Do #eCommerce #analytics prove #customer #affiliation?

Analytics for e-Commerce can be split between two contrasting and not necessarily companion aspects.

  • analytics of success in terms of market development, profile or response through hit or link ratings
  • analytics of sales or revenue

Dependant upon the purpose of a company’s e-commerce deployment one or other but rarely both of these analytics will seek to prove the status of the investment and allow informed strategic decision making.

According to the United Kingdoms Office for National Statistics (ONS) in 2002 “e-commerce is likely to have a huge impact on the way we do business. It has the potential to lead to dramatic growth in trade, increase markets, improve efficiency, effectiveness and transform business processes”. This notion of increasing markets and improving efficiency and effectiveness through internet technology is underwritten in ‘Blown to Bits: How the New Economics of Information Transforms Strategy’. (Evens, 1997). Here the authors determine that three key factors create success in e-commerce businesses: reach, richness and affiliation.

  • Reach indicates numbers of users interacting in e-commerce.
  • Richness is the depth or quality of information user’s gain.
  • Affiliation is a measure of correlation or affinity with users.

The internet’s ease of accessibility for consumers means there is a levelling of reach between small and larger enterprises. This is due to small companies being able to gain the sort of competitive advantage associated with advertising in other media such as newspapers or television. Richness or detail of information is also easier to provide online with multiple data sources, animations, video and databases linked to internet and extranet applications. Only affiliation with the customer is unchanged and this creates a greater effectiveness and success through tailored services such as the Amazon personal front page (Wolverton, 2000) by making a new store for each user. It is in an attempt to measure affiliation or user buy in, that many commercially based internet statistics are created. This pressure to create statistics often to prove the effectiveness of the method by companies with a vested interest renders many internet statistics unreliable . To quantify affiliation and supposedly shed light on the nature of business transformation, increased markets and improved efficiency of e-commerce as a marketing channel numerous calculation methods are in use.

Credibility and Trustworthiness

There are two distinct terms associated with the effectual usage and profitability of e-commerce, credibility and trustworthiness. James Rosenfields (2001) determined that much of the available internet statistics are “lies or damned lies”. This is because they are “an amalgam of guesswork, wishful thinking, pie-in-the-sky optimism, end-of-the-world pessimism and trivial commonplaces” (Rosenfields, 2001) and therefore lack credibility. Much of the underlying basis of e-commerce statistics are not transparent and in many cases appears to be clouded by commercial purpose. James Rosenfields further suggests that the longevity of the medium, newspaper being old verses the internet being new has a direct correlation in its perceived trustworthiness. This notion of trustworthiness is essential in developing business to consumer (B2C) e-commerce in a way that many companies are used to working by letters of agreement, tacit contracts or unsecured credit whereas consumers are not. There is a general perception by consumers that more credit fraud is created by unsecured internet connections than them not keeping a watchful eye on their credit cards in shops and bars. In fact 5% of all internet sales using credit cards are fraudulent and carried out by consumers against merchants (Sandoval, 2001). In consequence credibility and trustworthiness are often outside the experience of many web retailers and sets an environment of fear in e-commerce.

Foundational Analytics

Given the commercial value of internet analytics governmental websites have been chosen as a prime source of information. While governmental statistics may be open to influences by national economic concerns, the first and most fundamental statistic of total population holds no specific national benefit so may be considered viable. The U.S Census Bureau issues world population data which has been factored to be 6,446,131,400 (U.S Census Bureau, 2004) at a mid point in the year of 2005. Conversely this data is used with an algorithm by David Levin from the University of North Carolina in his population clock which goes up three to four people per second. At 12.30pm 7th April 2005 the population stood at 6,511,632,444 (Levin, 0000). While this format of data presentation is quite compelling it is without consensus and as such flawed. If the population of the world is not an agreed figure it is therefore difficult, if not impossible to determine a percentage usage of the internet as a derivative of world population. Further more to calculate return on investment (ROI) in association with internet usage, sales figures and market penetration by population methods would also be impossibly flawed.

Analytics of Success

There are many methods of analytical calculation in use on the internet e.g. click through rates (CTR) which are derived from using coded links which log activity and flow direction, website visitors which is an extrapolation of site hits, data requests and site session logs. However the underlying data is captured in only two arenas. Data is derived from either locally assessed user activity or remotely assessed communication packets or logging. These two data sources then form the basis of calculations in association with national and worldwide population statistics.

