The labels for the web sites are going to be designed using several methods. First of all some labels are defined using conten authors’s knowledge of the content that the web site is going to contain. According to Morville & Rosenfeld this method is useful for finding candidates for labels but not the final labels. This I because the content authors are professional indexers and don’t necessarily speak the language of the user (2007, p. 105). The labels defined by the content authors are tested with users by using an open card sorting as described earlier. The data from the card sorting are analysed and held together with log data from the former web site. More specifically the log data are analysed using Search Log Analysis (SLA) following recommendations from the article by Jansen, Understanding User – Web interaction via Web Analytics (2009). SLA has a grounded theory approach where theories are grounded in observations from the real world rather than hypotheses. This means that the method has an inductive approach as well as the DBR method (p. 65). Using the knowledge from these three methods I as a designer define the final labels in collaboration with the content author.
When the web site is launched and has been running for about a month it is possible to test whether the instructional goals has been fulfilled. As described earlier the instructional goals are: 1) the users should be able to find information that they are looking for and 2) be inspired to explore the web site and find additional information of interest. Both goals should be fulfilled leaving the user with a feeling of having had a good experience where their individual wishes have been taken into consideration.
A useful method for doing this is web analysis. When doing a web analysis Jansen is describing several metrics (2009). There is a lot of useful and interesting metrics that directly or indirectly tell something about goal 1) and 2). In the following I am going to describe some of the most obvious ones.
Keyword Analysis tells us what keywords the user are searching for when finding our site. By looking at these keywords in relation to the labels it tells us whether we have succeeded in speaking the language of the users when designing the labels. Top pages is a metric that shows which pages of the web sites that receive the most traffic. By looking at this it is possible to analyse whether the pages has been sorted so that the post popular pages are the ones that are easiest to access. It is also possible to get an indication of whether we have succeeded in designing the page so that it suits the different target groups. E.g. if the main menu Handikapvenlig has very low traffic compared to the other main menus maybe the disabled people doesn’t feel that the web site design suits them. Visit length is an important metric as it tells us whether the user is spending time at the web sites or just entering and leaving right away. This information can be combined with another metric Referring URL which tells what sites have directed traffic to the web site. My combining these two set of data it is possible to see which sites that are referring the most time-consuming users. E.g. does a site for disabled users refer a lot of users that spend less than five seconds on the site? If this is the case, it could also be an indication that the design is not appropriate for the segment of disabled people. The last metric that I am going to mention is Visitor path. This gives us information about how the user navigates through the web site and can give an indication of whether the user finds what he/she is looking for.
Even though web analysis provides some valuable data it has some shortcomings. Jansen stresses that “These shortcomings include failing to understand the affective, situational, and cognitive aspects of system users.” (2009, p. 51). He therefore recommends that the method is combined with other data such ad surveys and laboratory studies (ibid. p. 51ff). In the case of testing whether the instructional goals of the web site for Feriecenter Slettestrand has been fulfilled it would be ideal to combine the web analysis with a usability test followed by an interview or a post questionnaire.
This post is going to be updated with reflections upon the interpretive framework used in this design...