Bursting the Bubble of Assumptions about 'Sponsored Search'

In theory, it seems like a simple proposition for any advertiser who wants to reach target audiences by placing ads on pages that provide search-engine results: Bid high enough to win the top spot in these "sponsored search" listings, and you will attract more customers -- and generate higher profits. No wonder more and more advertisers are paying ever-increasing prices to rank higher in the listings, which appear alongside the results to queries entered in a search engine. According to eMarketer.com, 40% of the projected $34 billion that will be spent on Internet advertising in 2014 will consist of "sponsored search."

But do higher-ranked sponsored search listings really pay off in terms of a company's bottom line? When Wharton operations and information management professor Kartik Hosanagar, University of Texas professor Ashish Agarwal and Carnegie Mellon professor Michael D. Smith tested the common assumptions about online ad auctions, they discovered that the reality was quite different from the hype. "We found that the ads that are in the top position get disproportionately higher clicks, but that they are not necessarily maximizing revenues" for those advertisers, notes Hosanagar. In their study, the conversion rate -- the percentage of clicks that led to purchases -- and revenue were often higher for ads when they were in the second, third or fourth position than when they were at the top of the list. Their study, "Location, Location, Location: An Analysis of Profitability of Position in Online Advertising Markets," was published in the December 2011 issue of the Journal of Marketing Research.

The researchers discovered two major reasons behind that surprising pattern, according to Hosanagar. For one thing, ads that win the top position in search engine bidding auctions often attract consumers who are simply browsing for information about a particular product, with no real intentions to buy. Second, many serious shoppers will click on more than just the ad positioned first before proceeding with a purchase. And rather than return to the link positioned at the top, the consumer will buy from a lower-ranked site. The authors attribute this to a "recency bias" among consumers. For online shoppers, "the cost of evaluating an extra option is very little," notes Hosanagar. Consumers will often evaluate product offerings from more than one advertiser and "purchase from the most recently evaluated advertiser if all evaluated options appear reasonable," adds Smith. What most shoppers don't realize, however, is the ad positioned at the top of the page may have cost the company significantly more than what firms paid to be positioned second or third.

In conducting their research, Hosanagar, Agarwal and Smith worked with two different advertisers over a period of six weeks. During each week of the study, the advertisers submitted random bids of various values for over 100 keywords, constantly changing the value of their bids so that their ad copy showed up at different positions on the sponsored Google search listings. The researchers observed the number of clicks by page visitors, and the number of purchases ("conversions") by those who "click through" to the advertiser's page. They then measured how click-through rates, conversions and revenues changed in relation to ad position.

Beyond bursting the bubble of hype surrounding sponsored search auctions, their research "also shows that different kinds of people behave differently [online], depending on their intentions," Agarwal points out. Marketers and advertisers, he adds, need to tease out the differences between the ways a "window shopper" searches versus someone who is ready to buy. "Understand your target audience," he says. "Know what your actual buyers are doing; don't just go by the number of clicks." Hosanagar adds that serious buyers are more likely to click through to a website for more information, or to make a purchase, even if a company's ad is not positioned at the top of the search-engine listings.

What are the implications for search engine companies such as Google? The researchers note that most sponsored search is based on "pay-per-click" calculations: For example, if an advertiser bids $10 per click and there are ultimately 3,000 clicks on the company's ad, the firm winds up paying close to $30,000. If there is only half that number of clicks, the advertiser pays half that amount. "Google does not actively encourage 'pay for action'" contracts, says Hosanagar, in part because there is no proven technology for accurately measuring the impact of online ads on actual purchases.

Such a technology would not only have to measure how many consumers immediately made a purchase after clicking on a sponsored search ad, but also how many website visitors attracted by the ad went on to purchase that product days or weeks later -- either online or at a local bricks-and-mortar store. Despite the technological challenges involved with such an approach, "we are encouraging the entire search engine ecosystem and its advertisers to take pay-for-action more seriously," Hosanagar notes. "Technology for measuring pay-for-action is feasible in the long term, but search engines and advertisers will have to work together to create some recognized and transparent metrics for performance."

For search engines, the study could spell the end of justification for paying the high price for top positioning in sponsored search listings, Hosanagar notes. The research has other implications for advertisers, he adds, depending on what they hope to achieve by advertising on search engines. Companies that are employing sponsored search as a way to create greater brand recognition might be willing to pay for top positioning, he says. On the other hand, advertisers that are focused on generating "transactional revenues" -- attracting people who will click through an ad and buy something -- "should be cautious about bidding for the top position." Hosanagar advises companies to "be aware of how much you are paying and what you are getting for it. If you are a transactional advertiser, be careful to ensure that you are not over-paying. If you are not getting enough revenue, you should be lowering your bids."