Regional Analysis on Chinese Internet Ads with Click-Through Rates

Wei Li Qiang Wei Yaoqiang Chen School of Economics and Management, Tsinghua University, Beijing 100084, China Email: {liw.09, weiq, chenyq.03}@sem.tsinghua.edu.cn

Abstract Internet Ads market has continued high- speed growth of 50% roughly, occupying Internet Ads is one of the important forms the total advertising market share from of E-business activities, due to its high 3.8% in 2006 to over 10% in 2009 [2]. targeting, rich contents, flexible media as Nowadays, Internet Ads is going to show well as high interactivity. Since the econ- its unique advantages and expand the omy and Internet development of differ- market share significantly. ent regions in China are quite differen- Although Internet Ads market is ex- tiated, the Internet Ads of different re- panding rapidly, the effect of Internet Ads gions have significant regional characte- is not satisfactory, since Internet users ristics. Based on the real log data from a become to more and more immune from Chinese Internet Ads intermediate, this Internet Ads. Furthermore, users fre- paper empirically analyzes the differenc- quently complain the inconvenience es and characteristics of Internet Ads of brought by Internet Ads. One evidence is China’s regions with the Click-Through that the Click-Through Rates (CTRs) of Rates. Moreover, this paper proposes Internet Ads decline significantly, e.g., some suggestions for decision supporting from 30% ~ 50% on Banner Ads in mid- Chinese Internet Ads players, i.e., web- dle 1990s to less than 0.1% currently [3]. sites, intermediates and advertisers, from CTR is actually one of the key meas- a managerial perspective. ures to evaluate the performance of Inter- net Ads, which is impacted by many fac- Keywords: E-business, Internet Ads, tors, e.g., visibility of Ads, attitudes of Click-Through Rate, Regional Analysis users, interactive environment, etc [4,5,6]. Besides, region is also one of the impor- 1. Introduction tant factors which show some interesting relationship with CTRs, e.g., the Internet Internet Ads is one of the important forms behaviors of people in different regions of e-business activities, due to its high will show different patterns on brows- targeting, rich contents, flexible media as ing/clicking Ads, leading to different well as high interactivity, which is keep- CTRs. This paper is to empirically ana- ing pace with four major traditional me- lyze the difference, characteristics of In- dia ads, such as Television Ads, Radio ternet Ads of China’s different regions Ads, Newspaper Ads and Magazine Ads with CTRs data from some real log data [1]. According to industry reports, in of a Chinese Internet Ads intermediate. 2009, even with financial crisis, Chinese The paper is organized as follows. Sec- Internet Ads market is still booming, tion 1 will investigate the relationship be- reaching 20.73 billion RMBs, continuing tween regions and CTRs. The data de- rapid growth of 21.9% [2]. Since 2003, scription and methodology used in this

390 © Atlantis Press, 2010 The 2010 International Conference on E-Business Intelligence paper will be presented in Section 3. The ing the performance, e.g., CTRs, of Inter- research questions and hypotheses will be net Ads. proposed in Section 4 with some empiri- Besides, due to the imbalanced region- cal analysis. Section 5 will briefly eva- al development of Internet in China, re- luate the results and present some coun- gion factor plays a specifically important termeasures for Internet players from a role in China. Moreover, due to the lack managerial perspective. of real log data of Internet Ads, empirical studies could seldom be conducted. 2. Regions and Internet Ads’ CTRs 3. Data and Methodology Internet Ads is one kind of information dissemination activities for advertisers In this research, a real log data of Internet paying to persuade the public to buy Ads was provided by a professional In- products, aimed at influencing people’s ternet Ads intermediate in China, which attitude to products or services in the ad- provides a platform with typical Internet vertisement so that advertisers can make Ads services, such as posting, monitoring, profit. Therefore, the effects of Internet performance evaluation and optimization, Ads, e.g., CTRs, are closely paid atten- etc. There are more that 100 allied web- tion to by Internet Ads players, i.e., web- sites on the platform with thousands of sites, intermediates and advertisers. Internet Ads provided by hundreds of ad- The factors that influence CTRs can be vertisers. The Internet Ads on this plat- classified into three general categories: form almost cover all types of industries. the advertisement features [4], which are The users of the platform are from differ- the attributes and features of Internet Ads ent regions in China. The intermediate themselves, such as color, size, animation, charges advertisers with CPC. On this sound, scenes, interactive and virtual cha- platform, CTR is regarded as one of the racters, etc.; consumer-related factors [5], most important index for evaluating In- which are attitudes of users, such as per- ternet Ads. sonal knowledge structure, personal atti- Totally, we have collected 27 days’ da- tudes to Internet Ads; and situation- ta (March 30 – April 26, 2009) with over- related factors [6], which are Internet Ads all 500GB data including displaying data environments, such as the relevance be- and clicking data. MySQL was used for tween Internet Ads and web pages’ con- data cleaning and pre-processing. Since tents, the location of Internet Ads, etc. the data from Hong Kong, Macao, Tai- Nevertheless, among the researches on wan are quite limited, thus we focus on factors impacting CTRs of Internet Ads, the 31 regions/provinces in Chinese there is few talking about region factors. mainland. Chinese Silly Catalog concept proposed After pre-processing, the generated that regionalization is of great importance CTRs of each province are listed in Table to Internet services [7], especially in Chi- 1. na. Li and Wei indicated that different regions could create different and indi- Day Shanghai Beijing … Yunnan vidual distinctive marketing styles due to 1 3.00678 3.63746 … 7.07463 different regional cultures in China, 2 3.53536 3.33896 … 5.50929 which also took significant impacts on … …… …… … …… Internet Ads. Thereafter, regions should 26 4.15278 3.85921 … 6.85270 be taken into consideration while analyz- Table 1: Regional Average CTRs (1/104).

