Almost Half of Search Queries Are Repeats

    August 2, 2007
    WebProNews Staff

Forty percent of all search queries are repeat queries from users trying to find information they have found before, according to a new study. But if there has been a change in search result rankings since the last time the query was entered, it significantly hinders the searcher from re-finding the information they seek.

Almost Half of Search Queries Are Repeats
Almost Half of Search Queries Are Repeats

That’s not necessarily a bad thing, according to the study’s authors, hailing from the Massachusetts Institute of Technology, the University of Washington, and Yahoo. It really depends on how much time has elapsed since their first search.

If users repeat a query very soon after the initial search, they are most likely looking for new information, rather than trying to re-find. However, if they don’t search again for a span of days, they either don’t remember exactly what they saw before, or are more motivated to find that exact website.

Sometimes it just depends on how exact they are at remembering what query string they used before, but most of them (around 70 percent) are pretty decent at remembering.

The study seems to affirm that the higher the rank, the higher the likelihood a result will be clicked. This is demonstrated on re-finding missions when a previously clicked result shows up farther down the SERP. It is less likely to be clicked than a previously un-clicked result that now appears closer to the top.

This is in contrast to when the SERP is exactly the same as before. The previously clicked result is more likely to be clicked again.

The authors write:

We found that it was much more likely for a repeat result to be clicked if there was no change in rank: 88% percent of the clicks for overlapping-click queries were repeat clicks if there was no change in rank, while only 53% of the clicks were repeat clicks if there was a change in rank.If the rank of the result had not changed, the second click occurred relatively quickly, while if the rank had changed, it took significantly (p<0.01) longer. Changes to result ordering appear to slow re-finding.

The researchers advise that search engines should be more mindful of repeat queries in the future by tailoring results that are predictive of what searchers are trying to re-find. They suggest that the best way to do this may be by providing software to the end-user that keeps of record of their individual search queries and reproduces either queries or direct links to websites previously visited.

They downplay, however, the practice of using popular results, generated from users on the whole, to influence search results, as search spam could unduly influence them.

"In contrast, personalizing search results based on search history can help avoid potential problems caused by spam."