PLANNED OBSOLESCENCE: BUILT TO BREAK, DESIGNED TO PROFIT
The popular version of planned obsolescence is an engineer instructed to make something fail. That version is rare, hard to keep secret, and in several jurisdictions illegal. The real mechanism is duller and far more robust: nobody decides to make the product break, and somebody decides not to spend nine euros preventing it. The nine euros is visible in the bill of materials. The thing the nine euros would have cost the company is visible only if you do a second sum that nobody is required to do.
That second sum is the subject of this note.
The mechanism
Revenue from a durable good in a mature market is not a function of how many people want one. Nearly everyone who wants one has one. It is a function of how often the installed base turns over.
Annual replacement demand equals installed base divided by service life. Service life sits in the denominator. That single structural fact means every engineering decision that lengthens life divides revenue, and every decision that shortens it multiplies revenue, with no marketing, no new customers, and no visible change to the product’s specification sheet.
The canonical documented case is the Phoebus cartel, formed in 1924 among major incandescent lamp manufacturers, which coordinated a 1,000-hour service life standard for lamps, supported by testing of member output. [UNVERIFIED: the cartel’s internal penalty schedules, per-member fine amounts, and exact dissolution date were not accessed for this note; the existence of the cartel and its 1,000-hour standard is the widely documented part.] What makes Phoebus historically unusual is not the incentive, which is permanent, but that the incentive was written down and signed. Modern versions do not require a signature. They require a spreadsheet.
The arithmetic of service life
Model a durable good: unit price 600 EUR, gross margin 35%, so gross profit per unit is 600 x 0.35 = 210 EUR. Model an installed base of 2,000,000 units in a saturated market where all demand is replacement demand.
| Service life L | Annual replacement units (2,000,000 / L) | Annual gross profit (units x 210 EUR) |
|---|---|---|
| 8 years | 250,000 | 52.5 M EUR |
| 6 years | 333,333 | 70.0 M EUR |
| 5 years | 400,000 | 84.0 M EUR |
| 4 years | 500,000 | 105.0 M EUR |
| 3 years | 666,667 | 140.0 M EUR |
Moving service life from 8 years to 5 raises annual gross profit from 52.5 M to 84.0 M EUR, a gain of 31.5 M EUR per year, or 8 / 5 - 1 = 60%. The product did not change. The market did not grow. Nothing was announced.
Now price the durability that was not bought. Suppose extending life from 5 years to 8 requires 9 EUR of additional bill of materials per unit: better capacitors, a sealed bearing instead of an open one, a gasket rated for the actual thermal cycle. At the 5-year production volume of 400,000 units per year, that is 400,000 x 9 = 3,600,000 EUR per year of additional cost.
Set the two against each other: 31,500,000 forgone gross profit against 3,600,000 of added cost, a ratio of 31,500,000 / 3,600,000 = 8.75 to 1.
This is the whole mechanism. The nine euro capacitor is not rejected because nine euros is a lot of money. It is rejected because the finance case for including it is a proposal to spend 3.6 M EUR in order to lose 31.5 M EUR. Any competent cost engineer will decline, and will be right, and will never once have thought about making the product fail. The decision presents itself as cost discipline. It is demand management wearing cost discipline’s coat.
The arithmetic of repair, and where it is actually decided
The second lever is not durability at all. It is the parts catalogue.
Model a common field failure. Repair requires one component and 1.5 hours of labour.
| Scenario | Parts | Labour (1.5 h x 52 EUR/h) | Repair total | vs 600 EUR replacement |
|---|---|---|---|---|
| Component sold separately | 62 EUR | 78 EUR | 140 EUR | 23.3% |
| Only a module is sold | 310 EUR | 78 EUR | 388 EUR | 64.7% |
At 23.3% of replacement cost the customer repairs without hesitating. At 64.7% the decision becomes a genuine calculation, and the calculation must include remaining life. A repaired unit at year six of an eight-year design has roughly 2 years left. A new unit has 8.
- Repair, component available: 140 / 2 = 70 EUR per remaining year
- Repair, module only: 388 / 2 = 194 EUR per remaining year
- Replacement: 600 / 8 = 75 EUR per remaining year
With the component available, repair wins at 70 against 75 EUR per year. With only the module available, replacement wins by 194 / 75 = 2.59 times.
No durability specification changed between those two rows. No engineer was told to make anything fail. A parts catalogue was written at module level rather than component level, and the customer, calculating freely and correctly, chose to replace. The choice was genuinely the customer’s. The arithmetic behind it was not.
The strongest counter-argument
There are three serious defences and they are not decorative.
The first is that replacement is often welfare-improving. A unit designed eight years ago is measurably less efficient, less safe, and less connected than its successor. Extending its life extends the life of the worse version. In categories where efficiency improves several percent per year, a long-lived product can cost the owner more in running cost than the replacement costs to buy.
The second is that module-level service is not primarily a revenue decision. Field technicians misdiagnose component-level faults at a materially higher rate than they misdiagnose module-level faults, so component-level repair produces more repeat visits, more warranty leakage, and more units that come back worse than they went in. Modularisation buys diagnostic reliability, and manufacturers pay for it in inventory value. [UNVERIFIED: comparative field misdiagnosis rates by service granularity were not accessed for this note.]
The third is the uncomfortable one. Where durable and disposable variants have both been offered at transparent prices, buyers have frequently taken the cheaper-now option. A manufacturer that unilaterally adds 9 EUR of unadvertised durability to a 600 EUR product in a price-sorted market is not being ethical. It is being outsold by 1.5%.
All three are true. None of them changes the 8.75 to 1 ratio, and none of them requires that ratio to stay unpublished.
Consequence
The lever that actually moves this system is not a durability mandate. Durability mandates are litigated into meaninglessness because service life is a distribution, not a number, and any manufacturer can find a duty cycle under which its product meets the standard.
The lever is disclosure of the two quantities the model above turns on: designed service life, and the price and availability horizon of serviceable parts, both stated at the point of sale. That does not tell the manufacturer what to build. It changes what the customer’s L is, and L is the denominator on which the entire 8.75 to 1 calculation rests. A market in which the eight-year unit can be identified and priced is a market in which the 9 EUR capacitor becomes a feature to advertise rather than a cost to remove.
Right to repair legislation is often argued as an environmental measure. On the arithmetic above it is more precisely a parts-catalogue measure, and the catalogue, not the capacitor, is where the decision was made.
The product was never built to break. It was built to be replaced, which is a different sentence, and a much harder one to prosecute.