Activity Analytics

Activity statistics are derived through the use of pre-defined or targeted user groups surveys or reviews, cookies, banners, html email, adware and spyware.

Predefined groups are used by Nielson NetRatings to produce commercial guidelines for internet design. They include recommendation of functionality, content design and creation and accessibility. Commonly, internet experts are used to assess websites, they perform heuristic evaluations based upon defined criteria. This method is highly favoured in commercial arenas as it provides absolutes. An unfortunate aspect of this sort of evaluation is a level of condescension is required to dumb down the findings to allow for non technical usage of the internet. While this kind of study may produce insights into professional aspirations for internet usage it says very little about mass consumption. The motivation for this kind of study is consultancy or publication fees. The use of this sort of data to describe national or global interactions is highly suspect.

Targeted user group are used to perform ethnographic reviews commonly in users office or home that study interactions based upon goals, walkthroughs or scenarios. This type of study is highly effectual in producing notional interaction behaviour. Until recently the use ethnographic reviews has been associated with a response to an existing website rather than the underlying philosophy or ethos.

Academic References (some links require fees)

Cranor, L., Langheinrich, M., Marchiori, M., Presler-Marshall, M. (16th April 2002) The Platform for Privacy Preferences 1.0 (P3P1.0) Specification. W3C Recommendation. Published on W3C. [Electronic version]. Retrieved March 10th, 2005 from

Sandoval, G., (5th October, 2001) As Net fraud grows, so do e-tailers’ fears. CNET Published on ZDNet News. [Electronic version]. Retrieved March 10th, 2005 from

Wolverton, T. (13th June, 2000). Amazon, others add personal touch to home pages. CNET. Published on [Electronic version]. Retrieved March 10th, 2005 from,+others+add+personal+touch+to+home+pages/2100-1017_3-241869.html

Rosenfield, J. R., (November 2001). Lies damned lies, and internet statistics. Direct Marketing. Published: Garden City. Vol 64, Iss 7 pg 61 – 64. [Electronic version]. Retrieved March 10th, 2005 from Proquest.

Mullarkey, G., W., (2004). Internet Measurement data – practical and technical issues. Marketing, Intelligence & Planning. Vol 22, No 1, pg 42-58. [Electronic version]. Retrieved March 10th, 2005 from Emerald Fulltext

Frangos, A. (16th September 2002). E-Commerce (A Special Report): Selling Strategies — Search Engines: A Question for Google — Why do you think your Q&A site will succeed — after so many others have failed? Wall Street Journal (Eastern edition). Published: New York, N.Y. pg. R.8. [Electronic version]. Retrieved March 10th, 2005 from Proquest.

Phippen, A., Sheppard, L., Furnell, S. (2004). A practical evaluation of Web analytics. Internet Research: Electronic Networking Applications and Policy. Vol 14 No 4, pg 284 – 293. [Electronic version]. Retrieved March 10th, 2005 from Emerald Fulltext

Hallerman, D., (29th November 2004). Why online advertising matters. Marketing. Published: Toronto 2004 Vol 109. Iss 39. pg E34  [Electronic version]. Retrieved March 10th, 2005 from Proquest.

McClelland, S., (Febuaray 2004) Mobile internet statistics flatter to deceive. Telecommunications International. Published: Norwood. Vol 38. Iss 2. pg 15. [Electronic version]. Retrieved March 10th, 2005 from Proquest

Office for National Statistics (2002). e-Commerce And Internet Why we measure web business and internet access. Office for National Statistics [Electronic version]. Retrieved March 10th, 2005 from No longer functional

U.S Census Bureau (2004). Total Midyear Population for the World: 1950-2050
(Data updated 9-30-2004) [Electronic version]. Retrieved April 7th, 2005 from No longer functional

Levin, D.(00). Population for the World counter based upon Data updated 9-30-2004 University of North Carolina [Electronic version]. Retrieved April 7th, 2005 from No longer functional

Nielsen/NetRatings (March 2005) U.S. Internet Usage Shows Mature Growth, Forcing Innovation of New Web Offerings.  Nielsen/NetRatings [Electronic version]. Retrieved April 7th, 2005 from

U.S Census Bureau (2004) Quarterly Retail E-Commerce Sales. U.S Census Bureau. [Electronic version]. Retrieved April 7th, 2005 from No longer functional

E-Stats (2004) Statistical Methodology – ASM. U.S Census Bureau. [Electronic version]. Retrieved April 7th, 2005 from No longer functional

Foley, P. (2001) Internet and e-commerce statistics. European Business Review. Vol 13, No. 2. [Electronic version]. Retrieved March 10th, 2005 from Emerald Fulltext.

