Research by Siva Viswanathan and Goudong “Gordon” Gao
A buyer venting through a recent eBay forum about a shipping delay contemplated posting an online review criticizing the seller for “laziness and lack of professionalism.” One thing holding the buyer back was fear of a retaliatory review from the merchant that would “tank my rating” as a future seller.
Repliers in the forum cited eBay’s recent ban on such seller retaliation. They advised the buyer to claim more power in the situation but urged restraint — distinguishing between “mistake” and “fraud” in the transaction. Such a negative review, one responder said, could destroy that merchant’s business.
The discussion illustrates the premise of recent Smith research into manipulative behaviors in online buyer-seller reputation systems, where a high reputation score means more sales and ability to raise prices.
Such a system fuels an online marketplace, but participants “gaming the system” can make it sputter. In 2008 eBay navigated that storm when it acted against “strategic sellers” who were threatening buyers with retaliatory feedback to coerce positive transaction reviews.
Smith Information Systems Professors Siva Viswanathan and Goudong “Gordon” Gao measured buyer-seller reaction to the intervention through eBay’s seller-to-buyer power shift. Smith PhD student Shun Ye, a faculty member at George Mason University's School of Management, also contributed to the study, initially titled “The Good, the Bad or the Ugly” to reference three distinct seller behaviors.
Unlike the good sellers, strategic merchants had manipulated the review system. The researchers set out to determine whether these players were “bad” (inherently low-quality sellers) or “ugly” (opportunistic sellers exploiting the loophole in the old system and then reforming their behavior after the ban on seller retaliation). “The strategic sellers were indeed ugly,” Viswanathan says.
Before eBay intervened, buyers and sellers were ranked by the same feedback rating system. Counterparts in a transaction could revoke or mutually withdraw negative feedback and ratings. They relied on one another to avoid becoming red-flagged in the eBay community because of a negative, or even neutral, review.
The merchants who gamed that two-way system threatened its integrity. “The tactic enabled low-quality sellers who, for example, cut corners in shipping fragile items to manipulate their reputations and masquerade as high-quality sellers,” Viswanathan says.
Purchasers not playing along were paying a price, as evidenced by a four-fold increase in unwarranted negative feedback left for them in a retaliatory way, an eBay spokesman reported in a 2008 interview for Fortune Magazine.
Merchant outrage erupted upon eBay’s subsequent ban on negative and neutral reviews by sellers. Company forums generated discussion supporting a user stoppage. Sellers argued that eBay had stripped their protection from deadbeat bidders or unreasonable customer demands.
The company countered that retaliatory seller feedback, more than a bad transaction, would drive a buyer away from eBay.
Following a weeklong seller boycott and policy implementation, eBay reviewers became more emboldened. “Buyers were more likely to leave negative feedback with sellers stripped of retaliatory power,” Gao says. “More surprisingly, those strategic sellers who initially reacted strongly by opposing eBay’s policy change were found to exert more effort to improve their service quality after the ban on retaliation and revoking.”
The study is among the first to measure how an online reputation system’s design affects seller behavior. Some eBay merchants exploited a seemingly even, two-way rating system between buyers and sellers and compromised their marketplace.
Banning the mutual revoking of negative reviews and curbing sellers’ ability to retaliate changed their behavior for the better, but at the expense of unduly shifting the balance of power to the consumer, Gao says.
Regardless, Gao says the policy appears to be the most effective option for eBay and other online markets. “There is no easy and straightforward way to calculate ‘revoked feedback’ data in a reputation rating system,” he says.
“Strategic Behavior in Online Reputation Systems: Evidence from Revoking on eBay” is forthcoming in MIS Quarterly.