Meet Lily, a small toy seller on a major online marketplace. She relies on two key services: customer trend data to know what products to stock, and short-term loans to buy inventory before the holiday rush.

But new research by HKUST’s Professor Lijian Lu and colleagues shows these services don’t always reinforce each other. Using a game-theory model of sellers, banks, and platforms, the researchers examined how interest rates, loan sizes, and sales volumes shift when platforms provide both demand forecasts and financing.

Here’s what they found. When Lily gets access to the platform’s demand forecasts—say, that dinosaur puzzles will be hot sellers—she can align production more closely with expected sales. This is the “lower operational mismatch” effect: less waste, smoother operations, and higher revenue potential for both Lily and the platform.

But there’s a hidden cost. With more accurate forecasts, Lily is less likely to walk away from borrowing. That makes lenders—whether banks or the platform—comfortable setting higher rates. In the model, equilibrium loan rates climb above lenders’ own capital costs once information is shared. This is the “higher financial cost” effect, which quietly raises borrowing expenses and offsets part of the production gains.

Now consider David, a seller of custom phone cases. Traditional banks won’t fund him, so platform loans are his only option. Amazon, for example, launched a lending program in 2011 and has since issued more than $3 billion in cumulative loans to marketplace sellers in the U.S., U.K., and Japan. For small sellers, programs like these can feel like a lifeline. But when forecasts suggest rising demand, David suddenly needs a much bigger loan. That means higher exposure for the platform. If the platform keeps rates low, it risks financial losses; if it raises rates, sellers like David may find loans unaffordable.

The study shows this tug-of-war leads to misalignment: platforms are less inclined to share data if they also lend, and less willing to lend if they also share data.

One solution works. By charging a fixed subscription fee for data access, platforms can stabilize the system. A moderate fee lowers loan rates, reduces distortions, and makes both services complement each other. Sellers get reliable insights and affordable financing, while platforms avoid losses.

The lesson: more services aren’t always better. For e-commerce platforms, the real challenge isn’t how much support they offer—it’s how smartly those services are designed so they reinforce each other rather than pull in opposite directions.