基于渠道点击路径数据的多渠道网络广告组合效应测评研究

doi:10.16381/j.cnki.issn1003-207x.2022.0304

摘要/Abstract

摘要：

互联网、移动互联网和社交媒体的飞速发展，为电子商务企业带来了新型的网络广告营销渠道。为了探究多渠道网络广告的组合效应，本文根据选择集理论提出基于网络广告渠道分类的广告组合效应的相关研究假设及其概念模型，进而构建COX模型，基于电子商务企业的个体用户渠道点击路径数据，提取相关变量，对COX模型进行实证验证，分析了同渠道/跨渠道以及基于渠道分类的渠道组合效应。研究结果表明，企业发起类广告渠道到用户发起类品牌搜索渠道的渠道顺序会对购买产生正向的组合效应，而用户发起类广告渠道到企业发起类广告渠道的渠道顺序则会对购买产生负向影响。此外，通用型搜索转向品牌型搜索的渠道顺序会对购买产生正向的组合效应。最后，本文还按性别对数据进行分组，探究了不同性别用户网络广告效应的差异。

关键词: 多渠道网络广告, 组合效应, COX模型, 点击路径数据

Abstract:

The rapid development of the Internet， mobile and social media has brought many new online advertising and marketing channels to e-commerce enterprises. Under the background of integrated marketing and multi-channel online advertising， users may not only contact the online advertising channels once before purchase， and the channels contacted successively will affect consumers jointly but not independently. Therefore， in the effectiveness study of multi-channel online advertising， it is also necessary to consider the multi-channel combination effect of online advertising.The click-stream data of this study comes from the third party insurance agency website in Nanjing. After extracting， cleaning and transforming， the basic variables of this study are obtained.To explore the combinatorial effect of multi-channel online advertising， according to the choice set theory， related research hypotheses based on the classified combination effects of online advertising channels are proposed， and then the COX model is constructed. Based on the channel click path data of individual users in e-commerce enterprises， the relevant variables are extracted， and the COX model is verified by empirical analysis. Then the channel combination effect of the same channel， cross-channel and channel classification is analyzed.The results show that the combination of advertising channels from firm-initiated channels to brand search of customer-initiated channels will have a positive combination effect on purchases， while the combination of advertising channels from customer-initiated channels to firm-initiated channels will have a negative impact on purchases. In addition， the order of channels that move from general-purpose search to brand-type search has a positive combination effect on purchases. The data are grouped by gender to explore the differences in advertising effects of users of different genders. The results can help enterprises establish their marketing mix of advertising channels and mitigate the cost of ineffective advertising channel mix.

Key words: multi-channel online advertising, advertising combination effect, COX model, click path data

中图分类号:

F274

李莉, 曹秉坤, 杨振怡. 基于渠道点击路径数据的多渠道网络广告组合效应测评研究[J]. 中国管理科学, 2025, 33(9): 121-134.

Li Li, Bingkun Cao, Zhenyi Yang. Research on the Evaluation of Multi-channel Online Advertising Combination Effects Based on Channel Click Path[J]. Chinese Journal of Management Science, 2025, 33(9): 121-134.

图/表 8

图1

表1

表2

图2

表3

表4

扩展模型（一）估计结果"

