Artificial Intelligence and Machine Learning

A Typology of Complaints About Ebay Sellers

More could be done to reduce rising online fraud rates.
  1. Introduction
  2. Conclusion
  3. References
  4. Authors
  5. Footnotes
  6. Figures
  7. Tables

Complaints are expressions of dissatisfaction stemming from a feeling of having been wronged. Complaints can express dissatisfaction with a company’s customer service or allege that a company has defrauded a customer. Understanding complaints allows firms to improve their business practices to better meet customer needs. Online complaints are often recorded in reputation systems that collect, distribute, and aggregate feedback about an online business’s past behavior [9]. These online complaints help consumers engaging in transactions over the Internet to “decide whom to trust, encourage trustworthy behavior, and deter participation by those who are unskilled or dishonest” [9].

In 2007, an estimated $108.7 billion in goods were sold online with approximately a quarter of sales occurring at online auctions [2, 3, 10]. Online auctions provide unparalleled selection and potential value for buyers, while offering sellers a way to reach millions of buyers. However, the anonymity of online auctions gives less scrupulous sellers the opportunity to take advantage of buyers, either by intentionally misstating the quality and condition of their products or by selling products they have no intention of delivering. Both of these practices are forms of online fraud, and represent a growing problem for online consumers. Online auctions are especially prone to fraud because it is easy for businesses large and small to establish and run online auction stores. In 2006, online auction fraud was the most reported online offense, comprising 44.9% of complaints referred to the Internet Crime Complaint Center [5].

Online auction reputation systems contain feedback profiles made up of comments from transaction partners (buyers or sellers). In order to understand the complaints being left in online auction reputation systems, it is necessary to determine what problems are currently being reported as negative feedback in reputation systems, and the proportion of problems that contain allegations of fraud. This research develops a typology of complaints about sellers in online auctions that can be used to classify negative feedback posted at online auction sites. Complaints placed in eBay’s reputation system in 2003 and 2005 were analyzed using the typology of complaints. The types of fraud a buyer can encounter when making a purchase at an online auction include [6]:

  • Non-delivery. The seller places an item up for bid with no intention of delivering it.
  • Misrepresentation. The seller deceives the buyer as to the true value of an item.
  • Black-market goods. Illegal goods sold on online auction sites.
  • Fee stacking. The seller adds hidden charges to the item after the auction is over.
  • Triangulation. Stolen credit is used to buy from an online merchant and the item is resold at auction.
  • Shill bidding. Intentional fake bidding by sellers to drive up the price of their items.

In addition to fraud, online auction buyers can experience problems that are undesirable but legal (such as difficulty contacting the seller). To analyze the characteristics of negative comments placed at online auction sites, we developed a complaint categorization scheme that describes the problems buyers can have when purchasing from an online auction. It includes problems related to the payment and shipping of items, problems with the item received, along with refunds or exchanges, communication problems, and bidding. The typology of complaints is shown in Figure 1. It should be noted that the complaint classification system is designed to classify allegations of misconduct that have not been proven. It is possible that some allegations are unfounded; nevertheless, the allegations are useful for understanding overall patterns of misconduct and estimating whether those patterns change over time.

Data for this study was gathered from eBay’s reputation system. eBay allows users to post positive, neutral, or negative comments about their transactions. This study classified negative feedback posted about sellers and estimated the frequency of the different complaint types. Initial samples of 6,571 negative comments from May 2003 were used to validate the complaint typology, and to provide estimates of the complaint rates for each of the complaint categories. An additional 867 negative comments from July 2005 were used to determine if the frequency and distribution of complaints changed significantly over the intervening two years. The complaints examined were extracted from a sample of over one million eBay comments (across both studies), providing an overall complaint rate for this study was 0.73 complaints for every 100 comments made.

The eBay data was analyzed using content analysis—a manual process in which every complaint was read by at least one of the researchers and categorized into one or more of the complaint categories shown in Figure 1. The content analysis found negative feedback records often referred to more than one seller problem (such as both a communication problem and non-delivery). In these cases, the complaint was placed into more than one complaint category. There were 11,371 different negative comments made in the 7,438 complaints examined. The complaints were grouped into five broad categories: about receiving items, about the item received, about returning items, about communication, and complaints about bidding.

