Companies with messy BOMs usually get the same thing wrong

"We have two BOMs. Engineering has one, production has one." That sentence is the starting point of almost every BOM mess. The problem is not whose copy is more accurate. It is that nobody has been assigned to keep the two consistent. This post breaks down how two BOMs grow apart, why the real version-control problem is the effective date and not the version number, why substitute parts never make it past word of mouth, and the most painful question of all: what to do with work in progress after the BOM changes.

Johnny Yang11 min read
Companies with messy BOMs usually get the same thing wrong

"We have two BOMs. Engineering keeps one in the drawings folder. Production has its own Excel file."

The person saying this was a vice president, as calmly as if he were talking about the weather. I asked him which one was correct. He thought for about three seconds and said, "Depends what you're asking. For costing, use engineering's. For issuing materials, use production's." Then I asked how far apart the two were. He laughed a little and said nobody had ever checked.

So we checked. We picked the ten finished products with the highest shipping volume, pulled the material lists from both sides, and compared them. Only a handful matched. In one, a screw was missing. Engineering had removed that screw three years ago, but production never deleted it, so the warehouse issued one extra every time, and whoever picked it up tossed it into a parts bin. In another, there was an extra cable, because the floor had long since switched to a different spec, but the engineering drawing was never updated, so costing still used the old one. The worst case was the same part number using different units in the two lists. One side counted "pieces." The other counted "meters."

I'm Johnny Yang, Brightstream Technology in Taiwan. I've led implementation teams for seven years, across 252+ projects in 32+ industries. I've done this kind of reconciliation many times, so here is the conclusion up front: When BOM management goes wrong, it almost always goes wrong in the same place. Your company has two BOMs, and nobody's job description says "keep these two consistent." It's not that nobody is responsible. It's that everyone is responsible only for their own copy. Engineering owns "what this product should look like." Production owns "what materials this batch actually needs to pull." Both sides do their jobs well. But the path in between, the one that translates a design change into a picking list, has never been assigned to anyone. So the differences pile up day by day. Costs are off, stock counts don't reconcile, wrong materials get blamed on purchasing, defects get blamed on the floor. Before you implement any system, you need to answer one very plain question: if engineering changes a part tomorrow morning, whose hands does that change pass through, and in how many days does it become a warehouse picking slip? If you can't answer that, what you implement will just be a tool that moves the mess online.

First, you're right: engineering and production should see things differently

I know some readers want to push back here: two BOMs isn't necessarily bad. An engineering BOM and a manufacturing BOM are supposed to be different.

That's correct, and it matters. Engineering's version describes the product's structure. It cares about which parts make up the machine, and each part's spec and drawing number. Production's version describes the manufacturing process. It has to deal with which station each material goes in at, whether there's a scrap rate, whether packaging counts, and whether outsourced subassemblies get broken out. For the same machine, engineering might write three levels, while production splits it into five to match the actual station order. Nobody copied anything wrong. These are two different uses.

So I won't tell you to merge them into one. The real problem isn't "there are two." It's that there is no clearly defined conversion rule between them, and nobody is responsible for running that conversion again when the source changes. Engineering changes a part. How does that information reach production? The answer I hear most on site is "we tell them." Telling means saying it out loud, in the hallway, or dropping a screenshot in a group chat. After that, nobody goes back to confirm whether production's sheet actually got updated, because confirming it is nobody's job.

To put it more bluntly: in many companies, the production BOM isn't maintained. It's worn into shape by "fix it once we find out it's wrong." A bad build, a short pick, a customer complaint, and only then does someone add a line. A sheet like that looks more accurate over time. In reality it's only accurate where you've already stepped. Everywhere you haven't stepped yet is a landmine.

Version control is really about the effective date, not the version number

When version control comes up, many people's first reaction is "We number our revisions. Rev A, Rev B, it's all written down." Numbering revisions is good. But it only solves half the problem.

A revision number answers "how many times has this list changed." It doesn't answer the harder question: from when does this revision count? The day it's signed off? The next new work order? After the next batch of incoming material is used up? Those three answers lead to completely different behavior. What I usually find on site is that nobody has ever said, so everyone has a different answer in their head. Engineering thinks it's effective on sign-off. Purchasing thinks it waits until old stock is used up. Production control thinks work orders already open obviously stay on the old version. The warehouse issues whatever the person in front of them asks for.

So you get a very typical picture: the same finished product, shipped in three batches this month, built from three different combinations. And afterward nobody can tell which batch was which, because the work order only recorded the finished part number, not which BOM revision was applied. When a customer reports a problem with one batch and you need to trace which batch and which materials, all you have is the warehouse keeper's memory and purchasing's receiving slips to piece it together.

So the first thing version control needs isn't prettier revision numbers. It's writing the "effective method" explicitly into the change order. The same change can be effective immediately, effective once old stock is used up, or effective after a specified work order. Each fits different situations. The key is that someone ticks a box on every change, instead of leaving it to a default.

Effective methodWhen it appliesWhat happens if nobody specifies
Effective immediatelyMandatory changes for safety, regulation, or customer requirementAll work in progress already started becomes out of spec, and nobody knows
Effective when old stock is used upPure cost optimization or alternate specsPurchasing keeps buying new material while old stock sits in the warehouse until it's scrapped
Effective after a specified work orderCustomer-requested phasing, old and new models running in parallelThe floor decides which version to use on its own, and one batch ends up with mixed materials

Substitute parts: the piece of the BOM most likely to live only in conversation

Substitute parts are where I most often see something that "doesn't exist in the system but gets used on the floor every day."

