headshots of two professors


Ensuring fairness in product assortment: a ‘win-win-win’

Why it matters:

Product placement and display significantly influence what we choose to purchase. Ozge Sahin and Ruxian Wang study an algorithm that ensures minimum market exposure for all sellers which could benefit consumers.

Whether we’re shopping online or strolling the aisles of a brick-and-mortar store, product placement and display has a significant influence on what we choose to purchase, note Ozge Sahin and Ruxian Wang, both faculty members at Johns Hopkins Carey Business School, and Wentao Lu, a post-doctoral fellow at Carey.

Traditionally, the primary objective driving assortment planning or product recommendations has been to maximize revenue or profit — but this favors big sellers and can keep minor players and new or local players out of the market, say the researchers, who recently completed a study that examines what happens when “seller (producer) fairness” is introduced to ensure market exposure for all sellers.

It’s a timely issue. “The concern from regulators, as well as the importance of building justice into the platforms’ economic model, make fairness among sellers increasingly critical for platform management,” write Sahin and Wang, and Carey Business School colleague Wentao Lu, in their working paper, “A Simple Way Towards Fair Assortment Planning: Algorithms and Welfare Implications.”

Through their modeling and analysis, the researchers found that instituting seller fairness to product assortment or recommendation decisions provided a “win-win-win” solution — in which all involved parties are better off: consumers, new and existing sellers, and the selling platform itself.

More choices, happier consumers

In noting the importance of product assortment and display to consumer decision-making, the researchers point to data showing that only one-third of customers searching online for lodging options ever go beyond the first page of search results returned by a platform.

“Because there are so many similar products in any given category, product visibility is very important,” says Wang, a professor of operations management and business analytics at Carey Business School.

Traditionally the focus in the assortment planning literature has been on maximizing revenue, profit, or market share, notes Sahin, “and we have not been paying enough attention to the implications for sellers and their products.” In particular, minor, new, or local sellers suffer: If their products are not shown to consumers, they get no market exposure and therefore make little or no revenue, the researchers note, writing, “This has important economic implications for potentially monopolizing the market, in the long term harming even marketplaces.”

In their study, the researchers introduce an algorithm that ensures minimum market exposure for all sellers in their models. “We were pleasantly surprised by our findings,” says Sahin, a professor of operations management and business analytics at Carey Business School. “We found that when we take care of the sellers on the platform, consumers always win.”

That’s because buyers get a broader range of choices. “Normally they might be seeing two options,” says Sahin, “but when the fairness constraint is introduced, the platform might offer 10 product options. Consumers have many more options to choose from, with products that meet a broader range of their tastes.”

Crucially, the selling platform also benefits, notes Wang. “The platform is able to attract new sellers and new consumers,” he explains.

What to Read Next

A simple solution

Sahin says that the researchers’ findings have been well received, with some platforms indicating that the fairness approach described in their paper is “consistent with their current practices.”

She and Wang both note that the algorithm they developed and modeled to ensure fairness in their study is straightforward and “not overly complicated,” which would make deployment relatively easy.

“Instead of over-regulating and creating complicated policies, with our fairness algorithm a relatively simple intervention can produce outcomes that benefit everyone in the system,” Sahin says.


Discover Related Content