Complaints about receiving items. Problems related to receiving the products are summarized in Table 1. The principal complaint was that the item was paid for but never received. This accounted for 36.52% of all complaints made. Although this could have been due either to seller or shipper error, the buyers largely assumed that sellers had never shipped the item. Complaints that the item was paid for but never received usually represent an accusation of non-delivery fraud. In a very small fraction of the complaints, buyers complained of non-delivery without allowing sufficient time for the seller to ship the item, as in this complaint: “It has been over a week. Where is the jacket?” However, in these cases the complaint was coded as slow shipping as opposed to non-delivery. The average amount of time buyers waited before placing a complaint accusing a seller of non-delivery was 36 days and 74.5% of buyers waited at least 21 days before complaining about non-delivery. A comparison of complaint rates for 2003 with those of 2005 showed that a higher proportion of buyers made allegations of non-delivery fraud at the later date. However, the overall rate of complaints about receiving items remained relatively constant.

Complaints that the auction was won but the transaction was not completed because the buyer never heard from the seller, the seller did not follow through with the sale because the price was too low, or the seller no longer had the product to sell were found in 9.01% of all complaints. These are not fraud because the buyer did not lose any money. However, it does violate eBay’s policy on non-selling sellers, which states: “It is not permitted for a seller on eBay to refuse payment or delivery of an item at the end of a successful sale.”

There were a variety of problems reported related to the shipment of the item, with 6.7% of buyers complaining that their products were shipped late or to the wrong address. There were also complaints that sellers overcharged for shipping, with buyers charging that the seller used excessive shipping charges to make extra money on the sale. In general, shipping problems do not constitute fraud, nor do they violate any stated eBay policies. If the complaint indicated that the excessive shipping charge was different than the one listed in the auction it was categorized as fee stacking, which is fraudulent, not as shipping overcharge.

General complaints about items. In the sample of 7,438 complaints were several reported problems related to the products themselves (see Table 2). These include cases where the item received was undamaged but the buyer was dissatisfied, the item received was damaged, and the item received appeared to be counterfeit or stolen. In 26.27% of complaints, a working product was received but the complaint indicated that the item had been misrepresented either in the item title, description, or photo. Most often, the buyer complained that the seller had omitted important product details from the description. Buyers also complained that the item received was not the item they purchased. Usually, the incorrect item received was of lower quality or had fewer features than the item that the buyer actually purchased. There were also complaints that the item was of lower quality than the buyer expected. These complaints included words like “junk” or “trash.” Buyers also complained that they did not receive everything they had purchased or that that the item had an unexpected color, texture, size, or smell. Some of these fraud allegations could have been the result of buyer error, but the majority are likely cases of misrepresentation fraud in which the seller intentionally (or unintentionally) misrepresented the product to receive a higher price.

Buyers reported receiving damaged or defective goods in 7.76% of the complaints filed. The cases where buyers reported the damage was present before shipping were recorded as a problem with the item (instead of a problem that occurred during shipping). These would also be cases of misrepresentation fraud. In all, 30.3% of the complaints alleged there had been some type of misrepresentation fraud committed.

This complaint category also included complaints from buyers who believed their products to be forged, copied, or stolen. A total of 146 complaints (2.39% of the total) related to illegal products, or black market fraud. None of the complaints about illegal products indicated that the products had been purchased with stolen credit; this would be a different type of fraud: triangulation.

Complaints about returns or refunds. Table 3a summarizes the complaints related to returning the product or receiving a refund. These complaints were often coupled with some other type of complaint. For example, the following complaint indicates that the seller appears to be making income from fraudulently misrepresenting products and then charging customers to correct his error: “Sent wrong speakers he wants 2 charge me 25% restocking 4 his mistake! Beware!!” Other complaints related to refunds and exchanges include sellers accepting a product return but never sending the refund, sellers who refuse refunds altogether, and cases where refunds were only obtained after eBay, PayPal or law enforcement were involved. Most of the refund-related complaints began with the fraudulent non-delivery of goods or with goods that were misrepresented (732 of 796 refund complaints).

