In almost every ERP system in machine building, the same bolt lives several lives. Search for M12x40 and you will find it as "Bolt M12x40 8.8", as "M12 x 40 hex zinc plated", as "Hex bolt DIN 933 M12x40" and once more under a description only its creator understands. Four numbers, one physical part, and nobody who dares to say which one is the real one.
Duplicates like these rarely come from sloppiness. They exist because at the moment of placing a part, searching takes more effort than creating, and because nobody feels ownership of the item master. The good news: preventing duplicate items in your ERP does not start in purchasing or administration, but on the CAD side, where the item first lands in an assembly. With clear ground rules and a link between Inventor and your ERP, like the one Thundercad provides, reuse becomes the easiest route instead of merely the virtuous one.
In this article: how duplicates arise, what they really cost and how to prevent them structurally with three measures: enforce search before create, make reuse easier than creating new items, and dedupe on a fixed rhythm.
How one bolt can exist four times
Trace the history of a duplicate item and you almost always land on one of these four causes:
- Search is slow or unreliable. Whoever searches three times for "M12x40" and finds nothing because the item was entered as "M12 x 40" concludes it does not exist, and creates it.
- Descriptions have no fixed structure. One person starts with "bolt", another with the standard, a third with the size. Sorting and searching then always break down somewhere.
- Time pressure wins. Right before a deadline, creating an item takes two minutes; searching thoroughly and checking a doubtful case takes fifteen.
- Nobody is the owner. When everyone may create items and nobody manages the file, nobody corrects the growth either.
The deeper cause is usually the distance between CAD and ERP: the engineer works in Inventor while the item master lives in another system, with another screen and another search box. Every switch between those two worlds is a moment where "quickly creating a new one" becomes more attractive than searching properly.
What a duplicate item really costs
A duplicate bolt looks harmless until you follow the chain. Inventory: the same part sits in two locations with two reorder points, so there is structurally too much of it. Purchasing: the total volume is split across two numbers, so negotiations happen twice and the volume discount is never reached. Shop floor: two bins holding the same part, and still the occasional "out of stock". Engineering: where-used analyses that tell only half the truth, and BOMs of two nearly identical machines that differ on paper while physically nothing differs.
A quick calculation, with assumptions you are free to adjust: suppose sorting out one doubtful case (is "M12 x 40" the same item as "Bolt M12x40"?) takes ten minutes on average, and that this happens a few times a week across engineering, purchasing and work preparation. That adds up to a few hours of pure detective work every week, apart from the errors that occur when someone under pressure picks the wrong number.
Search first, create second: make it the easiest route
Behaviour follows the path of least resistance. As long as creating takes two minutes and proper searching ten, every work instruction loses to daily practice. The fix is therefore twofold: agree that nobody creates an item without searching first, and make searching so fast that the agreement enforces itself.
Speed starts with findability: one fixed structure for descriptions (type, standard, size, material, finish), so that "hex bolt DIN 933 M12x40 8.8 zinc plated" can only exist in one form. Teams that agree on such a format and gradually align the old descriptions notice that searching suddenly does return results.
The second accelerator is searching in the place where the engineer already works. With Place article from Thundercad you consult your ERP's item master directly from Inventor and place the item you found straight into your assembly. Using the existing item then literally takes fewer clicks than creating a new one, and that tips the choice towards reuse for good.
When the existing item is found and placed within seconds, "quickly creating a new one" loses its appeal all by itself. See what that does to your item master.
Try 30 days freeGround rules that channel item creation
Speeding up search is half the job; the other half is guiding the creation itself. Three ground rules that prove themselves in practice:
- One owner. Appoint one role, often work preparation or a senior engineer, who creates new items or at least approves them. Requests come in with a short description in the fixed format; whoever reviews the request searches once more first.
- Mandatory fields. An item without a standard, size or material cannot be found and is therefore a duplicate in the making. Make those fields mandatory at creation.
- Fast turnaround. A request process that takes days gets bypassed. Agree that a request is handled the same working day, otherwise the process itself creates time pressure again, and time pressure is exactly where duplicates thrive.
And the item number itself? The less meaning it carries, the less debate at creation and the smaller the chance that two people construct "the same clever number" slightly differently. Whether to give numbers meaning or deliberately not, and what that means as your file grows, is what we discussed in Part numbers that scale with you: meaningful or meaningless numbering?.
Dedupe periodically, without drama
Prevention limits the inflow, but the existing file only gets clean through a recurring dedupe round. A workable recipe: sort the item master by normalised description (spaces removed, everything lowercase, the "x" made uniform) and by supplier number, and lay doubtful cases side by side. Pick one survivor per group, usually the number with the most stock or the most usage.
Do not delete the losers; block them for new use and point to the survivor in the description. That keeps old orders and history intact, while new BOMs automatically land on the right number. Update live BOMs at the next natural moment, such as a revision.
Deduplication ultimately touches the broader question of how CAD and ERP form one single truth: who manages which field, and which system leads. We wrote about that earlier in Article management between CAD and ERP: one item, one truth.
Frequently asked questions
Is one duplicate bolt really a problem?
One duplicate bolt is not; the pattern is. Where one is visible there are usually dozens, and every duplicate number multiplies through BOMs, stock locations and orders. The damage is not in the part itself but in all the detective work and the wrong picks around it.
Should I delete duplicate items or block them?
Block them. Deleting breaks the link with old orders, stock history and existing BOMs, and that makes the problem bigger than it was. A blocked item pointing to the survivor steers everyone in the right direction by itself.
Does searching from Inventor work with our ERP system?
You connect Thundercad to your ERP or other business software through the Dashboard, after which Place article searches the item master directly from Inventor. Whether that runs smoothly in your environment is easiest to find out in practice: try it 30 days free.