Feeds:
Posts
Comments

There are many terms for Keyword Advertising.

Below is the list which all refer to the same meaning ‘keyword advertising’

  • paid placement
  • paid search
  • sponsored search
  • sponsored links
  • search advertisement
  • search marketing
  • keyword auction
  • keyword advertising
  • keyword marketing
  • PPC (Pay per click)

Research Background
As the Internet grows rapidly, a sustainable long-term business model is a key issue for entrepreneurs after the Internet bubble in 2001. It has been noticed that Internet advertising is a proven business model for those websites with sufficient traffic and audience. Portal websites, like Google, Yahoo!, MSN and AOL all take advertising fee as main income to keep them survive. According to annual report of Yahoo! 2005, it has been announced that there are more than 80% income came from marketing services since 2003, and the rest came from charging fees, including Internet broadband services, premium mail, music, as well as personals offerings.

On the other hand, advertisers also take Internet as important media to reach the masses, and it resulted in increasing budgets on Internet marketing year by year. Internet is not only new media for advertising, but also an effective and traceable marketing tool as well. In the past, advertisers could raise awareness and increase impressions via traditional marketing tools, but they only could get the information of probable segments of who will receive communication message depending on the exposure marketing channels. However, Internet advertising can provide information of who will receive and who will react to your marketing message exactly. In practice, Internet advertisers can tracks both impressions and click-though which respectively represent how many people are under ads broadcast and how many people are really attracted, and they are the issues advertisers care most.

There are many different forms of online advertising, including CPM (cost-per-thousand-impression) which counts on how many time the ads display, CPC (cost-per click) which used in search engine and paid by the times user click on, CPA (cost-per-action) which advertise lead users to take some action, such as purchasing a product or signing up, and COP (cost-per-order) is charged once order is purchased (Matthew et al., 2007). Nowadays, CPM and CPC are two primary models in Internet advertising industry, while two offer different merits.

CPM evolves from images to animations and videos, its main function is to drive traffic and catch audience’s attention, and therefore it was offered by giant websites with high expense. On the other hand, CPC offers a precise way to target certain segments when the ads showing up on search result pages. Besides, compared to CPM, CPC is a more affordable marketing tool for Small and Medium Enterprise due to it charges few dollars, depending on its bid in auction, whenever one clicks on the ads. Target marketing and affordable cost are the two reasons why CPC booms up in recent years. Take Google for example, its CPC product, AdWords, earned $1.63 billion in revenue for the third quarter of 2006 (Baker, 2006). It is estimated that that online advertising will exceed $55 billion globally by 2010 (Jansen, 2007). With its prevalence and potential, this research will exclusively focus on CPC model to reveal its current situation and discuss issues that advertisers care about.

The economic impact of CPC is immense, and it also plays an important role as a revenue generator for portal websites. Google received 99% of its $3.1-billion revenue from sponsored search in 2004 and Yahoo! received 84% of its $3 billion, according to Tim McCarty of Time magazine (Jansen, 2007). From an executive’s point of view, it is necessary to learn how to use this new marketing tool correctly and effectively with limited budgets.

However, CPC operation is a complicated and time-consuming routine work with enormous data processing, rather than just randomly picking keywords. According to the interview with a keyword agency, generally, advertisers have to selected hundreds of, or even thousands of keywords out of numberless vocabulary database once a time. Besides, they have to monitor the performance of each keyword, check daily report and adjust the settings immediately if necessary. Hence, in this market, advertisers bid lots numbers of keywords, and each keyword may be bidden by other competitors. As the result, the situation of keyword selecting by different companies will be the first step when researching this topic, how to make keyword selection strategy and its performance tracking is also worth to analysis.

This research tries to reveal the keyword bidding market situation at first, then, a keyword network will be formed via bidding behaviors and the relationship between the distribution in networks, search volume and performance indicator will be examined. A clear picture of paid search market and the keyword selection rules will be provides in this empirical study.

For advertisers, there are several issues when advertisers pick up their keyword sets for paid search.

    *How to decide a value of a keyword?
    * Is general keyword with huge impressions or specific keyword with niche target audience more effective? When to use * general keywords and when to use specific keywords? Is there a portfolio for keyword selecting?
    * Keywords with high impression will cause high price in auction?
    * Keywords with high impression can bring high Click-through rate?
    * If advertiser bid for keywords with high price, does that mean it can bring high Click-through rate?
    * Can we use the indicators of social network to explain the outcome of Click-through rate?

Even though paid search is still immature area with difficulties of data collection, it is very meaningful that this research can be a pioneer to kick-start the inside of paid search market.

The research objectives are as follows:

    * Provide an empirical competitive situation analysis of keyword advertising in 6 industries.
    * Clarify paradoxes that keywords with impression are effective keywords and provide correct keyword selecting strategies for advertisers.

This research procedure will go through the following seven steps, and the figure shows below.
Figure1-1

Introduction
This chapter describes the research background, motivation and objectives of this research.

