Going heavy to fight a competitor's strength is the one move that gets reversed
Shopify built a logistics network to take on Amazon and sold it a year later for equity, cutting a fifth of its staff. Coca-Cola gave bottling away in 1899 and has been buying it back ever since. Sixteen decisions on record.
Asset-light is a position, not a philosophy
Sixteen asset-light decisions are on record and nine were never reversed. The set is small and unusually clear, because the same companies keep moving in both directions and the record catches them doing it.
Coca-Cola granted national bottling rights away in 1899 because Asa Candler believed bottling would fail and thought the men asking were doomed. Over 1,200 bottling operations existed by 1920. Coca-Cola has repeatedly bought bottlers back and still owns consolidated bottling operations today. Domestic bottlers fell from 278 in 1985 to 79 by 2002. The structure was not a principle, it was a hundred-year oscillation.
The reversals were all built to fight somebody's core competence
Shopify decided in May 2022 to build its own logistics network rather than stay asset-light, to compete head-on with Amazon's fulfillment empire. In May 2023 it sold the entire logistics business to Flexport for a 13% equity stake rather than cash, and cut 20 to 23% of its workforce.
Wayfair built an owned fulfillment network for bulky home goods from 2015 on the reasoning that owning the supply chain provides delivery speed and control that keeps Amazon from taking share. CastleGate grew, and the bespoke-moat stance was reversed by opening it to outside suppliers. Wayfair lost $492 million in fiscal 2024, narrowed from $738 million and $1.331 billion in the two years before.
Neither was a bad execution. Both were decisions to build, at a fraction of the scale, the specific capability their largest competitor had spent twenty years and tens of billions perfecting.
What the durable light structures kept
Hilton spun off its real estate in 2017 and the fee revenue and unit growth followed through 2024. Red Bull owns its content machine and never spun it off. Mars entered pet care in 1935 and it grew into the majority of company revenue. Seven-Eleven Japan went the other way entirely and acquired its own licensor.
The common thread is not how much is owned. It is that each company owns the thing that decides — the brand standard, the content, the format, the licensing terms — and is indifferent about owning the thing that executes. Marriott appears here having been contractually forced to buy a hotel it did not want, which is what it looks like when the boundary is drawn in a contract rather than a strategy.
How to run this decision
Drawn from what separated the records that stuck from the ones that were undone. A step nobody in the record can be seen taking is not here.
- 01
State which capability you are buying and who already has it at scale
If the answer is your largest competitor, and the reason for building is to stop them, price the gap honestly. Shopify and Wayfair both entered fulfillment against a company whose fulfillment is the product.
- 02
Distinguish owning the decision from owning the execution
Hilton sets the standard and does not own the buildings. Red Bull owns the content and not the distribution. Write down which of the two you are actually buying, because the case for them is completely different and companies routinely make the first case while buying the second.
- 03
Assume you will reverse it, and check what reversing costs
Shopify's reversal took twelve months and was paid in equity rather than cash, plus a fifth of the workforce. Coca-Cola's has taken a century. The reversal cost is the real price of the decision, and it is rarely in the paper.
- 04
Check whether going light is giving away a term, not an asset
Coca-Cola did not give away bottling plants in 1899; it gave away perpetual rights on terms nobody revisited for decades. An asset can be repurchased at a price. A term is repurchased at whatever the other side asks.
- 05
Size the build against the volume you have, not the volume you plan
Fulfillment, logistics and manufacturing are scale businesses, and a network sized for the forecast is a cost structure sized for a company that does not exist yet.
- 06
Put the boundary in a contract with a review date
Marriott ended up owning a hotel because of an obligation nobody expected to trigger. The boundary between what you own and what you rent should be written where it can be read, with a date at which it gets looked at again.
What it looks like when this is going wrong
- The build exists to match a competitor whose core competence it is.Shopify, 2022. Wayfair, 2015.
- The case is about control and speed, with no unit cost at your volume.Wayfair, 2015.
- Nobody has priced what reversing the decision would cost.Shopify, 2022.
- Rights are being granted in perpetuity because the activity looks unpromising.Coca-Cola, 1899.
- An ownership obligation sits in a contract nobody has read recently.Marriott, 2024.