分类	变量	风险率	系数	95%置信区间
CIC Brand	BrandSEA	1.495609	0.4025332	[1.459579,1.532528]
	BrandSEO	1.410488	0.3439355	[1.346172,1.477876]
	Direct	1.055243	0.0537712	[1.051508,1.058991]
CIC Generic	GenericSEA	1.323395	0.2802002	[1.266034,1.383354]
CIC Generic	GenericSEO	1.248082	0.2216081	[1.231896,1.264481]
FIC	SMS	0.9008651	-0.1043997	[0.8423919,0.9633971]
	EMA	0.9737893	-0.0265604	[0.906867,1.04565]
	Wechat	1.052956	0.0516013	[1.043588,1.062408]
	Referer	1.149355	0.139201	[1.090619,1.211255]
CIC Brand (Past)	PastBrandSEA	1.125024	0.1178047	[1.114671,1.135474]
	PastBrandSEO	1.120869	0.114104	[1.09891,1.143266]
	PastDirect	1.031976	0.0314749	[1.029824,1.034131]
CIC Generic (Past)	PastGenericSEA	1.020664	0.0204532	[1.006186,1.03535]
CIC Generic (Past)	PastGenericSEO	1.134267	0.1259864	[1.122828,1.145822]
FIC (Past)	PastSMS	1.00994	0.0098909	[0.9761242,1.044927]
	PastEMA	1.08972	0.0859211	[1.063675,1.116403]
	PastWechat	1.016902	0.0167605	[1.011276,1.022559]
	PastReferer	1.080558	0.0774777	[1.063972,1.097403]
控制变量	Tvc(PastPurchase)	1.032877	0.0323483	[1.030085,1.035676]
	PastCollect	1.034653	0.0340663	[0.9833217,1.088664]
	PastShare	0.7951518	-0.2292222	[0.737499,0.8573116]
	Age	1.019605	0.0194149	[1.018276,1.020935]
	Gender	0.7386099	-0.3029854	[0.7157744,0.7621739]
交互项（同渠道组合）	BrandSEA×PastBrandSEA	0.976354	-0.02393	[0.9701595,0.9825881]
	BrandSEO×PastBrandSEO	0.9854322	-0.0146749	[0.976238,0.9947131]
	Direct×PastDirect	0.9989715	-0.001029	[0.9988066,0.9991364]
	GenericSEA×PastGenericSEA	1.000192	0.0001916	[0.9916995,1.008757]
	GenericSEO×PastGenericSEO	0.9911484	-0.008891	[0.9870659,0.9952479]
	SMS×PastSMS	1.01099	0.0109301	[0.9918164,1.030534]
	EMA×PastEMA	0.9865158	-0.0135759	[0.9728393,1.000385]
	Wechat×PastWechat	1.001467	0.0014656	[1.000889,1.002045]
	Refer×PastRefer	1.017111	0.0169667	[1.008446,1.025852]
交互项（跨渠道组合）	BrandSEA×PastNoBrandSEA	0.9868122	-0.0132755	[0.9824447,0.9911992]
	BrandSEO×PastNoBrandSEO	0.9781465	-0.0220958	[0.9704117,0.9859429]
	Direct×PastNoDirect	0.9995069	-0.0004932	[0.9985183,1.000497]
	GenericSEA×PastNoGenericsea	0.975973	-0.0243204	[0.9654965,0.9865632]
	GenericSEO×PastNoGenericseo	0.9930447	-0.0069796	[0.9922149,0.9938753]
	SMS×PastNoSMS	1.011966	0.011895	[1.005377,1.018598]
	EMA×PastNoEMA	1.032431	0.0319161	[1.017919,1.047149]
	Wechat×PastNoWechat	0.9908988	-0.0091428	[0.9884687,0.993335]
	Refer×PastNoRefer	0.9902901	-0.0097573	[0.9835826,0.9970435]

