On 12 September 2025, the European Commission closed its antitrust case against Microsoft without a fine. The price of closing it: Microsoft must keep selling Office 365 and Microsoft 365 suites without Teams, at a lower price, for seven years (The Register). The case started with a complaint Slack filed in 2020. By the time it ended, Microsoft had already pulled Teams out of its suites in Europe in 2023 and stopped selling Teams-included suites to new customers worldwide on 1 April 2024 (TechCrunch).
Most product managers will never face a regulator over a bundle. The Teams case is still the cleanest illustration of what a bundle actually does: it changes which product the customer evaluates. A company already paying for Microsoft 365 never ran a chat tool comparison. Teams was simply there. That is the power of bundling, and it is also why a product bundling strategy deserves more thought than “package these together and add a discount.”
The economics fit in two papers
Adams and Yellen showed in 1976 that mixed bundling (offering the bundle and the separate pieces) can beat both selling only the pieces and selling only the bundle (Quarterly Journal of Economics). The mechanism is disagreement between customers. If one buyer values product A at $30 and product B at $10, and another values them the opposite way, separate prices force you to pick whose valuation you capture. A $40 bundle captures both.
Bakos and Brynjolfsson pushed the idea into digital goods in 1999. Bundle many items with near-zero marginal cost and a customer’s total valuation of the bundle becomes far more predictable than their valuation of any single item, which makes it easier to price (Management Science, via NYU Stern). They were explicit about the limit: for most physical goods, the cost of producing things the buyer never uses wipes out the gain.
So the question worth asking is not “do these products go together?” It is “do our customers disagree about which piece is worth more, and does each extra unit cost us close to nothing?” If both answers are yes, bundling is probably a good bet. If neither is, you are mostly giving away margin.
What a bundle buys you, and what it hides
The first thing it buys is retention. Ampere Analysis looked at people who came back to Disney+ after cancelling and found those who returned on the Disney+, Hulu and ESPN+ bundle were 59% less likely to churn again within 12 months than those who returned to Disney+ alone (Next TV). Leaving a bundle means giving up three things, not one.
The second is distribution for the weaker piece. That is the Teams lesson. A bundle can carry a new product into accounts it could never win on its own, which is useful at launch and dangerous later, because it hides whether anyone actually chose the product.
That hiding is the part I would underline. Telecom is the industry that turned internet, TV and home phone into the standard “triple play,” and running IT operations at a large telecom taught me what bundles cost underneath the sales story. Every package variant, promotion and grandfathered plan became another rule that billing and provisioning had to honor. When someone proposed retiring a component, the first question was what it would do to the bundles, and the honest answer was usually that nobody knew which piece was holding which customer. Usage data inside a bundle tells you what people touch. It does not tell you what they would pay for.
Then comes the exit cost. When Microsoft separated Teams, Forrester’s analysts called the change “a sneaky price hike” for new customers who still wanted Teams. Agree with them or not, the pattern is general: once customers anchor on a bundle price, almost any attempt to separate the pieces reads as a price increase to somebody.
Five checks before you package anything
- Valuation spread. Ask for willingness to pay by component, not for the package. If every segment ranks the components in the same order, a bundle discount buys you very little.
- Marginal cost. Software seats and content libraries pass easily. Anything that carries support hours, services or hardware needs real math before it goes in.
- Component measurement. Instrument engagement per component before launch. If you cannot tell which piece retains a customer, you cannot manage the bundle or ever safely change it.
- A standalone option. Mixed bundling is what the theory favors, and it doubles as research: the split between bundle and standalone purchases is the clearest signal you will get about what customers value.
- A written exit. Decide now how you would unbundle and what the price story would be. If there is no answer, treat this as a decision you cannot easily reverse.
Bundling is a pricing decision in packaging clothes. If you have already worked through where your free tier fence should sit and why pricing belongs to product strategy, this is the same question one level up: who values what, and are you charging each of them for it?