The floor's logic is actually very reasonable. This capacitor is out of stock. Another brand has the same spec. The veteran technician knows it can be swapped and it won't affect function. That judgment is usually right, often closer to reality than the official process. The problem is that the judgment never enters any document. The BOM still lists the original part. The picking slip still lists the original part. The warehouse issues the other one, and the books deduct the original. Three months later at stock count, the original is on the books but not on the shelf, and the other one isn't on the books but is piled up in a corner. That's when people start asking whether someone is stealing materials.

And substitution isn't just one question of "can it be swapped." Some substitutes are fully equivalent and can be swapped anytime. Some require the customer's written approval, especially customers with spec sheets or sample approval records. Some can be swapped but cost very differently, which affects the quotation. And some require adjusting process parameters, which changes labor hours. When all of these are mixed together and passed along with one line, "that one's fine to swap," something will eventually go wrong.

What needs to happen in practice isn't complicated, but someone has to make the call: move substitution relationships out of the veteran technicians' heads and onto a sheet, with the substitution level and constraints noted. This has nothing to do with systems. You can start on paper or in Excel. The real difficulty is that someone has to go ask every veteran technician and write down what they all just know, and in most companies that role has never been assigned.

What happens to work in progress after the BOM changes

This is the most painful question in this post, and the one least often thought through in advance. The moment a change order is signed, things are already running on the line. The warehouse has kits already prepared. Some work orders are at station three. Some are finished and waiting for inspection. Some subassemblies are out at a subcontractor's plant. Without rules for handling these, every change turns into another meeting, and every meeting reaches a slightly different conclusion.

There are really only a few ways to handle it. What matters is deciding in advance who makes the call and on what basis. Option one: finish everything already started on the old revision, and only new work orders use the new one. This is the least effort, but if the change was made for quality or safety reasons, carrying on just sends the problem downstream. Option two: stop everything and rework. Disassemble what can be disassembled, scrap what can't. This is the cleanest and the most expensive, and someone has to own the scrap on the books. Option three: split at a boundary. Anything past a certain station stays on the old revision, anything not yet there switches to the new one. This fits reality best, but it has one prerequisite: you have to be able to say exactly which station every open work order is at right now.

That prerequisite is exactly where many factories fall down. Either production reports are entered all at once before the end of the shift, so times and sequence don't line up, or reporting technically exists but only records quantities, not stations. I broke down how to diagnose this in "Everyone reports production, so why can't you trust the numbers?" Here I'll only cover the effect on BOMs: if your work-in-progress positions aren't real, you can never calculate the scope of a change. You can only bet the whole batch one way.

One more warning. During implementation, many companies ask for the system to "automatically handle version switching for work in progress," and then commission a pile of custom-built work for it. But if your team never agreed internally on how to handle it, what gets built is one person's opinion at that moment, hard-coded into the software. Three months later a different situation comes up, and it has to be changed again. I have a way to decide what should be custom-built and what should be fixed in the process first in "How far should ERP customization go?" Here's the one-line version: don't rush to write rules into the system before there's agreement on them.

So when should you actually implement a system

There are times you should. I don't want to make it sound like BOM systems are never necessary.

If your products are multi-level and highly variable, with a dozen-plus option combinations on the same model, so that each customer order with different boxes ticked becomes a different material list, then maintaining that in Excel without errors is nearly impossible, because the combinations explode. In that case the system's value isn't "managing versions for you." It's letting you maintain the shared parts only once.

The other case is when traceability is already chasing you. Customers start requiring that shipment lot numbers map to the raw material lot numbers and the BOM revision used, or your industry already has audit requirements. Flipping through paper by hand can't keep up with that, and it usually fails exactly when you can least afford a mistake.

On the other hand, if your current pain is just "the two sheets don't match," don't rush to buy anything. Two sheets that don't match is a process problem, not a tool problem. Move it into a system and you'll get two screens that don't match, and they'll be harder to change than Excel.

So what should you do tomorrow

Do two things tomorrow. Neither costs money.

First, pick the ten finished products with the highest shipping volume, pull up engineering's version and production's version, and compare them line by line. You don't have to finish in one go. Two a day is fine. As you go, record what kind each difference is: engineering changed it and production didn't follow, the floor changed it and engineering doesn't know, or the unit or part number was simply wrong to begin with. These three kinds have completely different causes. Lump them together and all you'll see is "it's a mess." Separate them and you'll quickly see which end the leak is on.

Second, pick one person and write "keep the two BOMs consistent" into their job description. Not as a side task. Write it down explicitly: no design change order is closed until this person confirms production's version is synced and the effective method is noted. This person doesn't need to be senior, but they need the authority to send back change orders that aren't filled out properly. What most companies lack isn't capability. It's that authority to send things back.

Do both of these for a full month, and you'll have a list of differences and a change process that has actually been walked end to end. Then talk about whether to implement a system and what to buy, and the questions you ask vendors will be completely different. If you get halfway through comparing and find so many differences you don't know where to start, talk to me for thirty minutes, and I'll help you figure out which end the leak is on first.

Found this useful? Pass it on
30 minutes · free

Bring the one thing you are stuck on — a free 30-minute diagnostic

You do not need to know what system you want. Tell Johnny Yang about the one thing that keeps blocking you, and in 30 minutes you will know whether to skip it, build it with coaching, or pay for a custom build. No pitch — if we cannot help, you will hear that straight.

Johnny Yang replies himself · limited spots each month · or just book a free 30-minute diagnostic

Johnny Yang (熊董)
Johnny Yang (熊董)
Founder, Brightstream Technology

Serial founder leading a small team that builds systems into an owner's everyday work.

As a founder
7 yrs
Projects delivered
252+
Industries served
32+
About Johnny