Complaints about communication. Table 3b summarizes the complaints that included some reference to a communication difficulty. Often the buyer’s first sign that he or she would be a victim of non-delivery fraud was a difficulty communicating with the seller. In 1,287 of the 2,716 reported instances of non-delivery, the buyer also reported difficulties in contacting the seller either via email or phone. In 404 cases, communication difficulty (either nonresponse to email or rudeness) was the only problem reported. However, more often communication difficulties co-occurred with other types of reported problems. One type of communication problem indicates a type of fraud, known as fee stacking fraud. This occurs when the seller changes the price of the item, shipping cost, or payment methods accepted after the close of the auction. Fee stacking was reported in 1.26% of online auction complaints.

Complaints about bidding. No problems related to bidding were reported in any of the complaints from 2003. However, two complaints about shill bidding were found in the 867 complaints from 2005. There were also five complaints about bid cancellation. In these cases the sellers cancelled the winning bidders bid because the price was either too low or because they made a side deal to sell to a different seller.

Back to Top


Results of this study indicate that more than 97% of complaints allege serious problems with the seller. Comments often indicate that sellers lack business training and clear commerce standards, like proper communication skills (44.2%) and appropriate return policies (10.5%). This suggests that legitimate online auction sellers interested in establishing a “good reputation” should maintain good communication throughout the auction process and be willing to accept returns—especially in the case of seller or shipper error.

However, a greater proportion of the complaints contain allegations of fraud. This study shows that 69.7% of negative comments posted in eBay’s feedback forum indicate the seller may have defrauded the buyer by failing to deliver the item, misrepresenting the item in the product description, selling illegal goods, adding charges after the close of the auction, or by shill bidding. The proportion of complaints that allege fraud increased slightly between 2003 and 2005 (from 69.4% to 71.8%); however, the increase was not statistically significant (p=0.158). The fraud allegation rates (as a percentage of complaints made) for each online auction fraud type are summarized in Figure 2. This data indicates that both non-delivery fraud (found in 36.5% of complaints) and misrepresentation fraud (found in 30.3% of complaints) comprise the vast majority of frauds reported. Black-market goods (stolen, counterfeit, or pirated goods) were only reported in 2.4% of the negative comments analyzed. This may mean that buyers were unaware that the goods were illegal or that they did not care that the goods were illegal as long as they got a good price. Fee stacking was only reported in 1.2% of complaints. This type of fraud may be less prevalent because it only generates a small return (the extra fee sellers tack on after the auction close). Shill bidding was only reported in 0.03% of the auction complaints. Shill bidding is difficult for the average buyer to detect, and thus, actual instances of shill bidding may be much higher than those reported as negative feedback. Finally, triangulation was not reported in any of the complaints examined. In general, buyers would only know they were a victim of triangulation if contacted by the merchant victim or law enforcement; something that would happen long after feedback was provided to the seller. As with black market goods and shill bidding, it is likely that triangulation fraud is underreported in the eBay reputation system.

The primary contribution of this research is that it demonstrates that reputation systems contain information about online auction fraud that is not found anywhere else. eBay has consistently maintained that less than 0.01% of its auctions are fraudulent [1, 7]. However, this study suggests that the problem of online auction fraud may be more severe than the number of officially reported cases would indicate. Prior research shows that between 41.8% and 52.1% of all successful auctions receive feedback [4, 8]. The rate of negative feedback (as a percentage of all feedback left) found during this study was 0.73%, and 69.7% of negative comments alleged fraud. Thus, the rate of fraud accusations (as a percentage of completed auctions) made in the eBay reputation system was closer to 0.2%, 20 times higher than the rate reported through official channels.