Literature Review
Literature Review contains two parts. One is the introduction of basic but important variables of paid search and its previous related research. The other part is the theoretical review of social network.

Research methodology
First of all, Keyword networks will be defined, including nodes and ties. Data collection and processing method will also be explained in this chapter, following by establishment of research hypotheses.

Data Analysis
Research validation will be examined in the beginning of this chapter. Analysis result will be addressed and the characteristics of results will also be further discussed.

Conclusion and future work
Based on the results, not only conclusion of keyword analysis via social network methodology will be presented, but the suggestion to future work will be raised in the end of this research.

Click-through rate (CTR) is the most important measure in Internet advertising, and it represent for the percentage of users click on a giving list which can be formulated as Clicks/Impressoions . For example, one ad is delivered 100 times, and one user clicks it, the CTR will be 1%.

Due to the important role CTR plays in paid search, many researchers started to research what users’ behavior when they evaluate Internet contents, and also they studied the influencing factors and forecast model of CTR. Tombros, Ruthven, and Jose (2005) reported that there were 5 categories (text, structure, quality, non-textual items, and physical properties) used by the searchers to determine the utility of Web documents. Matthew et al. (2007) considered term, ad, order, and external features as factors to CTR, and used these variables in predicting CTR for new ads. In their model, the CTR has been improved 42% under over 1000 views. Furthermore, In 2008, Anindya et al. (2008) found in their experiments that the presence of retailer specific information in the keyword increases click-through rates, and the presence of brand-specific information in the keyword increases conversion rate .

Although there many related researches, most of them develop the algorithms in experimental environment and ignoring the complicated scenario in reality. In this research, we will depict in current market and analysis that outcomes of its bidding cost and exposures and CTR performance.

Yahoo! and Google have traditionally employed different mechanisms to determine the placement of the advertisement (Animesh et al., 2007), and Yahoo! adopted rank by bid” (RBB) mechanism and Google adopts rank by revenue” (RBR) (Lahaie, 2006). It has been found that ranking paid placement links by the product of willingness to pay and relevance is better, in most cases, than ranking by willingness to pay alone, which performs best only when the correlation between the provider’s relevance and willingness to pay is large (Juan et al, 2006). Lahaie (2006) proposed an equilibrium analysis revealed that RBB has much weaker informational requirements than RBR, because bidders need not know any information about relevance to play the Bayes-Nash equilibrium. However, RBR leads to an efficient allocation in equilibrium, whereas RBB does not.

According the latest release of Yahoo’s new search marketing platform in 2007, the ranking differs because they start to consider both of the two factors. One is how much advertisers’ willing to pay and the other is quality of advertisements which consists of predicted and previous performances. It allows more relevant advertisements could have upper exposure opportunities even with lower bid.

Although it seems this market is blooming, ranking providers like Google and Yahoo even offer substantial signup bonuses to new advertisers (Abrams, 2006), the influence of increasing bidders in this market is studies by Edelman, when a new advertiser arrives, a search engine’s revenue increases, but the author suggested that adding another bidder is not always preferable to setting an optimal reserve, in multi-unit context. Subsequently, it is easy to realize that too competitive bidding market will not result in a better income bring-in for ranking providers; on the other hand, advertisers also can not be benefited whenever the bid is spurred up to irrational price. The best way to prevent this vicious price competition is to spread constrained budget on various different keywords. Unfortunately, this topic has never been discussed in previous researches.

Keyword selections are the most critical process when launching advertisements on CPC platform, because it affects the total times of impression and the probability of clicking on. A correct keyword selection can increase ads exposure to target audience and estimate Click-through rate and conversion rate even under limited budget. In general, keyword agency suggests selecting hundreds of keyword a time to when launching ads.

As the result, many researchers started to study on keyword recommendation algorithms and develop keyword selection systems. Dwihananto et al (2007) recommended a new system that learns how to extract keywords from Web pages for keyword advertising, but their system only consider search volume of keywords as their main criteria of selection. Vibhanshu and Kartik (2007) used a web based kernel function to establish semantic similarity between terms and develop a method for generation of several terms from a seed keyword. Yifan and Gui-Rong (2008) proposed a novel keyword suggestion method that fully exploits the semantic knowledge among concept hierarchy. However, these two researches only use algorithms to generate a bundle of keywords without evaluating the performance of each word.

Rusmevichientong and Williamson developed an algorithm that adaptively identifies the set of keywords to bid based on historical performance. The algorithm prioritizes keywords based on a prefix ordering, sorting of keywords in a descending order of profit-to-cost ratio (Paat et al., 2007). This algorithm can increase profits by about 7%. Their formulation assumed that the bid price of each keyword remains constant and the ad will always appear in the same spot on search result page. In other words, they did not put cost into consideration, even cost is one of the main issues advertisers care about. Therefore, we will provide a cost and performance perspective to research this topic.

Follow

Get every new post delivered to your Inbox.