user's gaze 과 Naver integrated search vs. Google Web Search - Comparison of user's eye tracking results, I also send my opinion while reading these two articles.
Google's eye tracking data Search result page eye tracking and blog post title It seems to be old data used by . ITools sold a report that said that the Google search results were an inverted triangle for the first time.
What is Heatmap?
There are several methods of eye tracking analysis, and the published result looks like a heat wire, which is called a heatmap.
The heatmap is color-coded with high frequency. A hot color is more frequent, and a cold color is more frequent. So red is the highest. Good interaction design pattern.
Since the heatmap is a color-coded frequency, the result is different depending on which frequency is color-coded.
Yahoo! The search result pages of Google and MSN were made by iTools, Criteria for a heatmap in iTools is the percentage by participants. That is, how many people have seen it. You can say that the red in the picture is where all the participants saw it.
However, Naver search results and heatmap results do not seem to indicate how many people have seen the place. Because Naver did not outsource iTools and did it directly using Toby's Clear Viewer. Toby's Clear Viewer does not draw the places that many people have seen so beautifully, but draws them in a circle.
This is the content I have been requesting for about 2004 years from 4 to the present, but it has not been done yet. This month, there are 5 people (?) from all over the world, and even writing the documentation for the beta test, and even the latest beta version that is being tested still can't draw it pretty.
Without a complicated story, the heatmap on the Naver search results page is not a place that many people have seen, but a place that has been viewed many times or has been viewed for a long time. This is a heatmap drawn based on fixation count or fixation duration time. Therefore, when interpreting the Naver data, you should interpret that reddish spot as a place that everyone has seen, not a place that everyone has seen, but a place that has been seen many times or has been seen for a long time.
Which Search Results Pages Are Effective?
If it was a search task instead of just looking at it, the most inefficient thing to do when looking at the heatmap above is the Naver search result.
I don't know how much fixation duration time is in total, but looking at the picture alone, a lot of the area is red, so I've been looking for a long time or several places to find the search results. I don't know whether the heatmap criteria for Naver search results is fixation count or fixation duration time, but you can see that you are searching the entire page to find the search result you want because your gaze is all over the page until you click it.
The intention to see the entire page may have succeeded. However, it is quite inefficient for users.
And the heatmap does not only show the result that the gaze of the search result page is an inverted triangle. What is an inverted triangle? Is it important to put it on top of a folder? There are important findings in the search, which seem to be missing.
Web Page Search vs. Search
There is also the issue of whether it is a single source of web pages like Google or Yahoo! or an integrated search that organizes several things into modules.
Korean integrated search groups the source of search results first. So, the module of the source is bundled with correlation and placed. If you arrange by module rather than correlation of search results, the integrated search results will become terribly long.
In the end, users find search results in groups grouped by data source, not at the end node of the search results, so there seems to be more areas to look at like the Naver search result page eye tracking heatmap.
In terms of search results, it is managed by module, not the correlation or accuracy of the end node of the search result, so the scope is wider, but the cost of finding a user can increase.
I would like to see if the search results are listed by correlation of search terms, not by data origin. If you want to search by ethnicity, you might be able to click on something like a tab. If we were to look at the ingredients from the beginning, we would start from such a place. But many people just do a search and expect to find what they need in the results. If it is not grouping by origin, I think it would be better if the search results end node is listed by relevance.
Have you ever used Zeroboard's integrated bulletin board search function or functions that allow you to search all bulletin boards of Zeroboard at once? The search results do not come out by correlation, but are distinguished by each bulletin board, and there is a correlation therein. That is, grouping by source first. In fact, it is because I searched for each search DB for a simple implementation. Anyway, I have to look at the posts to find what I want, but I have the inconvenience of having to check which bulletin board first. I think that Zero Board's integrated search for bulletin boards is similar to that of Korean portals.
The purpose is to find posts, but I don't think it's very effective for people who search for them to be divided by bulletin board.
Regardless of which algorithm is used, users should be able to find the search results they need regardless of the source, just like the eye tracking data of Yahoo! or Google search results pages.