When the Side Business Becomes the Better Business

One claim circulates in the secondary literature: that Sabre's revenue consistently accounted for more than half of American's total revenues, and that American made more money from Sabre than from flying passengers. It traces to a third-party academic compilation citing Hopper (1990) and Sabre (2002), and the primary accounting is not visible. Treat with caution. REPORTED

Sabre displayed American's flights preferentially, which produced the screen-bias controversies and regulation of computer reservation systems. The parent's ownership was the reason the conflict existed. FACT

As an independent company, Sabre has had a difficult run: taken private in 2006, re-listed, and traded poorly since. Separation was not a guarantee of durability. FACT

Amazon never separated AWS, and there is no public evidence that competitor discomfort has materially capped its growth, though some large retailers have publicly moved workloads elsewhere. FACT

The case demonstrates that an adjacent business with better economics can be retained indefinitely when three conditions hold: the parent can fund it, the parent's identity does not constrain it in its own market, and the parent reports it separately so the market can price it. INFERENCE

The part usually left out: on January 31, 2018, eBay announced Adyen would become its primary payments processor. The captive volume that had made PayPal attractive inside eBay was removed by eBay itself, three years after separation. FACT

The parent's volume was the adjacent business's scaffolding and also its limit. Independence forced PayPal to replace captive volume with outside merchants, which it did. It might not have. INFERENCE

A compensation agreement in late July 2011 gave Alibaba licensing revenue from Alipay and guaranteed proceeds from any future IPO or sale. Ant was reportedly raising at a $60 billion valuation by 2018. Alibaba's compensation was a negotiated cap agreed at 2011 prices. FACT

Yahoo and SoftBank were shareholders in Alibaba. They were not shareholders in the entity that came to hold Alipay. That distinction is invisible while interests align and becomes the entire question the moment they diverge. FACT

The lesson is about timing, not villainy: ownership resolved under deadline is resolved at the other party's price. INFERENCE

IBM did not lose because it failed to recognize software mattered. It lost because at the moment of contracting, exclusivity looked like a detail and the hardware looked like the business. INFERENCE

Moody's finished 2000 with $602.3 million of revenue and delivered $797 million of revenue and $212 million of net income in 2001, its first full year standalone. As of a September 2025 third-party compilation, Moody's market capitalization stood around $86 billion against Dun & Bradstreet at around $5 billion. FACT REPORTED

Two businesses that both looked like credit information had entirely different economics. Ratings are issuer-paid, embedded in regulation, and scale without proportional cost. Business credit reporting is a subscription data business with real competition. Thirty-eight years of common ownership obscured the difference. INFERENCE

In a spin-off, shares go to existing shareholders, so D&B holders were not deprived. The company as an institution was. INFERENCE

The stated mechanism is the most useful sentence in this study: as a subsidiary of an investment bank, MSCI faced limitations when competing for clients that were also Morgan Stanley's rivals, and when pursuing partnerships or acquisitions that raised conflict questions. REPORTED

The parent's identity was the cap. Not capital, not talent, not product. The adjacent business could not fully address its market while owned by a participant in that market. INFERENCE

This is the boundary condition: the only case where the core was fully shut down, showing what the pattern looks like with no ambiguity at all. INFERENCE

It does not show that internal tools generally become better businesses. Thousands of companies built the same thing and shipped nothing. The selection bias here is total and should be treated as such. INFERENCE

Sears did not fail because it spun off Allstate and Discover. It failed for reasons internal to retail. But the proceeds of separating structurally superior businesses were deployed into a structurally inferior one, and the institution that made that choice no longer exists while both separated businesses do. INFERENCE

Gross margin is the wrong test. The right test is contribution after the costs the parent does not have to spend to acquire that revenue. An adjacent business riding on the core's customer acquisition can be structurally better at the operating line while looking worse at the gross line, and structurally worse while looking better on growth. Both errors are available from the same income statement. INFERENCE

Rich Barton, in his own words: “We've determined the unpredictability in forecasting home prices far exceeds what we anticipated and continuing to scale Zillow Offers would result in too much earnings and balance-sheet volatility.” And on the call: “too risky, too volatile to our earnings and operations, too low of a return on equity opportunity and too narrow in its ability to serve our customers.” FACT

Revenue growth was the loudest signal and the wrong one. The adjacent business converted a capital-light marketplace into a balance sheet, and the core competence, a pricing model, did not transfer into a business where being wrong costs inventory rather than credibility. INFERENCE