Robert S Owen. Electronic Commerce: Concepts, Methodologies, Tools, and Applications. Editor: S Ann Becker. Volume 3. Hershey, PA: Information Science Reference, 2008.

Introduction

This article discusses illegitimate activities associated with online promotion activities, specifically those related to click-through fraud. Whether one is starting a new brick-and-mortar restaurant, managing a charity organization, or building an online information portal, prospective customers, donors, or users must be informed, persuaded, and reminded of the features and benefits that are offered by the business. Online promotion has some unique and advantageous characteristics over traditional promotion media, but those characteristics create a potential for abuse. Click-through fraud, more commonly termed click fraud, is a relatively recent and evolving problem that currently has few deterrents. This article outlines why and how click fraud is done, and suggests some measures that can be taken to at least recognize the potential for click fraud.

The interactive nature of the Web would seem to make it an ideal advertising medium with a potential to completely eliminate advertising waste. Newspaper ads for, say, a pet store reach many people who do not have pets: the number of people in this category is termed as waste. With online advertising, it is possible—in an ideal world—to place an ad and to pay for the ad space only when a prospective customer shows interest by clicking on a linked advertisement that transports him/her to the advertiser’s online place of business. This pay-per-click (PPC) pricing model would seem to completely eliminate advertising waste.

Unfortunately, advertisers, ad hosts, and fraudsters are discovering that the PPC model is open to abuse. Those who click on an online advertisement could include:

• Prospective customers who actually have an interest in the product
• Competitors who want to generate high advertising costs for the advertiser
• Ad hosts who earn a commission from displaying pay-per-click advertising

The latter two categories consist of entities that have absolutely no interest in the advertiser’s offerings and absolutely no intention of ever performing any target actions such as purchasing a product. These fraudulent click-throughs or debiting clicks could turn an ad campaign from one that has almost no waste into one that has almost 100% waste. In a survey of advertisers, the Search Engine Marketing Professional Organization found that a quarter of respondents have tracked fraud as a problem (SEMPO, 2004).

Background

Contextual advertising is one way of increasing the likelihood that an ad is reaching people in its target audience. In Web-based advertising, contextual advertising would include banner ads or search engine links that are displayed on a page that has a context that is related to the ad; people on that page are already searching for or browsing through ad-related content. Better reaching the target audience in this way should result in less waste and, in the case of online advertising, in a higher click-through rate (cf. Newcomb, 2003; Smith, 2004; Sullivan, 2004).

Search engines engage in contextual advertising through the pay-per-click (PPC) or cost-per-click (CPC) pricing model. When someone does a search, many search engines not only return links to pages that have been indexed on the search terms, but also return sponsored links to pages that the advertiser has paid to have listed at high ranking (cf., Brendler, 2005). When someone clicks on a sponsored link in the PPC model, this click-through is a debiting click that subtracts from the advertiser’s prepaid advertising click budget. The price per click is set by a bidding process, whereby the advertiser who bids to pay the most for particular keywords will be ranked the highest in the list of sponsored links returned by the search engine (cf., Alchemist Media, n.d.). The Fathom Online Keyword Price Index shows that in March 2005, advertisers in the consumer retail industry were paying an average of $0.51 per click, while advertisers in the mortgage industry were paying an average of $5.39 per click (Fathom, 2005). Bids on some keywords can reach into the fifty-dollar range (cf., Associated Press, 2005a; Bruce Clay, n.d.).

Both advertisers and ad hosts have experienced fraud under the PPC advertising model, but legal action is relatively new at the time of this writing. In February 2005, the gift shop Lane’s Gifts and Collectables fled suit against Google, Yahoo, and others for allegedly billing for inflated click-throughs for PPC advertising, charging that many click-throughs were not generated by bone fide potential customers (Associated Press, 2005b). But just a few months earlier, Google had itself fled one of the first click fraud lawsuits against one of its advertising affiliates who displayed Google-generated sponsored links on its own Web site. The affiliate, Auctions Expert, contracted with Google to display PPC links, but then allegedly clicked on those links itself in order to generate a commission (Olsen, 2004). In March 2004, Michael Bradley was arrested by the U.S. Secret Service in association with his Google Clique software that he claimed could be used to generate false clicks for Google affiliates (Nariane, 2004).

Advertising Click Fraud

Before looking at why and how click fraud is done, we need an understanding of some basic cost or pricing methods used in online advertising.