391 The 2010 International Conference on E-Business Intelligence

Since no evidence shows that the data following normal distribution, this paper The test results of 31 provincial regions adopts the non-parametric test methods, are shown in Table 2. The result show i.e., Friedman test method and Wilcoxon that the p-value is less than 0.05, indicat- test method. SPSS statistical software ing the hypothesis could be accepted, was used. meaning that at least two provincial re- gions have significant difference on 4. Hypotheses and Analysis CTRs.

First of all, it is necessary to test whether N Chi-square Df Asymp. Sig. the CTRs of provincial regions are signif- 26 564.010 30 .000 icantly different. Then the first hypothesis 1 could be proposed. Table 2: Friedman Test of 31 provinces.

Moreover, CTRs varies in different re- H1: Different provinces have signifi- cantly different CTRs. gions, as shown in Figure 1.

Fig. 1: CTRs of 31 provincial regions in China (*10-4).

Roughly, it could be found that the H2: Different regions have significantly western regions, e.g., , Tibet, etc., different CTRs. show higher CTRs over the eastern re- gions, e.g., Shanghai, Beijing, Tianjin, The Friedman test results are shown in , etc. Table 4, which indicates that hypothesis 2 To further validate this phenomenon, could be accepted, since p-value is less we grouped provinces into several re- than 0.05, meaning at least two regions gions, i.e., eastern region, central region, have significantly different CTRs. and western region (as shown in Table 3), following the official criteria [9]. Eastern Central Western Thus, to further evaluate the differenc- Beijing Shanxi Guizhou es among different regions in China, a Tianjing Jilin Qinghai new hypothesis is proposed. Shanghai Heilongjiang Sichuan Jiangsu Anhui Ningxia Guangdong Shaanxi

392 The 2010 International Conference on E-Business Intelligence

Hebei Henan Gansu First, the differences of economic le- Liaoning Hubei Inner Mongolia vels of different regions in China lead to Zhejiang Hunan Guangxi different activeness of Internet business. Yunnan For example, Eastern region in China in- Shandong Chongqing clude the most productive local econom- Hainan Xinjiang ics, e.g., Shanghai, Beijing, Fujian, Tibet Guangdong, etc., while Central and Western regions do not. Higher economic Table 3: Classification of China’s Regions. level means more Internet business, lead-

ing to more Internet Ads to be displayed N Chi-square Df Asymp. Sig. 26 48.308 2 .000 to the Internet users. Second, the network penetration rates Table 4: Friedman test of 3 regions. of different regions are also quite differ- ent, as shown in Figure 2. Normally, the Moreover, to further analyze the re- Eastern region has the highest penetration gional difference on CTRs in details. The rate, which means that Internet users are Wilcoxon test is performed on paired more mature and are immune to Internet comparison on any two regions. The re- Ads to some extent. Nevertheless, due to sults are as shown in Table 5. the low network penetration rates in Cen- tral and Western regions, the Internet us- Z Asymp. Sig. (2-tailed) ers in those regions are still not quite an- Central – -4.457 tipathy to Internet Ads, which leads to .000 Eastern (a) higher CTRs than in Eastern region. Central – -4.457 .000 Third, different regions have different Western (a) regional cultures, which can also take ef- Western -4.254 .000 fect on the browsing and clicking beha- – Eastern (b) viors on Internet Ads. Generally speaking, the people in Eastern region have more Table 5: Wilcoxon test on 3 regions. intense pace of life than Central and Western region, which means that the In- Table 5 shows that the CTRs of any ternet users in Eastern region tend to re- two regions are significantly different. trieve useful information directly and ig- Specifically, Central region has the high- -4 nore Internet Ads in many situations. est average CTR (8.26081 * 10 ), West- However, the Internet users in Central ern region has the middle average CTR -4 and Western regions prefer more enter- (7.428064 * 10 ), while Eastern region tainment activities on Internet, thereafter has the lowest average CTR (5.610809 * -4 Internet Ads, which are usually colorful 10 ), showing Eastern region has clear and appealing, may have higher possibili- difference on CTRs with the other two ty to be browsed and clicked for enter- regions. Intuitively, it shows that people tainment purposes by Internet users in in Central and Western regions of China Central and Western regions. tend to click more on Internet Ads than To summarize the above discussions, the people in Eastern region of China. the Internet users in Eastern region of The major reasons that cause this situa- China are more immune to Internet Ads tion may stem from the maturity of adopt- than those in Central and Western regions ing Internet Ads, which can be further of China. discussed in three aspects.

