Applying Seven Powers to Platform Businesses

Executive overview

Getting to product-market fit and getting to power are two distinct problems that require two separate inventions. For platform businesses this distinction is especially treacherous: the very technology that lets a platform scale fast also makes it easy for competitors to replicate it and for customers to multi-home.

Hamilton Helmer and Chenyi Shi introduce three diagnostic questions for assessing platform power: how is economic value created and how does it scale with participants; how does each customer segment perceive that value; and what prevents a competitor from reaching equivalence.

Flywheels signal product-market fit, not power — a platform with genuine power must show a material, sustainable performance advantage over rivals that participants cannot costlessly arbitrage away.

What makes platforms different from other businesses

• A platform is any intermediary for transactions — the concept predates digital technology by millennia.
• Technology is the primary driver of new platforms because it radically lowers transaction costs (search, input, delivery, matching).
• Platforms create value through matching heterogeneous buyers and sellers; better matching is the core economic engine.
• Two distinct step changes in company value: achieving product-market fit, then achieving power. Each requires a separate invention.
• In platforms, these signals often point in opposite directions — rapid scaling can coexist with zero power.

The three diagnostic questions for platform power

1. How is economic value created on your platform, and how does it change as the platform scales? Quantifies the benefit side of power.
2. How does each customer segment perceive that value, and how does perception shift with scale? Requires segment-level granularity; the Amazon seller (maximise units at market price) optimises very differently from the eBay antique-watch seller (maximise price from a rare buyer).
3. What prevents competitors from reaching equivalence? This is the barrier side — the power question proper.

Multi-homing and why scale alone is not enough

• Multi-homing occurs when customers freely use competing platforms in parallel, arbitraging away any differential value.
• If both riders and drivers always open the Uber and Lyft apps simultaneously, relative density becomes irrelevant — both platforms feed the same effective pool.
• Ride-sharing has diminishing returns to density; once density is "good enough" in a region, additional scale yields little extra value, making a durable lead hard to sustain.
• Meta-search tools (e.g., a ride-aggregator app) are friction reducers that accelerate multi-homing and disintermediate platform power.
• Contractual lock-in or genuine switching friction on at least one side is required to prevent rivals from erasing a scale advantage.

Preference heterogeneity and why it determines whether scale compounds

• The rate at which returns to scale diminish depends on how heterogeneous the transactions are.
• Uber/Lyft: the ride is nearly undifferentiated — only distance to driver matters. Density curves flatten early.
• YouTube: content preferences span language, topic, creator personality, production quality, and mood. The space is so high-dimensional that even a massive content library is never "good enough" for all edge cases.
• High heterogeneity means a platform's scale lead keeps compounding: more watch-time data improves the algorithm, which improves matching, which attracts more creators and viewers.
• Operators should explicitly map the heterogeneity of their transactions to gauge how important scale will remain at maturity.

YouTube as a platform power case study

• YouTube monetises viewer attention with advertising, then redistributes revenue to creators — a pricing schema where one side pays and the other is paid.
• Viewers are split between intent-driven searchers (seek specific content) and passive consumers (rely on the algorithm to surface content). Both segments perceive high, sticky value.
• Even if a copycat held every YouTube video, it would lack accumulated watch-time signals and personalisation history — the search-cost advantage is embedded in the data, not the content.
• Creator multi-homing costs are low (upload the same video elsewhere), so power resides mainly on the viewer side.
• Rising ad loads have not driven viewer attrition — evidence that surplus leader margin remains large.

When and how to start capturing value

• Surplus leader margin: the maximum price premium a platform can charge versus the best alternative while maintaining leadership. It is set by differential scale, not by management discretion.
• During the takeoff phase, customers are underpriced; it is the right time to acquire share aggressively and sacrifice short-term margin.
• Tightening monetisation makes sense only after the performance gap versus rivals is wide enough that the price increase does not push customers to switch.
• TSMC's below-maximum pricing locks in long-horizon customer commitments, enabling billion-dollar fab investments and securing scarce upstream supply (ASML EUV machines) — a rational strategy given lumpy capital and predictable Moore's Law advances.
• Apple's 30% App Store rate is maximally extractive but sustainable because developer and consumer switching costs are both high and neither group finds alternatives compelling.

Network effects vs. network economies

• A network effect occurs when a new participant makes the platform more valuable to existing participants.
• Indirect network effects (a new Uber driver benefits passengers) are common but do not automatically produce power.
• Direct network effects — where a participant on the same side creates additive value (a new Facebook friend benefits existing friends) — are rarer and more likely to produce winner-take-all dynamics.
• Network economies (Hamilton's working term): power that results specifically from direct network effects. Both a material benefit and a barrier to arbitrage must exist.
• Network effects describe value creation only; power additionally requires that the value differential is not competitively arbitrageable.

The flywheel trap

• Nearly every startup pitch deck contains a flywheel, but a flywheel is a symptom of product-market fit, not evidence of power.
• Test: swap the company logo in the flywheel diagram with a competitor's logo — if the flywheel still works, there is no power embedded in it.
• The hard follow-on question is always: why does one platform scale faster or sustain a larger lead, and what prevents rivals from matching it?
• Achieving power is a second invention, every bit as difficult as achieving product-market fit. Founders who assume out-innovating rivals forever is sufficient usually face commodity competition eventually.