表4

表5

表6

参考文献 45

[1]	Lim J S， Ri S Y， Egan B D， et al. The cross-platform synergies of digital video advertising： Implications for cross-media campaigns in television， Internet and mobile TV［J］. Computers in Human Behavior， 2015， 48： 463-472.
[2]	Li H A， Kannan P K. Attributing conversions in a multichannel online marketing environment： An empirical model and a field experiment［J］. Journal of Marketing Research， 2014， 51（1）： 40-56.
[3]	Assael H. From silos to synergy［J］. Journal of Advertising Research， 2011， 51 （1）： 42-58.
[4]	Naik P A， Peters K. True synergy for real effects： How to control integrated marketing successfully［J］. GfK Marketing Intelligence Review， 2015， 7（1）： 34-41.
[5]	Danaher P J， Dagger T S. Comparing the relative effectiveness of advertising channels： A case study of a multimedia blitz campaign［J］.Journal of Marketing Research， 2013， 50（4）： 517-534.
[6]	Chatterjee P， Hoffman D L， Novak T P. Modeling the clickstream： Implications for web-based advertising efforts［J］. Marketing Science， 2003， 22（4）： 520-541.
[7]	Joo M， Wilbur K C， Cowgill B， et al. Television advertising and online search［J］. Management Science， 2013， 60（1）： 56-73.
[8]	Nottorf F. Modeling the clickstream across multiple online advertising channels using a binary logit with Bayesian mixture of normals［J］. Electronic Commerce Research and Applications， 2014， 13（1）： 45-55.
[9]	Xu L， Duan J A， Whinston A. Path to purchase： A mutually exciting point process model for online advertising and conversion［J］. Management Science， 2014， 60（6）： 1392-1412.
[10]	Bollinger B K， Cohen M A， Lai J. Measuring asymmetric persistence and interaction effects of media exposures across platforms［J］. SSRN Electronic Journal， 2013：2342349.
[11]	Stolyarova E， Rialp J. Synergies among advertising channels： An efficiency analysis［J］. Journal of Promotion Management， 2014， 20（2）： 200-218.
[12]	Micu A C. Testing for a synergistic effect between online publicity and advertising in an integrated marketing communications context ［D］. Columbia： University of Missouri - Columbia， 2005.
[13]	Micu A C， Thorson E. Leveraging news and advertising to introduce new brands on the web［J］. Journal of Interactive Advertising， 2008， 9（1）： 14-26.
[14]	Loda M D， Coleman B C. Sequence matters： A more effective way to use advertising and publicity［J］. Journal of Advertising Research， 2005， 45（4）： 362-372.
[15]	Micu A C， Pentina I. Examining search as opposed to experience goods when investigating synergies of Internet news articles and banner ads［J］. Internet Research， 2015， 25（3）： 378-398.
[16]	Danaher P J， Dagger T S. Comparing the relative effectiveness of advertising channels： A case study of a multimedia blitz campaign［J］. Journal of Marketing Research， 2013， 50（4）： 517-534.
[17]	Klapdor S， Anderl E， Schumann J H， et al. How to use multichannel behavior to predict online conversions［J］. Journal of Advertising Research， 2015， 55（4）： 433-442.
[18]	禹杭，陈香兰. 含蓄还是直白？ ——隐喻广告效果研究回顾与展望［J］. 外国经济与管理， 2018， 40（10）： 54-65.
	Yu H， Chen X L. Implicit or explicit： Literature review and prospects of research on the effects of advertising metaphors［J］. Foreign Economics & Management， 2018， 40（10）： 54-65.
[19]	Wu G， Sandeep B. The synergy effect of print and web advertising： A field experiment［C］//Proceedings of the 2007 Conference of American Academy of Advertising Conference， Burlington， VT，USA，April 12-15， 2007： 208-213.
[20]	Micu A C. Testing for a synergistic effect between online publicity and advertising in an integrated marketing communicationscontext［D］. Columbia： University of Missouri， 2005.
[21]	Voorveld H A M， Valkenburg S M F. The fit factor： The role of fit between ads in understanding cross-media synergy［J］. Journal of Advertising， 2015， 44（3）： 185-195.
[22]	Shankar V， Malthouse E C. The growth of interactions and dialogs in interactive marketing［J］. Journal of Interactive Marketing， 2007， 21（2）： 2-4.
[23]	Wiesel T， Pauwels K， Arts J. Practice prize paper—Marketing’s profit impact： Quantifying online and off-line funnel progression［J］. Marketing Science， 2011， 30（4）： 604-611.
[24]	Calder B J， Malthouse E C， Schaedel U. An experimental study of the relationship between online engagement and advertising effectiveness［J］. Journal of Interactive Marketing， 2009， 23（4）： 321-331.
[25]	Van den Poel D， Buckinx W. Predicting online-purchasing behaviour［J］. European Journal of Operational Research， 2005， 166（2）： 557-575.
[26]	Snyder J， Garcia-Garcia M. Advertising across platforms： Conditions for multimedia campaigns［J］. Journal of Advertising Research， 2016， 56（4）： 352-367.
[27]	Sridhar S， Germann F， Kang C， et al. Relating online， regional， and national advertising to firm value［J］. Journal of Marketing， 2016， 80（4）： 39-55.
[28]	王平，孙皖玲，范秀成，等.广告协同效应研究述评及展望［J］.外国经济与管理，2016，38（11）：14-28+128.
	Wang P， Sun W L， Fan X C， et al. A literature review of advertising synergy effect and prospects［J］. Foreign Economics & Management， 2016， 38（11）： 14-28+128.
[29]	Abou Nabout N， Lilienthal M， Skiera B. Empirical generalizations in search engine advertising［J］. Journal of Retailing， 2014， 90（2）： 206-216.
[30]	Ghose A， Yang S. An empirical analysis of search engine advertising： Sponsored search in electronic markets［J］. Management Science， 2009， 55（10）： 1605-1622.
[31]	Dou W， Lim K H， Su C， et al. Brand positioning strategy using search engine marketing［J］. MIS Quarterly， 2010， 34（2）： 261-279.
[32]	Lee W M， Sanderson M. Analyzing URL queries［J］. Journal of the American Society for Information Science and Technology， 2010， 61（11）： 2300-2310.
[33]	Libai B， Biyalogorsky E， Gerstner E. Setting referral fees in affiliate marketing［J］. Journal of Service Research， 2003， 5（4）： 303-315.
[34]	Van den Bulte C， Iyengar R. Tricked by truncation： Spurious duration dependence and social contagion in hazard models［J］. Marketing Science， 2010， 30（2）： 233-248.
[35]	Hauser J R， Wernerfelt B. An evaluation cost model of consideration sets［J］. Journal of Consumer Research， 1990， 16（4）： 393-408.
[36]	Moe W W， Fader P S. Dynamic conversion behavior at E-commerce sites［J］. Management Science， 2004， 50（3）： 326-335.
[37]	Yadav M S， Pavlou P A. Marketing in computer-mediated environments： Research synthesis and new directions［J］. Journal of Marketing， 2014， 78（1）： 20-40.
[38]	Rutz O J， Bucklin R E. From generic to branded： A model of spillover in paid search advertising［J］. Journal of Marketing Research， 2011， 48（1）： 87-102.
[39]	Baecke P， Van Den Poel D. Improving purchasing behavior predictions by data augmentation with situational variables［J］. International Journal of Information Technology & Decision Making， 2010， 9（6）： 853-872.
[40]	Johnson E J， Moe W W， Fader P S， et al. On the depth and dynamics of online search behavior［J］. Management Science， 2004， 50（3）： 299-308.
[41]	Murray K B， Häubl G. Explaining cognitive lock-In： The role of skill-based habits of use in consumer choice［J］. Journal of Consumer Research， 2007， 34（1）： 77-88.
[42]	Burnham T A， Frels J K， Mahajan V. Consumer switching costs： A typology， antecedents， and consequences［J］. Journal of the Academy of Marketing Science， 2003， 31（2）： 109-126..
[43]	Lambrecht A， Tucker C. When does retargeting work？ information specificity in online advertising［J］. Journal of Marketing Research，2013，50（5）： 561-576.
[44]	Hausman J A， Woutersen T. Estimating a semi-parametric duration model without specifying heterogeneity［J］.Journal of Econometrics，2014，178： 114-131.
[45]	Lambrecht A， Tucker C. When does retargeting work？ Information specificity in online advertising［J］. Journal of Marketing Research，2013，50（5）： 561-576.