Clayton, T., Tsvelik, M., Waldron, K. (2002) International e-Commerce Benchmarking Experimental Statistics Database. Office of National Statistics: New Economy Branch.  [Electronic version]. Retrieved March 10th, 2005 from National Statistics Online.

Clayton, T. (2002) Towards a Measurement Framework for International e-Commerce Benchmarking. Office of National Statistics: New Economy Branch.  [Electronic version]. Retrieved March 10th, 2005 from National Statistics Online.

Newman, Eric J; Jr, Donald E Stem; Sprott, David E (2004) Banner advertisement and Web site congruity effects on consumer Web site perceptions. Industrial Management & Data Systems; Vol 104. No. 3.  [Electronic version]. Retrieved March 10th, 2005 from Emerald Fulltext

Unknown. (22nd November 2004) Value of Internet sales doubles in 2003. Office of National Statistics.  [Electronic version]. Retrieved March 10th, 2005 from National Statistics Online.

Unknown. (16th Febuary 2005). Internet sales surge 21% in January. Interactive Media in Retail Group (IMRG). [Electronic version]. Retrieved April 7th, 2005 from National Statistics Online.

Unknown. (17th March 2005). Retail Sales: Underlying sales are in decline. Office of National Statistics. [Electronic version]. Retrieved April 7th, 2005 from National Statistics Online.

Tagged : / / / / / / / / / / /

#Converting #browsers to #buyers, exploring what drives #consumer #choice in internet e-commerce 2005

The following is a paper I wrote in 2005. From WWW/Internet 2005 Proceedings

Converting browsers to buyers: exploring what drives consumer choice in internet e-commerce

Why do internet users behave as they do, are their activities solely determined by website design? Or do they create their own pathways as a response to designated systems. For many, internet design is about the imposition of schemas, predetermined flows and consumer motifs, allowing the shepherding of an understood and mapped user towards buying products and services. However if this were true then every browser would also be a buyer. The underlying concepts of current website design rely on a number of pretexts which, when reviewed in relation to human activity and interaction, become questionable in their veracity.


There is recognition [16] that there is limited information in the understanding of the reciprocity of attitudes and behaviour constituting the relationship between internet shoppers and e-commerce websites. This is in contrast to commercially driven usability and web metrics companies who assert their findings based upon activity patterns often using statistically small samples [15]. Developing an understanding of the relationship between users and websites is key to determining patterns of interaction. Patterns of interaction are currently under investigation in two distinct ways, by using reflective and diagnostic methodologies. Reflection upon measurable activity, clicks, information foraging [5] and sales provide compelling insights for business metrics, can be limited by their subjective constituents. In turn diagnosis based on reviewers or heuristic interpretations with little user involvement [15] produce contentious results. This study will attempt to combine both forms of investigation with a large participant group study producing empirical data to be reviewed using both quantitative and qualitative approaches.

1.1         Research aims

The aims and objectives of the research can be summarised as follows:

  • Why do internet users behave as they do, are their activities solely determined by website design? Or do they create their own pathways as a response to designated systems.
  • List common behaviours and attributes to discern if there is a pattern that can be mapped and predicted.

These present some added details in comparison to the original project aims and objectives, described in the project proposal (appendix 6). Differences and the reasons for them are discussed in the final conclusions (chapter 6).

1.2      Conceptual investigation

  • Technological Determinism
  • How it came about
  • Engineering to computing rather than social science to computing
  • Boudieur absolutes – linking social theory with technology
  • Statistical absolutes taken by industry
  • Conclusion
  • Dialectic Computing
  • Social Computing
  • Tavistock Institute
  • Human Computer Interaction
  • Cultural Cognition
  • Embodied Interaction
  • Organic Systems
  • Commodities
  • Biography
  • Conclusion
  • Shopping
  • Consumption
  • Dialectic Interaction
  • Tensions
  • Dynamic
  • Conventions
  • Mediation
  • Trust
  • Conclusion
  • Investigation Methodologies
  • Reflective methodologies
  • Information Scent
  • Internet Statistics
  • Diagnostic methodologies
  • Observational
  • Active Narrative
  • Goals
  • Adaptations
  • Conventions
  • Conclusion
  • Lexical Distribution of Activity
  • Activity as a language
  • Conclusion


