictured: Iryna Honcharuk
Iryna Honcharuk, an estimator at a precision-machining shop in California, built her company’s internal quoting and cost-estimation system in 2024, cutting turnaround time on complex quotes and letting the shop take on more technically demanding work without adding staff.
An Order Can Fall Behind Before Production Ever Starts
Before accepting an order, a manufacturer has to answer more than “how much will this cost?” The harder question is “how exactly will we make it?” If a shop has built something similar before, it can draw on existing data – the material, equipment, processing time, and sequence of operations are already known. A new part is more complicated: its geometry, tight tolerance requirements, or a combination of different processes can call for an entirely different production route. A specialist has to work out the equipment, the number of setups, the sequence of operations, processing and inspection time, possible technical constraints, and the risk of defects before a price can be justified.
“You can add equipment or add staff, but that doesn’t solve the problem if the shop still can’t figure out fast enough how a new order should be made and what it should cost,” Honcharuk said.
In her view, how quickly and accurately a shop can price non-standard orders should count as part of its production capacity, alongside the machines themselves.
Why Software Can’t Price Everything
Digital systems can already speed up quotes significantly, especially for repeat orders, where a new part resembles ones the shop has built before and enough historical data exists. The difficulty shows up where there’s nothing to compare against. Two parts might look similar on a drawing and use roughly the same amount of material, yet one might be finished in a few standard operations while the other needs an extra setup, custom tooling, EDM work, outside processing, or tighter inspection because of its tolerances. In cases like that, the price difference comes less from the part itself than from how it has to be made. That’s why, for complex work, Honcharuk starts the calculation with the production route, and lets the price follow from there.
Also Read: How Research Quality Improves Biotech Reproducibility
“The less a new order resembles what the shop has already made, the less useful a straight comparison with past data becomes,” she explained. “You have to understand the manufacturing technology first, and only then calculate the cost.”
That distinction also marks where automation is useful and where it isn’t. Repeat decisions are worth handing to software. Wherever a truly new production situation appears, it still calls for engineering judgment.
Where Shops Lose Time
According to Honcharuk, one way to find the problem is to trace a customer’s request from start to finish, from the drawing to the finished quote. If most of the time isn’t going into the arithmetic itself but into clarifying the technology, searching for comparable past orders, coordinating operations, re-checking data, and figuring out whether the part can even be made a given way, the bottleneck is defined not by how fast a particular person is working. The crucial factor is how the process itself is organized.
“Every time a new order comes in, the first question is what we’ve already solved and what’s new,” Honcharuk said. “If we’ve kept the reasoning behind past decisions, not just the final numbers, a specialist can build on that instead of starting over. That’s where their time should go, toward situations nobody’s worked out yet.”
In practice, that comes down to a handful of principles. Standard and non-standard requests get separated as early as possible, so repeat orders follow proven rules instead of forcing a specialist to start each calculation from zero, while complex parts get routed quickly to someone who can assess the whole production path. The calculation starts with the sequence of operations, equipment, number of setups, quality control, and main risks, with the price built from that analysis rather than the other way around. Shops keep not just the final prices of past orders but the reasoning behind them, so the next specialist understands why a previous estimate looked the way it did instead of inheriting a number with no context. And automation gets aimed first at what repeats: trying to automate a non-standard process before a shop has worked out its own rules for it usually just moves the same uncertainty into a new piece of software.
The Root of a Pricing Error
Honcharuk also studies this problem as a researcher. In a paper published in 2025 on optimizing cost estimation in high-precision manufacturing, she examined how drawing interpretation, the sequence of production operations, resource use, and final cost connect to each other.
The practical takeaway is simple: a pricing error can form long before anyone starts doing the math. Get the tool access, the number of setups, the required precision, or the complexity of inspection wrong, and the error shows up first in the production plan. From there it carries into the time and labor calculations, and only then into the final price. So the accuracy of a quote depends as much on the engineering analysis behind it as on the pricing formula itself.
“Adding more specialists doesn’t change the underlying logic of the calculation,” Honcharuk said. “What matters is that machine time, labor, technical constraints, and risk get evaluated by clear, repeatable rules.”
In 2024, a similar principle was applied to rebuild the estimating process at Honcharuk’s company: recurring parameters started being evaluated against a shared set of rules, while individual expert judgment was reserved mainly for non-standard situations. The value of that shift shows up in where a specialist’s time actually goes: away from decisions the shop has made before, and toward the ones it hasn’t.
Where Judgment Still Matters
The future of manufacturing calculations probably isn’t a choice between a person and a program. Standard, repeating tasks will keep getting automated. If a shop has made the same type of part hundreds of times, the need to manually repeat the same analysis each time drops away..
At the same time, the specialist’s role is shifting. Their value moves from running repeat calculations toward working with uncertainty: understanding an unfamiliar production route, spotting risk, identifying the necessary operations, and setting the first rules for a new type of order. Once that’s done, parts of that decision can be reused, and eventually automated.
Honcharuk’s research in other areas of high-precision manufacturing touches a similar problem: finding the stages that disproportionately extend a production cycle. Sometimes that’s quality control, and sometimes it’s the processing technology itself. For cost estimation, that constraint appears before production ever starts.
That’s why backlog statistics only show part of the picture. They capture how many orders a shop hasn’t finished yet, but say almost nothing about the projects that haven’t even reached the schedule, because the company still needs to work out how to make them, how long it will take, and what price it can quote without real risk of a costly mistake. For standard production, that calculation can take minutes. In high-precision manufacturing, it often becomes an engineering problem of its own.
If the next wave of automation is going to meaningfully speed up complex manufacturing, shops will need to optimize more than their machines. They’ll need to optimize the decisions made long before those machines ever start.