变量类型	一级变量	描述	二级变量	描述
解释变量	CICBrand	当前用户发起类品牌型渠道点击量	BrandSEA	当前品牌付费搜索广告的点击量
			BrandSEO	当前品牌自然搜索次数
			Direct	当前直接访问次数
	CICGeneric	当前用户发起类通用型渠道点击量	GenericSEA	当前通用付费搜索广告的点击量
	CICGeneric	当前用户发起类通用型渠道点击量	GenericSEO	当前通用自然搜索次数
	FIC	当前企业发起类渠道点击量	SMS	当前短信广告的点击量
			EMA	当前电子邮件广告的点击量
			Wechat	当前微信推送点击量
			Referer	当前引荐广告的点击量
	PastCICBrand	过去用户发起的品牌型搜索类渠道点击量	PastBrandSEA	过去品牌付费搜索广告的点击量
			PastBrandSEO	过去品牌自然搜索次数
			PastDirect	过去直接访问次数
	PastCICGeneric	过去用户发起的通用型搜索类渠道点击量	PastGenericSEA	过去通用付费搜索广告的点击量
	PastCICGeneric	过去用户发起的通用型搜索类渠道点击量	PastGenericSEO	过去通用自然搜索次数
	PastFIC	过去企业发起类渠道点击量	PastSMS	过去短信广告的点击量
			PastEMA	过去电子邮件广告的点击量
			PastWechat	过去微信推送点击量
			PastReferer	过去引荐广告的点击量
		当前渠道的过去跨渠道点击量	PastNoBrandSEA	品牌付费搜索广告的过去跨渠道点击量
			PastNoBrandSEO	品牌自然搜索的过去跨渠道点击量
			PastNoDirect	直接访问的过去跨渠道点击量
			PastNoGenericSEA	通用付费搜索广告的过去跨渠道点击量
			PastNoGenericSEO	通用自然搜索过去跨渠道点击量
			PastNoSMS	短信广告的过去跨渠道点击量
			PastNoEMA	电子邮件广告的过去跨渠道点击量
			PastNoWechat	微信推送的过去跨渠道点击量
			PastNoReferer	引荐广告的过去跨渠道点击量