The main area of this research is to gain an understanding of how people use technology as an extension of their world [10], specifically in the mode of consumer e-commerce. This world view is augmented by a representation of social constructs by an evolutionary process of active agents of transformation [4] in technology and specifically the creation of an electronic media habitus. This constituency maintains its own cultural capital and produces a parallel distinctive counterpoint to popular and consumer cultures. The nature of human computer interactivity, it’s cultural, educational and gender attributes has a key influence on the unification of money, knowledge and technological aspiration. These factors are also represented in technological deprivation and the personalised safety of internalised ignorance.

To relate instances of electronic media habitus a combination of activity biography [1], [12] and negotiated conventions [7] enable the development of an activity definition index. The cultural disposition of technology, interactions and resultant pathways remain difficult to interpret without recourse to such a framework.


Existing methodologies produce results that use complex mathematics to create algorithms [6], create subjective rules of design [15] or usability inspection tools [3]. Which are normally only used with existing websites, only reflecting upon current interactions.

2.1 Information Scent

The cognitive walkthrough of the web has evolved based upon the notion that users decide on their course of action based upon cues, which derive behavioural patterns of interaction then form guide routes of information scent [3]. Information scent has also been developed using aspects of information foraging both structured and unstructured [6]. The Bloodhound project seeks to establish a clear method showing consistent, measurable elements that provide benefit in the design of websites. However there is contention in the effectiveness of this methodology [14] by commercially driven consultants.

2.2 Internet Statistics

There remains a problem with accessing “actionable statistics” [8] for businesses, and while their credibility and measurability remains opaque there will be a question regarding their veracity [18].


The general interpretation of an open and untamed [2] source of information like the World Wide Web (WWW) requires a systematic review of actions. Actions and user activity [11] in relation to an observable world require a common representation to determine navigation and related target acquisition. Ethnographic studies related by an in-series testing system can reduce the anomalous results associated with subjective reflective data. Ultimately a lexical definition of activity is needed; in the interim the term narrative enables an interpolative review of data which will provide a clearer definition of activity.

3.1 Narrative

The understanding of human interaction can be viewed as participation in the creation of personal historical elements, which allow dispersion in potential trajectories [10] evolving of a self imposed narrative. This narrative can be observed in linguistic and engendered functions which require definition and contextualisation. However to effectively map these functions a lexicon approach [9] as associated with endangered languages, would allow the use of rational linguistic dimensions including orthography, morphology, syntax and semantics. The creation of a lexical basis [17] makes individual actions communicable aspects of communities of actions with related compound, processed and adaptive meanings.

Several hypothetical goals, adaptations and conventions can be derived from this research which will further refine and delineate additional aspects of narrative behavior to produce foundational lexical and activity indices.

3.1.1 Goals

Goals can be a descriptor of predetermined final destinations which can subsequently be reduced to a form of knowledge morpheme. These inter-related sub-rationale units while distinct and finite offer an activity based response to catalytic impositions by addition and adaptation.

3.1.2 Adaptations

Adaptation allows the extension of narratives [13] creating alternative perspectives on the same object or situation. Further modifications can be made in a process of engagement, by determining the user’s perceptions or discernment of active adversity which produces redirection.

3.1.3 Conventions

Conventions allow the use of avatars [7] to create nodes within lexical frameworks, providing index points in a narrative activity. Agreement of conventions in social, emotional and commercial arenas for completion, enable a measurable resolution to tasks.


Data for this paper is being gathered through a three tiered research process. The target group for this research is consumers who purchase online, non-experienced WWW users based in the United Kingdom.

An initial pilot survey link has been introduced onto a number of United Kingdom based online shopping directories. Control questions have been used to define the target group and acquire basic demographic information. The survey consists of open ended questions with text areas to allow user to express their opinions on their online purchase experiences.

The main survey will be derived from the pilot by asking detailed questions related to the pilot results. This survey will use menu and dropdown tools with text areas to create both quantitative and qualitative primary data.

The final counterpoint survey will involve twelve participants (six consumers and six heuristic users) working on a series of scenario based activities derived from the main survey results. This ethnographic study will allow interactive testing and appraisal of user preferences, requirements and actual activity.


