UNC Kenan-Flagler Business School


The search is on


money in slotCompetitive battles are being waged in the $200 billion global digital marketing industry – and paid-search advertising has become a critical tactic used to win customers.

UNC Kenan-Flagler Business School marketing professor Shijie Lu studies competitive factors in search advertising. His findings suggest strategies that could help both advertisers and search-marketing providers improve revenue and profits.

Lu and his colleagues used datasets from keyword campaigns to test the behaviors of advertisers and consumers using mathematical models. “Data from an individual e-tailer was typically used in previous studies,” Lu says. Using data that encompassed all advertisers that used a given set of keywords allowed him to ask different questions.

Lu studied the effects of competition on paid-search advertising with Sha Yang of the University of Southern California and Xianghua Lu of Fudan University using a dataset with information about 1,573 search terms related to digital camera/video products and accessories. He also studied the spillover effect of keyword market entry in sponsored search advertising – or, in other words, how the likelihood of using a keyword is affected by competitors’ entry decisions – with Yang.

In both studies, the researchers considered each keyword to be an individual market in which advertisers chose whether to compete. For the competition study, they drew on data from an online marketing platform outside the U.S. that hosts keyword-based advertising. Marketers bid to have their ads appear alongside search results for a designated keyword. Ads higher on a page typically draw more clicks than ads lower on a website page.

Lu found that as more advertisers bid for a particular search term:

  • Individual advertisers’ revenues declined.
  • Their click volume declined.
  • Their cost-per-click (CPC) increased.

Advertisers who stayed at the top of a page, however, tended to benefit from competition because of the increase in click volume. Advertisers who remained at the bottom saw their click volume increase at first, then decrease as the number of competitors bidding for the same search term grew. Lu’s findings suggest that the impact of competition on an advertiser’s profit depends on the advertiser’s expected position in ad listings.

In the spillover effect study, the researchers used 1,252 popular keywords related to laptops and accessories on Google from September 2011 to April 2012. A leading U.S. search-advertising information provider – known as an infomediary – provided data about firms entering those keyword markets and ad rankings.

The Google data revealed that the advertiser’s position in the distribution channel influenced its success. Lu distinguishes between three types of companies that run search ads and compete for users’ limited attention on search engines’ result pages: manufacturers such as Lenovo, retailers like Best Buy and comparison sites like NexTag.

He discovered that:

  • Manufacturers tended to use keywords with more retailers and other manufacturers – and fewer with comparison sites.
  • Retailers tended to use keywords with more comparison sites and other retailers.
  • Comparison sites tended to use keywords with fewer advertisers from each of the three types.

These findings suggest that the presence of direct competitors – as well as advertisers from other parts of the distribution channel – should be a factor when considering whether to enter a particular keyword market.

Lu also found that manufacturers used a wider variety of keywords, possibly because they are more familiar with their products’ features. Retailers and comparison sites, he says, should consider reviewing keywords that have been chosen by manufacturers to increase their own advertising options.

Advertising platform strategies
The research also has implications for search advertising platforms, where revenues increase as the number of competitors for a keyword increases.

There are two ways the platforms could attract more advertisers.

First, they can use design changes to reduce the decay factor – a measure of how much the click volume declines going down the page from one ad to the next. This would drive more clicks to ads farther down the page, making them more attractive to advertisers. While advertising providers would see the CPC decline on those pages, the total click volume and revenues would increase despite the lower CPC. However, search ad providers are still dependent on consumers changing their behavior in response to design changes for this strategy to work.

Offering coupons is another way to attract more advertisers. Lu calculates that offering a new advertiser a coupon with a face value of 20 percent of the average advertising cost would improve host profits by 25 percent. This approach – unlike a design change to influence consumer behavior – is entirely within the control of the search platform provider, says Lu.

Finally, the researchers considered whether infomediaries selling keyword advertising information were hurting or helping search-advertising providers. The model Lu and Yang developed found that making data available – including information about what other advertisers are doing – increased search engine revenue by 5.7 percent. This fits with major search engines’ practice of allowing third-party companies to access and sell their keyword advertising data.

Academics have just begun to explore digital marketing rigorously, Lu says. And while large companies like Google have extensive internal research capabilities, many smaller firms do not.

“The internet was only accessible to the general population in the middle of the ’90s, and mobile took off in the first decade of the 21st century,” Lu says. “When new technology comes out, a lot of marketers will embrace it – and as academics, we have a responsibility to undertake rigorous studies of whether it works.”