393 The 2010 International Conference on E-Business Intelligence

Fig. 2: Network Penetration of China [1].

could be generated than the same Internet 5. Managerial Analysis Ads distributed to Central and Western regions, due to the lower CTR, meaning From a managerial perspective, this sec- more impressions have been shown to tion will further discuss the Internet Ads Internet users in Eastern region of China. marketing strategies and countermeasures, For intermediates, generally, Internet considering the regional impacts, for Ads intermediates make profits by pur- three major Internet Ads players, i.e., chasing advertising positions from web- websites as the network media, interme- sites and selling them to advertisers with diates who buy advertising positions in packaging and providing value-added mass and sell to advertiser separately; and Ads services. Usually, intermediates will the advertisers, who distribute their own buy some advertising position from web- Internet Ads. sites will a packaged price per month. For websites, if CPM (cost per thou- Thereafter, considering the regional im- sand impressions) pricing strategy is pacts on CTRs, intermediates may prefer adopted, website operators could think to charge the Internet Ads distributed to about to increase the CPM price of Inter- Central and Western regions by CPC net Ads distributed to Central and West- pricing strategy, while charging the Inter- ern region, since the number of clicks of a net Ads distributed to Eastern region by thousand impressions could be higher CPM pricing strategy, if possible, to max- than the same Internet Ads distributed to imize the possible arbitrage. Eastern region, due to the higher CTR, For advertisers, due to the high immun- which means the click performance the ity of Internet Ads in Eastern region, ad- Internet Ads could be better. If CPC (cost vertisers may prefer to pay with CPC per click) pricing strategy is adopted by a strategy, since low cost for clicks could website, the website operator can think be paid while high impressions could be about to increase the CPC price of Inter- achieved due to the lower CTR in Eastern net Ads distributed to Eastern region, region. In Central and Western regions, since more impressions for per click CPM could be more preferred by adver-

394 The 2010 International Conference on E-Business Intelligence tisers, since the cost for thousand impres- This work was partly support by the Na- sions could be fixed, while high clicks tional Natural Science Foundation of could be achieved, due to the higher CTR China (Grant 70890083/71072015), the in Central and Western regions. MOE Project of Key Research Institute of Humanities and Social Sciences at Uni- 6. Conclusion versity of China (7JJD63005).

Based on the 26-days logo data of from a References Chinese Internet Ads intermediate, this paper discussed the regional impact on [1] CNNIC, The 25th Chinese Internet CTRs of Internet Ads in China. The em- Development Statistic Report, 2010 pirical analysis shows that there exist sig- [2] iResearch, Chinese Internet Ads nificant differences among the 31 provin- Development Report in 2009 and cial regions in Chinese mainland. Moreo- 2010, 2010 ver, the test results show that the average [3] Gian M. Fulgoni, Marie Pauline CTR in Eastern region in China is signifi- Mörn. How online advertising cantly lower than those in Central and works, Empirical Generalizations in Western regions in China. This paper also Advertising Conference for Industry investigated the reasons that caused this and Academia, December 4-5, 2008 situation in China, e.g., the difference of [4] Cho, Chang-Hoan. Factors Influen- economics development, the different cing Clicking of Banner Ads on the network penetration rates, as well as the www. Cyberpsychology & Behavior, differences in regional cultures, of East- 2003, Vol 6, No 2, 201-215. ern, Central and Western regions of Chi- [5] Chang-Hoan, Hongsik John Cheon na. Thereafter, immunity of Internet Ads (2004). Why Do People Avoid Ad- leads to the lower CTR like in Eastern vertising on the Internet, Journal of region in China. Furthermore, based on Advertising, 33 (4), 89-97. the analysis, from a managerial perspec- [6] Cho, Chang-Hoan. (2003). Factors tive, some comments and countermeasure, Influencing Clicking of Banner Ads regarding the region factor, are proposed on the WWW. Cyberpsychology & for the three major Internet Ads players, Behavior, 6(2), 201-215. i.e., websites, intermediates and advertis- [7] Hong C. Silly Catalog promote In- ers. ternet Localization Process. China There still exist some limitations in this dayily Forum, 2008 paper. First, the log data is just from one [8] Xinwei L., Zhong W. Research on intermediate. To what extent the data can Regional Internet Marketing Fea- represent the overall situation of Internet tures among Beijing, Shanghai, Ads in China can hardly been validated. Guangdong. Shanghai Information, Moreover, there exist many other factors 2009. influencing CTRs, which have also been [9] Chun Lin, Hans Hendrischke, The proposed in this paper and need to be fur- Territories of the People's Republic ther analyzed together with region factor. of China. Europa Publications, First edition, 2002, London. 264 pages. Acknowledgements

395