How to Reduce BOM Cost

Geometry, bare-board fabrication and the choice of a silicon each deserve their own review. This article is the structure of the list: custom versus catalogue, minimum order, alternates, cables, packaging, and the unit cost you only get if every line arrives in the same week. Figures are hypothetical teaching maths, not BrahmWorks prices, yields, timelines or test results. How that list becomes something a factory can buy against is part of How to Take a Hardware Prototype to Production.

A line's cost is price times quantity consumed, plus whatever you had to buy and will not consume, plus the tooling or setup that exists only because the line is custom. A connector at INR 3 that forces a custom cable assembly at INR 40 is not a three-rupee decision. A display with a minimum order of five hundred, on a first build of eighty, is not the unit price on the distributor page. The other four hundred and twenty are inventory, cash and an obsolescence risk.

Rank lines by extended cost at the quantity you will build this year, not by unit price and not by the prototype invoice. Fasteners look cheap and multiply. A machined spacer looks minor and carries a drawing, a supplier and a lead time. Packaging is a line even when it lives in another spreadsheet. If it ships with the product, it is in the product's cost. So is a spare that you promise in the box and forget on the bill.

A catalogue part pools you with everyone else who buys it. You inherit their volume and their risk of discontinuation. A custom part can be the product: a moulding, a flex, a specific sensor. It can also be a habit, a bracket left over from a prototype that was never redesigned. Custom earns its place when it performs a function no catalogue part will, at a total cost you have written down. "It was in the first build" is not that function.

Alternates are not free savings. A second manufacturer that is not fit, form and function compatible is a second design. Putting it on the bill without a trial splits your quality in half and your buying power as well. A true alternate, qualified and documented, is a supply hedge. An unqualified "or equivalent" note is an invitation for the builder to choose. If you want them to choose, qualify the set. If you do not, remove the words.

Variants of your own product do the same damage from the inside. Two colours that need two overlay tools, three regional power ratings that need three magnetics, a "lite" model that still carries the expensive connector: each split divides the quantity that justified the tool and the reel. Delete a variant that does not have a buyer, or give it a bill that actually differs, so you are not paying for the full product and selling a discount.

The teaching product is a wall-mounted gateway. It is not a BrahmWorks product. The bill has a custom aluminium spacer under the radio module, two board-to-board connectors where a single flex would do, four screw types, a cable assembly drawn as "wire, 80 mm" with no manufacturer, and a carton specified by a render. The spacer exists because the first printed housing was short. The housing has since been retooled in CAD, and nobody deleted the spacer.

Teaching edits, not a cost-down project report. Delete the spacer and confirm the module still meets its keep-out in the new housing. Replace two connectors and a hand-made wire with one catalogue flex, accepting a higher line price in exchange for deleting an assembly step and a polarity error. Collapse four screw types to one, after checking that the longest boss still has engagement. Specify the cable as a manufacturer part, or accept that every builder will invent one. Add the carton, the insert and the label as lines so the "unit cost" is the thing that leaves the dock.

None of these edits touch the radio's specification. That is deliberate. The expensive integrated circuit may dominate the bill and still be the wrong place to hunt, if the function is the radio. Hunt the lines whose only function is to paper over an old geometry or an unfinished drawing.

Hypothetical teaching maths. Not a distributor contract and not a BrahmWorks result.

Assume a display at INR 420 each, with a minimum order of 500. The first production lot needs 80. Cash out = 420 × 500 = INR 2,10,000. Units not yet needed = 420. Inventory value parked = 420 × 420 = INR 1,76,400. If you instead buy 80 through a broker at an invented INR 610, cash out = 610 × 80 = INR 48,800. You pay more per unit and you do not carry 420 pieces of a display the mechanical design might still move. The cheaper unit price is the more expensive decision at this quantity. Revisit it when the design is stable and the next order will consume the minimum.

Assume the custom spacer costs INR 70 each at this quantity, plus a drawing. Deleting it saves 70 × 80 = INR 5,600 on the first lot, which will not impress anyone, and 70 × 2,000 = INR 1,40,000 at an assumed later volume. The larger reason to delete it is the lead time and the chance it is forgotten on a build. A small extended cost can still be a bad line if it is a process of one. Price is not the only column. It is the column that stops you calling every deletion a victory when the rupees are trivial, and that stops you ignoring a minimum order because the webpage looked cheap.

• Lines are ranked by extended cost at the quantity you will build, not by unit price.
• Minimum orders and leftover inventory are in the comparison.
• Packaging, labels and promised spares are on the bill.
• Each custom part has a function a catalogue part cannot perform.
• Alternates are qualified, or the words "or equivalent" are gone.
• Product variants that split volume have a real buyer and a real difference.
• A deleted part is deleted from the assembly, not only from the spreadsheet.
• The cheaper part still meets rating, footprint and the compliance claim.
• Cables are manufacturer parts, or you have accepted that the builder will invent them.
• The saving is checked against assembly time, not only against the part cell.

No. Unit price ignores minimum orders, scrap from a poor alternate, a cable that the connector forces, and a tool that only one custom part requires. Compute the cash you spend and the units you can actually build this year. A dearer catalogue part that removes a custom drawing can be the lower bill. A cheap part on a fourteen-week lead, bridged by a broker, often is not.

No. A second source you have not built is a fiction, and maintaining two qualified parts has a cost. Second-source the lines whose failure stops the line and whose alternates are real: a common connector, a passives case size, a memory that has a footprint twin. Do not invent a second source for a custom moulding. The second source for that is a second tool, which is a different decision.

Yes. Unlisted hardware is how two builders use two lengths and how a gasket durometer changes without a revision. If it is fitted, it is a line, with a manufacturer or a specification tight enough to buy against. "Screw, M3, black" is a family, not a part. Add the length, the head and the drive, or expect variety.

When it is the product's function, or when it replaces several catalogue parts and an assembly risk at a total you have added up. A custom housing is often that part. A custom spacer that recreates a catalogue standoff is not. Revisit custom parts at each architecture change. They have a way of surviving the problem they were drawn to solve.

Bring the bill, the quantity for this year, and the list of parts that exist only because the prototype needed them. BrahmWorks can separate a line that earns its cost from a line that is an old workaround with a price.

Review the next stage with BrahmWorks