变量类型	划分依据	变量	描述
控制变量	个体异质性特征	PastPurchase	用户过去是否发生过购买行为
		PastCollect	用户过去是否发生过收藏行为
		PastShare	用户过去是否发生过分享行为
		Gender	用户性别 0-女；1-男
		Age	用户年龄
被解释变量		Convert	用户最终是否发生购买转化

分类	变量	风险率	系数	95%置信区间
CIC Brand	BrandSEA	1.346447	0.297469	[1.318364,1.375127]
	BrandSEO	1.21968	0.1985883	[1.187041,1.253215]
	Direct	1.044702	0.0437318	[1.041057,1.04836]
CIC Generic	GenericSEA	1.229635	0.2067177	[1.183997,1.277033]
CIC Generic	GenericSEO	1.076133	0.0733738	[1.066395,1.085959]
FIC	SMS	0.9003422	-0.1049804	[0.8445306,0.9598422]
	EMA	0.9665695	-0.0340021	[0.9075077,1.029475]
	Wechat	1.026653	0.0263038	[1.016372,1.037037]
	Referer	1.072776	0.07025	[1.054619,1.091247]
CIC Brand (Past)	PastBrandSEA	1.09085	0.0869576	[1.084562,1.097175]
	PastBrandSEO	1.031748	0.0312549	[1.010217,1.053739]
	PastDirect	1.008229	0.0081954	[1.00734,1.009119]
CIC Generic (Past)	PastGenericSEA	1.019775	0.0195815	[1.005542,1.034209]
CIC Generic (Past)	PastGenericSEO	1.124814	0.1176173	[1.114052,1.135679]
FIC (Past)	PastSMS	1.039186	0.0384373	[1.020704,1.058002]
	PastEMA	1.09627	0.0919131	[1.07569,1.117243]
	PastWechat	1.012529	0.0124508	[1.009428,1.015639]
	PastReferer	1.089078	0.0853313	[1.073539,1.104841]
控制变量	Tvc(PastPurchase)	1.033397	0.0328512	[1.030599,1.036202]
	PastCollect	1.062955	0.0610529	[1.010622,1.117998]
	PastShare	0.7658462	-0.2667739	[0.7100619,0.8260131]
	Age	1.020624	0.0204143	[1.019302,1.021948]
	Gender	0.7420095	-0.2983932	[0.7190838,0.7656661]

