The difference in the perception of sites

In this post I want to try to show how the perception maps of sites vary depending on the target groups, and try to build the most inadequate map.



Some time ago, our company - www.eye-tracker.ru - conducted a test on the main page of the Russian-language site of the famous CMS Drupal - www.drupal.ru The task of the test was to find a link to the demo site of this CMS. As test subjects, two groups of 5 people each were taken - “computer scientists” and “managers”. Both groups were not related to the sites and did not work with CMS.



Testing was carried out using an eye-tracker , a device that tracks the direction of the subject's gaze. That is, in the process of testing it was determined how the subject studies the page and where he searches for the information he needs.



The result is the following perception maps:

All subjects	Computer scientists	Managers

')

But further we were interested in the question, is it possible, by combining different subjects, to create a perception map that would be radically different from the above? Since there was no desire to do combinatorics and check all 250 options (for cards of 5 users) (and it’s quite difficult to cite them in the article), we made maps of individual users and tried to select “non-standard” visually. So:

Respondent 1	Respondent 2	Respondent 3
Respondent 4	Respondent 5	Respondent 6
Respondent 7	Respondent 8	Respondent 9
Respondent 10

As respondents with differing ideologies, three were chosen: the third, the seventh, and the tenth. A map of their combined attention was as follows:







But since testing usually takes place on five or more respondents, we tried to create an “inadequate” attention card, adding two more. The most inadequate cards turned out when adding respondents 2, 6, and also 8, 9.

Respondents 2, 3, 6, 7. 10	Respondents 3, 7, 8, 9, 10	All respondents

One can clearly see that even consciously constructed “maps of inadequate perception” closely coincide with the generalized one, which largely echoes the famous article by Jacob Nielsen. Five users are all that is needed for the test , and is also confirmed by our observations.