AI orchestrator: How recommendation algorithms shape complementor strategy and market equality
Platform recommendation algorithms act as orchestration tools: popularity-based ones push complementors to specialise, personalisation-based ones spur broader new products.
Summary
The study examines a major food-sharing platform through two changes in its recommendation algorithm, from a popularity-based to a personalisation-based one, using more than 1.7 million observations of seller behaviour. Under the popularity algorithm, sellers concentrated on a smaller set of core products and were less likely to launch new ones. Under the personalisation algorithm they widened their portfolios and introduced more new offerings to reach diverse tastes. The market effects also differ: popularity ranking reduced inequality by exposing long-tail sellers and limiting superstar dominance, while personalisation increased it, because firms with stronger capabilities and richer data exploit it better. The authors conclude that recommendation algorithms work as a soft form of control that shapes competition and the distribution of value, so platform owners must balance generativity with control. The evidence is from a single platform.
Editor's opinion
A recommendation algorithm is a strategic lever, not a neutral ranking tool: the logic a platform chooses tells its sellers what to do. Popularity ranking narrows their offers but spreads visibility more evenly, while personalisation broadens their offers and tilts the market towards the best-resourced. Platform owners should choose the algorithm knowing which kind of ecosystem, and which degree of inequality, they are building.
Key concepts
recommendation algorithms · platform orchestration · complementor strategy · market equality