分类	变量	风险率	系数	95%置信区间
CIC Brand	BrandSEA	1.376832	0.3197853	[1.319545,1.436606]
	BrandSEO	1.234039	0.2102924	[1.173642,1.297543]
	Direct	1.056626	0.055081	[1.047285,1.066051]
CIC Generic	GenericSEA	1.30149	0.2635098	[1.186306,1.427858]
CIC Generic	GenericSEO	1.328776	0.2842584	[1.291911,1.366694]
FIC	SMS	0.9464716	-0.0550143	[0.8503287,1.053485]
	EMA	1.007904	0.0078733	[0.9439406,1.076202]
	Wechat	1.051919	0.0506165	[1.039848,1.064131]
	Referer	1.139992	0.1310214	[1.101564,1.17976]
CIC Brand (Past)	PastBrandSEA	1.08336	0.0800669	[1.06179,1.105367]
	PastBrandSEO	1.098837	0.0942524	[1.062654,1.136252]
	PastDirect	1.032823	0.0322963	[1.026494,1.039191]
CIC Generic (Past)	PastGenericSEA	1.030065	0.0296216	[0.997417,1.063781]
CIC Generic (Past)	PastGenericSEO	1.153627	0.1429112	[1.130881,1.176831]
FIC (Past)	PastSMS	1.02877	0.0283635	[1.004936,1.053168]
	PastEMA	1.083799	0.0804722	[1.060861,1.107233]
	PastWechat	1.013714	0.0136205	[1.007235,1.020234]
	PastReferer	1.08012	0.0770725	[1.047068,1.114216]
控制变量	Tvc(PastPurchase)	1.033188	0.032649	[1.030245,1.036139]
	PastCollect	1.049828	0.0486267	[0.9944068,1.108339]
	PastShare	0.7780524	-0.2509614	[0.7181272,0.8429781]
	Age	1.020356	0.0201518	[1.018926,1.021788]
	Gender	0.7441685	-0.2954877	[0.720327,0.7687992]
交互项（基于渠道分类的特定渠道顺序）	CICBrand×PastCICBrand	0.9988621	-0.0011385	[0.9985018,0.9992226]
	CICBrand×PastCICGeneric	0.989975	-0.0100756	[0.9830055,0.996994]
	CICBrand×PastFIC	1.000373	0.000373	[0.9995354,1.001211]
	CICGeneric×PastCICBrand	0.9913733	-0.0086641	[0.9901382,0.99261]
	CICGeneric×PastCICGeneric	0.9873087	-0.0127725	[0.9825931,0.992047]
	CICGeneric×PastFIC	0.9755327	-0.0247716	[0.9446239,1.007453]
	FIC×PastCICBrand	0.9917452	-0.008289	[0.9876548,0.9958526]
	FIC×PastCICGeneric	0.9940015	-0.0060166	[0.9660304,1.022782]
	FIC×PastFIC	1.001384	0.0013835	[1.000503,1.002266]

分类	变量	系数
分类	变量	男	女
CIC Brand	BrandSEA	0.3486749	0.3029304
	BrandSEO	0.1993119	0.2007566
	Direct	0.0600732	0.0525167
CIC Generic	GenericSEA	0.2297741	0.3453779
CIC Generic	GenericSEO	0.2755656	0.3027294
FIC	SMS	0.0100421	-0.1753532
	EMA	-0.009854	0.0118566
	Wechat	0.0571562	0.0417457
	Referer	0.1341778	0.1381804
CIC Brand (Past)	PastBrandSEA	0.121433	0.0716659
	PastBrandSEO	0.1194788	0.0896197
	PastDirect	0.0301799	0.03712
CIC Generic (Past)	PastGenericSEA	0.0080339	0.0339193
CIC Generic (Past)	PastGenericSEO	0.1500571	0.1384109
FIC (Past)	PastSMS	0.1092676	0.0063826
	PastEMA	0.0812032	0.0810788
	PastWechat	0.0271742	0.000853
	PastReferer	0.096782	0.0677919
控制变量	Tvc(PastPurchase)	0.0316208	0.0329459
	PastCollect	0.1184742	-0.0176073
	PastShare	-0.193947	-0.2853876
	Age	0.022278	0.0178563
交互项（基于渠道分类的特定渠道顺序）	CICBrand×PastCICBrand	-0.0010853	-0.001394
	CICBrand×PastCICGeneric	-0.0138712	-0.0072687
	CICBrand×PastFIC	-0.0015139	0.0013761
	CICGeneric×PastCICBrand	-0.0061048	-0.0088922
	CICGeneric×PastCICGeneric	-0.010905	-0.0154069
	CICGeneric×PastFIC	-0.0589481	0.0367443
	FIC×PastCICBrand	-0.0083223	-0.0094824
	FIC×PastCICGeneric	-0.0011416	-0.0024805
	FIC×PastFIC	0.0008727	0.0038986