The category error at the heart of affiliate media
Here is the conceptual mistake, stated plainly: treating “affiliate” as a business model instead of what it actually is, which is a distribution mechanism. A distribution mechanism is not a moat. It never was. It is a rented shelf in someone else’s store, and the rent, the shelf placement, and the store’s own survival are all decided by parties who owe you nothing contractually beyond the current commission agreement.
Byron Sharp’s mental availability framework makes a related point about brands: durable growth comes from being present and easy to buy across many contexts, not from maximizing efficiency in one. Apply the same logic one level up, to the business itself rather than the brand it promotes, and the conclusion is uncomfortable but simple. A media or affiliate business that is only present in one revenue context – one network, one attribution model, one payout structure – has optimized itself into fragility. It looks efficient on a spreadsheet. It behaves like a single point of failure in practice.
To diversify affiliate revenue is not to abandon the distribution mechanism that built the business. It is to stop mistaking the mechanism for the business itself, and to build the second and third contexts the model was always missing.
Three forces compressing the old model at once
Three separate pressures are converging on the same businesses in the same 18-month window, and each one alone would justify a strategy change. Together, they are not a cycle to wait out. They are a permanent repricing of the old model.
AI answer engines are disintermediating the click. Shopping-related queries on ChatGPT grew faster than any other query type in the year to mid-2025, and consumers already use AI assistants for price comparison (54%), deal-finding (41%), and review checks (41%). The recommendation increasingly happens inside the chat window. The click, and the commission attached to it, does not.
Attribution is degrading exactly as regulation tightens. Server-side tracking is expected to reach only 60-65% accuracy by mid-2026, connecting just 3-5 touchpoints across devices, at the same time the EU’s Digital Services Act (January 2026) and Meta’s own disclosure rules (March 2026) add mandatory compliance work with zero revenue attached to it.
The top of the market is pulling away from everyone else. The top 10% of affiliates already capture 67% of program revenue, and the global average commission rate sits at 8.3%, with no structural reason to expect it to rise. Scale, not effort, decides who benefits from the category’s growth.
No single one of these forces is fatal. Stacked together, they describe a model with a shrinking share of a growing pie for any operator that has not already added a second gear.