From Interactions to Transactions: Designing the Trust Experience for Business-to-Consumer Electronic Commerce
Egger, F.N. (2003). From Interactions to Transactions: Designing the Trust Experience for Business-to-Consumer Electronic Commerce. PhD Thesis, Eindhoven University of Technology (The Netherlands). ISBN 90-386-1778-X.
- Télécharger le fichier PDF (1.07 MB)
Business-to-consumer electronic commerce on the Internet has revolutionised the purchase of products and services by giving consumers round the clock access to worldwide providers. However, B2C e-commerce has also shown to be associated with a myriad of factors hindering adoption and usage by private customers. Such factors include concerns regarding security and privacy, the unfamiliarity of some online services, lack of direct interaction with products, salespeople and fellow shoppers and the generally low credibility of online information. These factors were collectively defined as “trust issues”, as they refer to a purchase decision customers have to make in a situation of uncertainty and risk.
The first objective of this research was to build up substantive knowledge about which specific factors make customers trust e-commerce websites. The second objective was to build up and validate methodological knowledge in the form of tools that HCI practitioners can use to design and evaluate trust-shaping factors in e-commerce websites. On the basis of literature on trust and e-commerce surveys, a first model of trust in e-commerce (MoTEC) was developed. Through user tests, the initial model was refined to increase its descriptive power. The final MoTEC model contains four main dimensions, containing components and subcomponents. It is structured as follows:
Pre-interactional Filters refer to factors that may affect a person’s trust in an online vendor before accessing its website. They are composed of User Psychology and Pre-purchase Knowledge.
Interface Properties refer to surface cues in the user interface, namely graphic design and ease of use. The corresponding components are Branding and Usability.
Informational Content refers to the different types of information contained in the website. There are to main types of information, each having two sub-components: Competence, containing information about Company and Products & Services; and Risk, containing information about Security and Privacy.
Relationship Management refers to interactions with the company over time, both before and after a purchase (Pre-purchase and Post-purchase Interactions).
The MoTEC model was then used to derive a Trust Toolbox containing a suite of concrete tools for designers. The first tool was called GuideTEC and was a set of trust design principles and guidelines. The second, CheckTEC, was a checklist evaluators can use to diagnose the trust performance of a website. Thirdly, QuoTEC was a questionnaire that can be administered to representative users, either after a user test with a facilitator or on its own.
The QuoTEC questionnaire was used to collect user feedback in two different studies. The first dealt with the service industry and examined user reactions to six hotel websites in Switzerland and in the Netherlands. The second dealt with the retail industry and examined user reactions to two computer websites and two online bookstores in the United Kingdom. The main objective of this double study was to reduce the number of items in the questionnaire. The original set of 23 items was reduced to 15 items, while keeping the effect of the reduction on the explained variance minimal.
These studies also uncovered differences in the factors underlying and predicting trust in the two industries. Trust in hotel websites was found to be best predicted by the components Company and Products & Services. On the other hand, trust in retail websites was best predicted by the components Privacy, Products & Services, Company and Usability. This difference was accounted for by the fact that hotel guests will physically stay at the hotel and interact with its staff and, often, only make a booking, while retail customers actually buy goods online from a company they will only interact with online.
A validation study demonstrated that evaluators using the CheckTEC checklist found four times as many problems as unguided evaluators, in half the time. Also, checklist-guided evaluators paid attention to a greater range of factors than unguided ones, who mostly noted factors related to Branding and Usability. Compared with the results from user tests, checklist- guided experts correctly predicted about 90% of all observed problems.
The QuoTEC questionnaire was also tested to compare the results produced after its administration after a user test, with a facilitator, and those produced in a remote evaluation set-up, without a facilitator. The findings indicate some differences in the results that were mostly due to the questionnaire- only participants not systematically following the set scenarios. Building in some controls in the remote evaluation would address this issue by forcing participants to evaluate websites more thoroughly, which would increase the reliability of the results.
Given the differences observed between the hotel and the retail websites, the MoTEC model should be applied to more varied types of industries and validated for each of them individually. This would show which constellation of factors are the most important in each case. As the GuideTEC guidelines were only indirectly validated through the checklist items, future research should examine the effect each guideline makes on perceived trustworthiness. In conclusion, concrete examples illustrate the ethical implications of making websites appear to be trustworthy.