Coil Slitting Optimization Software: A Buyer's Guide for Metal Service Centers (2026)

Most of what you can find online about this software category is old. The most-cited general-audience piece, The Fabricator's guide to coil optimization software, was published in 2007. The math has not changed since then. Almost everything else has: where the data lives, what a solver can chew through in a minute, and what a planner is expected to do in a day.

This is a buyer's guide, written from the planner's side of the desk. What the category does, the questions that actually separate vendors, how the four common approaches compare, and how to run a 30-day pilot that produces a number you can defend.

Quick answer: what is coil slitting optimization software?

Coil slitting optimization software decides which customer orders to cut from which master coils, and in what knife arrangement. It solves a constrained cutting-stock problem: minimize trim loss and knife changes while respecting gauge, grade, width tolerance, PIW, coil weight and due dates. The output is a runnable slitting schedule, not a suggestion.

What problem does it actually solve?

Take an order book with 300 open line items and an inventory of 400 master coils. Any coil can, in principle, serve dozens of those orders. Each combination implies a different arbor layout, a different amount of trim, a different set of knife changes, and a different amount of revenue booked this week instead of next.

That is the one-dimensional cutting-stock problem, and it is old. Gilmore and Gomory published the linear programming approach with delayed column generation in 1961, and the best algorithms today still build on that relaxation. It is also genuinely hard. The pricing subproblem inside column generation is itself NP-hard, and gets worse as the variety of widths and demands grows.

Which is why an experienced planner is good at this and still leaves money on the table. A person can hold maybe five or six coils in their head at once. The good combination is often the one involving a coil nobody was looking at, matched to an order that was not due for three weeks.

Two things separate real slitting from a textbook cutting-stock exercise:

• The constraints are physical. Gauge and grade have to match. Width tolerance is per customer, not global. There is a maximum number of knives, a minimum strip width, PIW and coil weight limits per strip, arbor and separator availability, and a slitter that can only take certain outside diameters.
• The objective is not just yield. A 98% yield plan that misses a truck is worse than a 95% yield plan that ships. Trim loss is the number everyone quotes. Due dates, knife changes and shipped tons are what the P&L actually reflects. We break the levers down line by line in our 12-point checklist for reducing scrap on a slitting line.

How is this different from what our ERP already does?

Your ERP knows the order book, the inventory, the tags and the genealogy. It records what happened. That is a different job from deciding what should happen.

Metals-specific ERPs do cover the processing side. Metalware from Paragon serves flat-roll, specialty metals, toll processors and fabricators, and the metals vertical at Jonas Software grew out of long-established service-center systems. These platforms let a planner build a slitting job, specify the arbor layout, record slit weights and run toll work.

The word doing the work in that sentence is "let". The system holds the layout the planner entered. It does not go find a better one. Deciding is a search problem, and search is what an optimizer is for.

What are the four ways service centers do this today?

1. Spreadsheet and experience. Free, flexible, and it works. A senior planner with fifteen years on the floor produces good plans. The problems are ceiling and continuity: they can only compare a handful of coils per order, and the method walks out the door when they retire. Most service centers are here.

2. An ERP processing module. You already own it, it already has your data, and it produces real work orders. It generally will not search the solution space for you. Good fit if your bottleneck is paperwork rather than yield.

3. A standalone optimizer. Purpose-built cutting-stock software. Coil Cut Optimizer from Fast-Square is one of the longer-running examples, described as software that "helps human programmers generate Optimized Cutting Patterns for Slitting and Cut-To-Length machines, minimizing Trim-Loss (Scrap) and Set-up time," respecting order requirements like minimum and maximum strip OD and weight, and machine constraints like maximum knife count and minimum strip width (Franz Inc case study). These tools are strong at the math. The cost is usually at the edges: someone has to feed them the order book and inventory, and carry the answer back into the ERP.

4. An AI command-center platform. The newer category, and the one we build: see what a command center for metal processing is for the longer definition. The optimizer is there, but so is the plumbing: pull orders and inventory from the ERP automatically, solve, let the planner edit, then write work orders back. LineSight's Command Center is one of these, built around a two-way ERP connection where plans flow back out as executable work orders. The bet is that the integration is the hard part, not the solve.

There is no universally right answer. If your yield is already 96% and your planners are drowning in transcription, option 2 or 4 beats option 3. If you have one line, one planner and a clean spreadsheet habit, be honest that option 1 may still be the best return on your attention this year.

What should you ask a vendor?

Ten questions. Ask them in this order, and make them answer with your data, not a demo dataset.