Pricing Models

Traditional Model: CPM and Pay-Per-Impression Advertising

A traditional measure of advertising cost for print and broadcast media has been CPM, or “cost-per-thousand.” In traditional media, this is the cost of reaching 1,000 individuals or households with the advertising message in a given medium. With online advertising, CPM tends to be associated with the cost of 1,000 impressions, or actual exposures to a particular advertisement. In pay-per-impression advertising, the advertiser pays for some predetermined number of ad impressions.

When this impression budget is exhausted, the ad is removed from the display list (cf., Internet Retailer, 2005).

Performance Model: Cost-Per-Transaction or Cost-Per-Action (CPT, CPA)

Although not often discussed and apparently not often used in online promotion, pricing that is based on the ultimate visitor’s target action (e.g., making a purchase) could be done. The terms cost-per-action and cost-per-acquisition(CPA) are sometimes used to describe this idea (cf., Gold, 2005; Stevens, 2001; Think Metrics, n.d.). Mand (1998) discussed a model that has been called cost-per-sale, cost-per-trade, and cost-per-transaction advertising (CPT). This equates to a sales commission in traditional media.

Bid-for-Placement Model: Pay-Per-Click or Cost-Per-Click Advertising (PPC, CPC)

In pay-per-click (PPC) advertising, also known as cost-per-click (CPC) advertising, the advertiser pays a fee for each time someone clicks an advertisement that links to the advertiser’s Web site. The advertiser pays for a predetermined number of clicks at a price that is often set by being the highest bidder in an auction. The advertisement host (e.g., a search engine) removes the ad from the display list when this click budget is exhausted. The advertisement host might also remove the ad from the display if the clicks-to-impressions (CTI) ratio or click-through rate (CTR) falls below some predetermined limit (because it is failing to generate sufficient revenues.)

Click Fraud Classification

Fraud can exist in all three types of pricing models, but click fraud is probably the most problematic at this time because the PPC model has become so widely used in online advertising. Many types of online advertising fraud are motivated by two basic objectives. One is associated with an attempt to either hurt a competitor or to force the competitor to decrease its advertising. Another is to profit from hosting advertising messages. The first type could be classified as competitive fraud and the second as affiliate fraud or network fraud (cf., Claburn, 2005; Lee 2005; Stricchiola, 2004).

Competitive Fraud

The objective of competitive fraud is often to hurt a competitor or to get a competitor to quit advertising. For example, one competitor can repeatedly click on another competitor’s PPC ad with the intent to inflate the competitor’s advertising cost rather than to obtain information about the advertised product. The objective of deliberately imposing this cost on the competitor could be to:

• Drain the competitor’s advertising budget in an effort to decrease the competitor’s profitability.
• Cause the competitor to see less value in bidding up the price to be at the top of the search engine list. This thereby lowers the prices for other competitors to bid for top listing.
• Cause the competitor to quit advertising or to quit doing as much advertising because it either cannot budget for the volume of clicks or sees PPC advertising as providing a poor return on investment. If the competitor bows out of the top spot, it is now open for the remaining competitors at lower prices.

Click Monkey, for example, advertises that it can be hired to “Google bomb your competition” for pennies on an advertisement’s bid price as a cost-effective way to lower high-bid PPC rates (Click Monkey, n.d.).

An interesting example of apparent competitor fraud is the alleged actions both by and against Blue Star Jets. Charter Auction, a charter jet broker, at one time spent $20,000 per month on search ads and $20 per click for some search terms. Charter Auction and another charter-jet broker, Air Royale International, believed that they were the victims of click fraud and identified an IP address (Internet address) assigned to Blue Star Jets that had repeatedly clicked on ads thousands of times. But Blue Star Jets itself was hit with click-through theft by a competitor who presumably believed that it was justified in retaliating against Blue Star Jets. Tag Aviation found that competitor Blue Star Jets had purchased clicks based on the search phrase for its own name, “Tag Aviation,” thereby stealing customers who might have intended to go to Tag Aviation. Tag Aviation staff were told to retaliate by clicking on these Blue Star ads in order to deplete Blue Star’s click budget for these ad keywords (McKelvey, 2005).

Note that we do not necessarily have to click on the competitor’s ads to drive up its costs. We could also quit using particular keywords ourselves, causing the competitor to get all the hits (non-fraudulent, legitimate ones). If the competitor has a high bid price on these keywords, its advertising costs might increase if it is getting all of the hits. If it cannot temporarily afford this increase, it might be coerced into decreasing its advertising, leaving the market open for the rest of us (cf., Crowell, 2005).

