Aytraking: how well the link is made

One of the most important issues that need to be addressed in usability research site, is how well made links. If a visitor does not notice that you have a link on your site, say, to an online store (which is quite common in our practice), then, most likely, it will not buy anything in it. And if you notice the link, but do not understand that this is a store (and such cases are also not uncommon), the result, again, will be disappointing ...

Being engaged in usability research with the use of ITracking, we have developed a method that allows you to quickly and visually assess the quality of links. Of course, this method has a number of limitations, but with proper use it allows you to very quickly identify areas to which additional attention should be paid.

So, to start a certain task. If we want to determine, for example, how well the link to the store is made, the respondent is offered to buy something on the site. And we begin to track how the respondent solves this problem.
')
Note. It should be borne in mind that the quality of the solution of the problem (to buy something on the website) and the quality of a specific link are, in general, different things. For example, a link to a store may not be one or a visitor instead of going to the store will click on an ad that leads to a specific product. In this article I will talk about the link and how to evaluate its perception by the visitor.

As illustrations of the following text, maps of views from the testing site of the Computer World store will be used. And we will evaluate the link "catalog of goods."

So, in the testing process, the following options are possible:
1. If the respondent did not look at the link, then he did not see it.
2. He looked at the link and clicked. So, oddly enough, he saw.
These are the two most simple options, which, in general, do not require special analysis ...

A map of views, which shows that the respondent did not look at the link under study and immediately entered another section.	The map of views, where the respondent saw the link and passed over it (fixations, starting from the 12th, were made after clicking on the link).

The third option. The respondent looked at the link, but did not click, but continued to examine the page. Possible options:
3.a did not notice (random fix)
3.b noticed, but assessed as wrong (I did not understand that this is a link, did not guess where it leads, etc.)
3.v noticed and “put it in my memory”, but decided to look for a more suitable one.

Option 3.a is fairly easy to determine visually - fixation in such cases is single and meek, as a rule, up to 200ms. In most cases, the direction of movement of the gaze at the same time practically does not change - the angle between the previous and subsequent saccade does not exceed 20-30 degrees. The return of a glance to the link either does not occur (option 1 - we can assume that the respondent did not look at the link), or it happens at intervals of more than two or three fixations. For control, at the end of testing, you can lose the session of the respondent and ask him to comment on his behavior.

A typical example of a random fix is ​​that the respondent looked at the link but did not pay attention to it.

Options 3.b and 3.c are visually similar — several consecutive short fixes in the region of the link, or one or several long 200-400 ms. In case 3.b, as a rule, there is a repeated return of the look to the link, in case 3.b the return is single, with a subsequent click (or the return does not occur if the other link found is perceived by the respondent as more appropriate). Also, in cases when the respondent is not sure of the correctness of the link, before a click, as a rule, there are multiple fixations on this link and, in some cases, “throwing” a glance between the two found links.

An example of a map of views when the respondent rated the link as incorrect.	Part of the page view by the previous respondent - only the piece of interest was left. The first viewing is fixation 2, then viewing other links, returning to the area of ​​interest — fixing 6, 7, 9 and leaving the area of ​​interest.

An example of a map of views on which the respondent is looking for a more appropriate link.	Part of the page view by the previous respondent - only the piece of interest was left. The first viewing is fixations 2 and 3, then a search for more suitable links, then a return of a look — fixations 27, 29 and several subsequent ones (removed to avoid cluttering the map) with a subsequent click.

Of course, such a “visual” analysis of aitracking data is not the only tool, but it can be very useful in identifying “hidden flaws”. In addition, its simplicity allows research customers to independently evaluate the functionality of the design, simply by viewing videos from test session records.