This research highlights the prevalence of various types of fraud at online auction sites and suggests more could be done to reduce current fraud rates. For example, since many of the fraud complaints relate to non-delivery of goods, online auction sites could more actively promote the use of escrow services, which reduce the ability of sellers to accept payment without delivering goods. This research also shows that reputation systems contain important information related to fraudulent activities, and thus improvements to these systems could make it easier for buyers to detect and avoid fraudulent sellers. For example, mixing negative comments with the large number of positive comments may make it more difficult for buyers to find comments about illegal behavior. Redesigning reputation systems so that recent negative feedback is highlighted could potentially improve a buyer’s ability to assess fraud likelihood.

One benefit of this study is that it presents a framework for classifying complaints and information about the rate of fraud occurring on the eBay online auction site during 2003 and 2005. This provides a baseline for future research on online auction fraud and will allow researchers to assess the effectiveness of fraud reduction measures.

One limitation of this study is that it focused exclusively on eBay, which was selected for this study because it owns nearly 75% of the global online auction market share. For this reason fraud rates for eBay dominate any determination of auction fraud rates in general and it is common research practice to use eBay data exclusively when studying online auction markets. However, each online auction site has different security precautions, which could change fraud rates for these sites. Doing an in-depth comparison of complaint rates across multiple online auction sites was beyond the scope of this study, but should be included in future research on online auction fraud.

This research shows that reputation systems serve an important function in today’s online world. They can allow buyers to assess the trustworthiness of unknown online auction sellers and can be used by sellers to improve their customer service. However, these systems play another important role. They contain information about potentially illegal activities. Since the rate of fraud reported in these systems is 20 times higher than the rate quoted by eBay, it is likely that instances of online auction fraud are often reported only in these systems. This makes these systems important to both online auction houses and to law enforcement as they try to combat rising levels of online auction fraud.

Back to Top

Back to Top

Back to Top

Back to Top


F1 Figure 1. Typology of online auction complaints.

F2 Figure 2. Online auction fraud reported as percentage of eBay complaints.

Back to Top


T1 Table 1. Complaints about receiving items.

T2 Table 2. General complaints about items.

T3 Table 3. (a) Complaints about returns or refunds, (b) Communication complaints.

Back to top

    1. Cox, B. And the online fraud goes on..., (Feb. 14, 2003); 1584531_2.

    2. eBay. eBay Outlines Global Business Strategy at 2005 Analyst Conference. eBay press release (Feb. 10, 2005);

    3. Gray, T. Online shopping boomed this holiday season., (Jan. 3, 2005); 3453931.

    4. Gregg, D.G. and Scott, J. The role of reputation systems in reducing online auction fraud. International Journal of Electronic Commerce 10, 3 (2006), 97–122.

    5. IC3 2004 Internet Fraud Crime Report, Jan. 1, 2006-Dec. 31, 2006. National White Collar Crime Center and the Federal Bureau of Investigation (2007);

    6. IFCC, Internet Auction Fraud. National White Collar Crime Center and the Federal Bureau of Investigation (2001);

    7. Konrad, R. EBay losing allure to some entrepreneurs. USA Today Online (June 27, 2005);

    8. Resnick, P. and Zeckhauser, R. Trust among strangers in internet transactions: empirical analysis of eBay's reputation system. The Economics of the Internet and E-Commerce, M.R. Baye, Ed. Advances in Applied Microeconomics, Vol. 11. JAI Press, Amsterdam, 2002.

    9. Resnick, P., Zeckhauser, R., Friedman, E., and Kuwabara, K. Reputation systems. Commun. ACM 43, 12 (Dec. 2000), 45–48.

    10. U.S. Census Bureau. Quarterly Retail E-Commerce Sales 4th Quarter 2006;


Join the Discussion (0)

Become a Member or Sign In to Post a Comment

The Latest from CACM

Shape the Future of Computing

ACM encourages its members to take a direct hand in shaping the future of the association. There are more ways than ever to get involved.

Get Involved

Communications of the ACM (CACM) is now a fully Open Access publication.

By opening CACM to the world, we hope to increase engagement among the broader computer science community and encourage non-members to discover the rich resources ACM has to offer.

Learn More