1. How does it get our orders and inventory? Named integration with our ERP, a nightly file, or manual entry? Ask who maintains the connection when the ERP is upgraded.
2. Which constraints does it enforce, and which does it ignore? Read the list out loud: gauge, grade, width tolerance per customer, knife count and arbor setup, minimum strip width, PIW, coil weight, OD limits, separators, multi-pass and re-slit, toll processing, dedicated coils. If any of those terms are unfamiliar to whoever is running the evaluation, the glossary is here. Every "we handle that in a future release" is a plan your planner will have to fix by hand.
3. How long does a solve take on our real order book? Not on 30 orders. On 300 orders and 400 coils. Under a minute is workable. Twenty minutes means nobody will re-run it after a change, and a plan you cannot re-run is a plan that goes stale by Wednesday.
4. Can a planner edit the plan and keep the math? Swap a coil, force an order onto a specific tag, drop a strip. Does the tool re-cost the plan after the edit, or does it just accept the change silently?
5. What comes out the other end? A PDF for the operator, a work order in the ERP, arbor setup instructions with knife positions and spacer stack. Ask to see the actual document a slitter operator would work from.
6. How does it handle must-ship orders? Real order books have lines that cannot slip. Ask what happens when the must-ship set alone is infeasible. A tool that returns nothing, with no explanation, will be abandoned in week two.
7. How does it handle partial coverage? Most weeks you cannot cover the whole book. Ask what it does with the remainder: does it tell you what to buy, what to leave open, and which orders it deliberately deferred and why?
8. What does a solve cost, and what does a seat cost? Per user, per line, per site, per ton, flat? Is the optimizer metered? Some vendors price against measured savings rather than seats, which is worth asking about if you have a clean baseline. Ours works that way: the first three months are free, the savings get measured on your live orders in that window, and we never charge more than we save. Whatever the model, get the renewal price in writing before the pilot, not after.
9. How long is implementation, and who does the ERP work? Ask for a date, a named integration owner, and what happens if your ERP is a 20-year-old on-prem system with no API. We publish under a month, including the custom ERP integration, configuration and training, and we do not charge until it is live in production. That is a reasonable benchmark to hold any vendor to, ours included. LineSight builds a dedicated integration for whatever ERP you already run.
10. What happens when the plan is wrong? Because it will be, in the first month. Ask how an override is captured and whether the system learns the floor-specific rule behind it, or whether your planner will make the same correction every week forever.

How do you run a 30-day pilot?

The goal of a pilot is not to like the software. It is to produce one number your owner believes.

Week 0: pick the scope and the baseline. One line. One product family. Pull the last 90 days of actual slitting jobs from your ERP: master coil widths and weights, orders assigned, trim scrapped, knife changes per shift. That is your baseline. Do not let it be an estimate.

Week 1: shadow mode. Run the software on the same order book your planner is working, but ship your planner's plan. Compare the two on trim, knife changes and covered tons. Expect the software to lose some of these. Note exactly why. Almost always it is a constraint nobody wrote down.

Week 2: fix the constraints. This is the real work of any implementation. The width tolerance that only applies to one customer. The coil nobody touches because it is allocated. The slitter that cannot run under 0.020". Feed these in and re-run week 1's book. If the vendor cannot encode a rule you consider basic, you have learned something important cheaply.

Weeks 3 and 4: run live on the pilot line. Your planner reviews and edits every plan before release. Track three things only: trim loss percentage, knife changes per shift, and planner minutes per plan. Also track overrides, and the reason for each.

The decision. At day 30 you should have a per-line delta on yield and hours, a list of constraints the tool could not express, and a count of overrides that were falling rather than steady. Falling overrides mean it is learning your floor. Steady overrides mean you are about to buy a tool your planner will fight every week.

Be suspicious of your own pilot if it was run on a quiet month, on your cleanest product family, or by the vendor rather than by your planner. All three inflate the result.

What it feels like when it works

The tell is not the yield number. It is that the argument about which coil to run stops happening. Abizer Ghadiali, VP at Elite Steel, put it this way: "LineSight has taken the guesswork away from trying to maximize a coil… because of LineSight, my employees are working more efficiently, and we're making more money on every coil." Elite Steel reports +$850K in annualized revenue added, a 2.9% average scrap-rate reduction, and a +20% gross profit margin increase. That is one operation with one product mix, which is why the pilot above matters more than anyone's case study.

That is the honest bar for this category, whoever you buy from. The planner stops guessing. The judgment calls stay human, and the search gets handed to a machine that can hold four hundred coils in its head at once.

FAQ

What is coil slitting optimization software? It is software that decides which customer orders to cut from which master coils and in what knife arrangement, minimizing trim loss and setup time while respecting gauge, grade, width tolerance, knife count, strip weight and due-date constraints. It produces a slitting schedule and arbor setup that an operator can run.

Does slitting optimization software replace our ERP? No. The ERP remains the system of record for orders, inventory, coil genealogy and invoicing. The optimizer reads that data, decides what to cut, and writes work orders back. Any vendor proposing to replace a working metals ERP is selling you a much larger and riskier project than the one you asked about.

How much trim loss can optimization software actually remove? It depends entirely on your starting point, product mix and inventory depth. A shop already running tight patterns on a narrow product range has less to gain than one with a wide book and deep inventory. Treat any headline percentage from any vendor, including us, as marketing until you reproduce it on your own 90 days of history. That is what the pilot is for.

Can this work with an older on-premise ERP that has no API? Usually yes, but ask specifically. Common paths are a read-only database connection, a scheduled file export, or a bridge that drives the ERP screens directly. LineSight builds a dedicated integration per customer, so an older on-premise system without an API is a normal case rather than a disqualifier. The same applies to getting orders in: see AI order entry for metal service centers. What matters is who owns the connection long term and what happens when the ERP is patched. Get that named in the contract.

What is the difference between a standalone optimizer and an AI command-center platform? A standalone optimizer solves the cutting-stock math and gives you patterns; you handle the data in and the work orders out. A command-center platform connects to the ERP in both directions and automates the surrounding steps, so the plan arrives as an executable work order rather than a file to retype. Standalone tools are often cheaper and narrower. Platforms cost more and remove more manual work.

Bilal Yousuf is co-founder of LineSight, which builds planning software for metal service centers.