Tilda Rasmussen has a bill of materials pinned to the wall of her office with the countries written in the margin in biro. The 2021 version runs to 240 lines across eleven countries and four continents. The hinge on the lid of her camp stove — a stamped steel part costing 31 cents — was, on that version, touching three of them before it reached her bench. She used to describe this to people as efficient, because that is the word everyone had agreed to use.
The 2026 version has 240 lines and two countries. About 61% of her component spend now comes from inside a 400-mile radius. Her unit cost went up 22%, which she can quantify to the cent, and she calls it the best money the company spends.
The cost that isn’t on the invoice
The thing a long chain optimises is the one number that shows up on a purchase order. Everything it costs you is somewhere else in the accounts, filed under headings that don’t look like procurement. Rasmussen’s stoves used to require sixteen weeks of lead time, which meant sixteen weeks of guessing. Guessing means inventory, and inventory means cash sitting in a warehouse in the shape of hinges. She was carrying about $410,000 of it against $2.9M of revenue. That figure is now $95,000, and the difference is not a saving so much as a return of her own money from a place she couldn’t reach it.
The Brackett Group has been running the same numbers across 640 manufacturers under $25M and finds the pattern is not subtle. Firms that cut their median supplier distance by more than half reported unit costs up a median of 17%, and cash conversion cycles down from 94 days to 38. Roughly two-thirds ended up with more cash in the business than before, while every single one of them got worse at the metric their old operations consultant had been hired to improve.
The other line item is defects, which behave strangely at distance. A bad batch of hinges found in week two of a sixteen-week chain is a phone call. Found on arrival, it is a quarter. Rasmussen’s warranty rate went from 4.1% to 1.3% after the move, and she is careful to say she doesn’t fully know why — her best guess is that a supplier forty minutes away has met her, seen the stove, and will now ring up to ask whether something looks wrong, which nobody in the old chain had any reason to do.
We saved eleven cents a hinge and paid for it with a warehouse, a planner and my entire 2022. Nobody put those on the same page, because they weren’t on the same page.
What comes home and what doesn’t
Nobody is repatriating everything, and the founders who tried are the ones with the best warnings. The line lands in a similar place each time:
- Anything bulky, heavy or awkward comes back first. Freight is the argument and it makes itself — Rasmussen’s sheet steel now travels 70 miles instead of 7,000.
- Anything that changes often comes back. If you revise a part twice a year, you cannot do it by email at nine hours’ offset.
- Anything that fails visibly at the customer comes back, because the warranty cost is invisible until it is enormous.
- Electronics, mostly, do not come back. Nobody has found a way to make a small run of boards nearby without paying four times over, and the founders who claim otherwise are usually excluding their own labour.
- Fasteners, bearings and motors do not come back and probably never will. The volumes that make them cheap exist somewhere else, and a stove company is not going to change that.
The obvious objection is price, and the honest answer is that the customer paid it. Rasmussen raised the stove from $189 to $229 in 2023 and sold more of them. She is not claiming a marketing victory — she thinks about a third of the increase was absorbed by margin she used to give away in freight, and the rest was accepted by people who were already choosing her over a $60 alternative and had therefore already made the relevant decision. This does not work if you are competing on price. If you are competing on price, this whole article is a description of how to go out of business slightly closer to home.
The real risk is the one that replaces the old one. A chain that runs 400 miles is a chain with fewer participants, and Rasmussen now has a business where one press shop in one town makes a part with no qualified second source. She knows. She has been trying to dual-source it for two years and keeps discovering that the second source is either eleven weeks away or the same shop under a different name. The retreat from four continents is not a retreat to safety. It is a swap: distance risk for concentration risk, four unknowable failures for one knowable one.
Rasmussen’s view is that she would rather have a problem she can drive to. The 2021 bill of materials stays on the wall, she says, for the same reason people keep photographs of themselves at their worst weight.