Bundling works for a reason that is not obvious and is worth stating precisely: it makes money out of the fact that buyers disagree.

Two buyers, two products. The first would pay $100 for product A and $20 for B. The second would pay $20 for A and $100 for B. Price them separately at $100 and you sell one of each: $200. Price them separately at $20 and you sell four units for $80. Bundle both at $120 and both buyers take it: $240, with no discount to anyone and nothing changed about the products.

The gain comes entirely from the disagreement between the buyers. Their valuations are uncorrelated, and the bundle averages them, which reduces the variation the seller has to price around.

When it works and when it does not

Uncorrelated valuations is the condition. Where buyers agree about what is valuable, bundling adds nothing — everyone wants the same component and the rest is packaging.

Low marginal cost helps. Adding a component that costs nothing to include converts unsold capacity into revenue. Where each component has real cost, the bundle has to be priced to cover them all and the arithmetic tightens.

A bundle must be worth more than the parts in the buyer's eyes. Convenience, integration, one relationship, one invoice — some real reason beyond arithmetic, or it is a discount with extra steps.

Unbundling is the same insight reversed

Markets alternate between the two, and the alternation is not fashion.

Bundles are attractive to incumbents because they average preferences and hide the price of individual components. That averaging is exactly what leaves an opening: a buyer who wants one component is paying for six, and a competitor offering only that component at a lower price wins them — which is how almost every incumbent bundle gets attacked.

So the useful question about a bundle is not whether it raises revenue now. It is which component a competitor could profitably sell on its own, because that is where the attack comes from and it is visible in advance.

Practical forms

Good-better-best. The most common structure, and less about features than about letting buyers reveal willingness to pay by self-selecting. The tiers should differ in something the buyer cares about, and the gaps should be wide enough that choosing is a real decision.

Core plus optional. A base everyone needs with genuinely optional additions. Cleaner than a tiered bundle, and it forfeits the averaging effect that makes bundling profitable in the first place.

All-in retainer. The service-business bundle: a fixed monthly fee covering a defined scope. It works on exactly the mechanism above — clients use different parts in different months, and the fee averages across them. It fails when scope is undefined, at which point it stops being a bundle and becomes unlimited access at a fixed price, which is a different and much worse product. See productization for the version that holds.

The failure worth avoiding

Bundling to move things that do not sell. It is the most common reason bundles get built and the reason buyers distrust them.

Buyers are good at identifying which parts they did not want. Once they have, the bundle reads as padding, the headline price reads as inflated, and the seller has created precisely the opening described above — a competitor offering just the wanted component, at a price that looks honest by comparison.