5 questions to go. Your result appears here.
What the record shows
Three questions the record says matter most: why you would hand it out, whether you have the scale to make it, and how many suppliers you could turn to.
Handing something out to cut cost is where buying goes wrong: four of ten did
Of 59 make-or-buy decisions with a known outcome, 14 were later reversed or went badly. Where a company handed something out to cut cost or free capital, four of ten ended that way: Borders, Boeing, Nvidia and Nike. Where it handed something out for control, a capability or supply, one of seven did.
Borders handed its online store to Amazon; Boeing sold its Wichita fuselage plant to become a "lean systems integrator"; Nvidia moved its gaming chips to a cheaper foundry and moved them back two years later. Each was later undone: Borders relaunched its own site in 2008, Boeing agreed to buy the Wichita operation back, and Nvidia's next generation returned to TSMC.
Deciding to secure supply worked every time in the record
Where the main reason was to secure supply or depend less on a supplier, none of eight decisions went wrong and seven paid off as intended: Nucor, KFC China, Nestlé, Netflix, Tesla, Eli Lilly and one more.
The number of suppliers points the same way. Where a company chose to make something itself because there were few capable suppliers, one of 17 decisions went wrong. Across all decisions, those taken where there were many suppliers went wrong in four of 13 cases; those where there were few, in three of 23.
Without the scale to do the work yourself, four of six decisions went wrong
Where the company lacked the volume to do the work efficiently itself, four of six decisions went wrong: Borders, Stanley Black & Decker, Peloton and Intel. Where it had the scale, four of 19 did.
Intel built fabs "sized for an external customer base it did not yet have". When Peloton's demand fell, its owned factories "didn't flex".
Making something to learn a skill you lack is harder than making it for control
Where a company chose to make something itself to gain a skill or technology it lacked, four of 16 decisions went wrong: Kodak, Stanley Black & Decker, Shopify and Intel. Where it made something for control, cost or supply, three of 24 did.
What did not separate outcomes: how central the activity was
The usual rule says make what is central and buy what is standard. In the record, decisions about central activities went wrong in ten of 43 cases, and decisions about standard ones in three of 13: the same share. Why the company decided, and whether it had the scale and the suppliers, mattered more than how central the activity was.
Making went wrong somewhat less often than buying: eight of 41 decisions to make, against six of 18 to buy.
Warning signs, in the sources' words
- The case for handing it out is the saving. “The 2001 deal was a sensible cost cut.” (Borders: online channel (2001))
- You would hand out a step your product depends on. “Own the design, own the brand, own final assembly — and let someone else own the buildings” (Boeing: Wichita fuselage plant (2005))
- You would build capacity for demand you do not yet have. “funding fabs sized for an external customer base it did not yet have” (Intel: IDM 2.0 strategy (2021))
- Owned capacity cannot shrink when demand does. “When Peloton's demand reversed, the owned plants didn't flex.” (Peloton: owned manufacturing (2022))
Questions for the board before it decides
If handing this out were not cheaper, would we still do it?
Four of ten decisions to buy from outside to cut cost were later reversed or went badly, against one of seven made for other reasons.
Do we have the volume to do this at a competitive cost today, not in the plan?
Where the company lacked the scale, four of six decisions went wrong; where it had it, four of 19.
How many suppliers could take this work tomorrow, and what would each of them charge us once we depend on them?
Decisions made to secure supply went wrong in none of eight cases.
Have we done anything like this before, or would we be learning it on the job?
Making something to gain a skill the company lacked went wrong in four of 16 cases; making it for control, cost or supply, in three of 24.
If we get this wrong, what would it cost to reverse it?
Four of 18 decisions to buy from outside were later reversed, among them Boeing's sale of its Wichita fuselage plant.
What it is not
A forecast. The record holds 63 decisions to make or buy, researched from published sources: company filings and statements, and reporting. Of those, 59 have a known outcome and 4 are too recent to call. Decisions about where to make something (reshoring, a new country) rather than who makes it are not counted, and those often ended badly, so the record here may read kinder than make-or-buy decisions at large. Well-documented decisions lean toward the famous ones. Where the sources do not say, that case is left out of that question rather than guessed.