While this paper seeks to review and define the boundaries of an ongoing associated data gathering exercise it has also produced a number of testable hypotheses to be reviewed after data acquisition.

The linking of action cues with ethnographics has the potential to define activity components, constituents, usage and compound derivatives which will allow measurable patterns of formation and defendable narrative component interpretation.

The use of lexical representations will provide a framework for the indexing of interconnected activity components which currently operate under diverse notations.

The ability to interpret interaction will form the basis of other studies to better understand and design e-commerce sites based on human interpretive activity.


[1]   Appadurai, A. (1988) The social life of things. Commodities in cultural perspective, pp. 64–91. Cambridge:   Cambridge University Press.

[2]   Benyon, D., Turner, P., Turner, S. (2005). Designing Interactive Systems.

[3]   Blackmon, M., H., Polson, P., G., Kitajima, M., Lewis, C. (April 2002) Cognitive walkthrough of the Web. Conference on Human factors in Computing Systems: Proceedings of the SIGCHI conference on Human Factors in computing systems: Changing our World, changing ourselves. ACM Press: Minnesota, USA.

[4]   Bourdieu, P. (1990). Distinction: A Social Critique of the Judgement of Taste. Harvard University Press.

[5]   Chi, H., Pirollie, P., Pitkow, J. (2000) The Scent of a Site: A System for Analyzing and Predicting Information Scent, Usage and Usability of a Web Site. Xerox Palo Alto Research Centre.

[6]   Chi, H., Rosien, A., Supattanasiri, G., Williams, A., Royer, C., Chow, C., Robles, E., Dalal, B., Chen, J., Cousins, S.  (April 2003). Web usability: The bloodhound project: Automating discovery of web usability issues using the InfoScent™ simulator. Proceedings of the SIGCHI conference on Human factors in computing systems. Pages: 505 – 512. ACM Press: New York, NY, USA.

[7]  Clanton, C., Marks, H., Murray, J., Flanagan, M., Arble, F. (1998). Interactive narrative: stepping into our own stories. Conference on Human Factors in Computing Systems: CHI 98 conference summary on Human factors in computing systems. Pages: 88 – 89.  ACM Press: New York, NY, USA

[8]   Foley, P. (2001) Internet and e-commerce statistics. European Business Review. Vol 13, No. 2. Published: Emerald Fulltext.

[9]   Gulrajani, G. (August 2003) SHAWEL: Sharable and interactive Web-Lexicons. Max-Planck-Institute for Psycholinguistics. Max-Planck-Institute: Nijmegen

[10]   Jennings, P. (2005). Constructed Narratives a Tangible Social Interface. Creativity and Cognition: Proceedings of   the 5th conference on Creativity & cognition. Pages: 263 – 266. ACM Press: New York, NY, USA.

[11]   Jul, S., and Furnas, G., W. (1997) Navigation in Electronic Worlds: A CHI 97 Workshop. SIGCHI Bulletin Vol 29, No 4 October.

[12]   Kopytoff, I. (1988) The cultural biography of things: commoditization as process. In: Appadurai, A. (ed.) The social life of things. Commodities in cultural perspective, pp. 64–91. Cambridge: Cambridge University Press.

[13]   Nakhimovsky, A. (June 1988) Special issue on tense and aspect: Aspect, aspectual class, and the temporal structure of narrative. Computational Linguistics, Volume 14 Issue 2 Pages: 29 – 43. MIT Press:   Cambridge, MA, USA

[14]   Nielsen, J. (August 2, 2004). Deceivingly Strong Information Scent Costs Sales. Alertbox from

[15]   Nielsen, J. (March 19, 2000). Why You Only Need to Test With 5 Users. Alertbox from

[16]   Perea y Monsuwe, T., Dellaert, B., G., C., Ruyter, K. (2004). What drives consumers to shop online? A literature review. International Journal of Service Industry Management. Vol 12, No 1, Pages 102-121. Emerald Fulltext.

[17]   Pustejovsky, J. (December 1991). The Generative Lexicon. Computational Linguistics. Volume 17 Issue 4 Pages: 409 – 441. MIT Press: Cambridge, MA, USA.

[18]   Rosenfield, J. R., (November 2001). Lies damned lies, and internet statistics. Direct Marketing. Published: Garden City. Vol 64, Iss 7 pg 61 – 64.

Tagged : / / / / / / / / / / / / / / / / / / / / / / / / / / /