It is also possible to hurt a competitor through a click restraint action rather than doing a click-through. When the clicks-to-impressions ratio falls below a certain threshold, the search engine or other advertising host might automatically drop the ad from the display list to be replaced by an ad that gets a higher click response (which therefore generates higher revenues). A click-through is a debiting click when it is used to consume a click from the PPC prepaid budget of clicks, (cf stricchiola, 2004). But in this case, the fraudster opens a Web page that causes an ad to be displayed, but then refrains from clicking on it; this decreases the click-to-impression ratio and therefore could cause the ad to be dropped or lowered on the list (cf., McKelvey, 2005).

Affiliate/Network Fraud

The objective in affiliate or network fraud is to obtain a financial commission on PPC advertising that is being hosted. Mentioned earlier, Auctions Expert allegedly recruited as many as 50 people to click on ads that it hosted (Vise, 2005). Vidyasagar (2004) suggests that a small industry is developing in India in which housewives can earn up to $0.25 per click to browse online advertising. With paid-to-read (PTR) programs, list subscribers are paid to click on links that are e-mailed to them (e.g., StopClickFraud, n.d.); the unscrupulous affiliate thereby lowers the risk of detection by generating clicks that appear to be legitimate in originating with real people at real computers in a wide geography. Click Monkey (n.d.) outright advertises its click farm services, advertising rates for as little as one dollar per thousand clicks. The higher rates, however, would have to be paid to those who would spend 60 or 90 seconds on an ad—long enough to give the appearance of a legitimately interested party. It is also possible to automate hits with spoofs of varying IP addresses (i.e., the software pretends to be making clicks from a variety of geographic locations instead of from a single computer), as when Michael Bradley claimed that his Google Clique software could be used to generate false clicks for Google affiliates (Nariane, 2004).

An advertising affiliate can also commit impression fraud and content fraud. In impression fraud, the advertisers pays for visitors who merely arrive at a page with an ad (an ad impression), not for a click-through of that ad. Impression fraud can be committed by the same methods used in click-through fraud (cf., Feldblum, 2005; Internet Retailer, 2005; Lee, 2005). In committing content fraud, the PPC advertising affiliate (host) indexes pages using keywords that are unrelated to the actual content of the page and the advertisement. In this way, the fraudster can increase the number of hits to that page (impressions), but the visitors who reach that page arrived with the expectation of different content; the impressions are wasted on an audience that has no interest and had expected something different (cf., Feldblum, 2005).

Click Fraud Detection Hints

An advertiser should normally look at measures and ratios in order to evaluate the cost effectiveness of an advertising campaign. Importantly, some of these could be useful in tracking the onset of fraud activities.

Click-Through Rate

The click-through rate (CTR) of an ad or sponsored link is the number of visitors who clicked on the link vs. the number of visitors who were exposed to the ad, or click-throughs vs. impressions, expressed as a percentage. A higher CTR could suggest that the ad or link is effectively placed in the context of the page that displays it. A sudden increase in CTR, however, could also suggest the onset of click fraud activities.

Bounce Rate

Some Web site visitors will enter and then back out (leave) without viewing any other pages on the site. The bounce rateor ratio is the number of visitors who leave without linking from anything on the page vs. the total number of visitors (orpageviews), expressed as a percentage (cf., IndexTools.com, n.d.; Think Metrics, n.d.). A sudden increase in bounce rate could suggest the onset of click fraud activities.

Customer Conversion Rate

Conversion has to do with visitors who become revenue-generating customers. The customer conversion rate (CCR) is the percentage of Web site visitors who engage in an ultimate target action, such as making a purchase (cf., Gold, 2005; Pelland, 2005; ThinkMetrics, n.d.). A sudden decrease in conversion rate, or an increase in non-converting traffic, could suggest the onset of click fraud activities.

Future Trends

If the value of online advertising decreases due to fraudulent activity, providers of online advertising services will ultimately have a financial motivation to take action to protect the interests of advertisers. As customers of these services, advertisers will gravitate to competing service providers who provide the best assurance that they are providing good value for the investment in advertising. How these services might best protect advertisers is beyond the scope of this article; this emerging issue currently does not appear to have any easy answers either in practice or in the courts.

Conclusion

While this article has not provided any answers regarding how advertising hosts can protect against fraud, it has hopefully helped to make advertisers aware of important emerging issues associated with online advertising. Although fraud is currently a cost of business that cannot be easily stopped, the advertiser can be wary of signs that suggest fraud and can take measures that compare different media or service providers to track the cost effectiveness of an advertising campaign. Advertising waste exists and is accepted as a cost factor in traditional advertising media; that fraud cannot easily be controlled in online advertising might ultimately (and unfortunately) force us to accept it as a mere nuisance factor of waste in contrasting the cost efficiency of various media.