The current lack of data from Google about the behavior of users has been the greatest frustration for SEOs. Although some previous studies have tried to analyze user behavior on Google, most of them are outdated as SERP features have changed over the past years.
This study intends to fill the gap and shed some timely and novel insights on how users interact with Google’s SERP features.
In this study, we address the following research questions:
Step 1 Survey Design: We provided the survey respondents with a set of Google search queries. In order to provide a broad set of search queries, we settled for the following ones:
You can find the survey questions here: Link
Step 2 Survey Tools and Screen Recording: Secondly, we decided to screen record survey participants to derive useful data points for our study. Our survey tool of choice was surveygizmo.com, as it nicely integrated with the screen recording software addpipe.com. US-Participants for the study were recruited on Mechnical Turk. We also run a pre-screening to make sure respondents were familiar with the instructions. The survey instructions can be found here and a sample screen recording video can be watched here.
Step 3 Data Annotation: Thirdly, to consider only valid responses for the data annotation task, we reviewed each of the videos individually to check whether instructions were followed properly. Of the 454 submitted videos, 259 passed our screening. We then annotated the remaining 259 videos and entered each data point into a spread sheet. We collected data with a total recording time of 2276 minutes (38 hours).
Respondents provided information about their use of search engines.
We observe that Google is most commonly used by 94% of users. Bing is the biggest competitor with a much smaller share of 2.7%.
76% of users use Google search multiple times a day, and 94% use it at least once every day.
We measure for each question what percentage of users used the search suggestion. We find that users use them between 18% and 31% of the time, globally 23% of the time on average.
Using the video timestamps we can assess the time when users click for the first time. We find that the on average a user clicks first after 15 seconds and that half of the users click between 5 seconds and 17 seconds after getting on the page.
The average time to first click, when we consider questions independently, varies from 10 to 25 seconds.
On average 9% of users reach the bottom of the page, as we can expect this varies however significantly depending on the query, the most specific queries leading to higher percentage. We find that to find a car phone holder, as little as 3% of the users reach the bottom, while 17% (almost 6 times more) reach it when searching for supplements to treat back pain.
In the majority of cases (85%), we find that users don’t change their search query at all. 11% change it once and only 4% change it 2 times or more.
Modifying one’s query or not depends a lot on the nature of the search. Our data shows that the percentage of users who change their query can go from under 5% to more than 35% when trying to find supplements for back pain.
We measure how many times searchers clicks a result, then bounces back to the search results page and chooses a different result. We find that overall the bounce rate is small, 83% never bounce while only 5% bounce more than once.
Depending on the question, this percentage of users who bounce at least once varies in the of 10% to 30%.
We find that most users visit only one page, but for some specific queries the average can go above 2.
We find that most users click on organic search results at least once. However two questions stand out, few users click on organic search results when looking for a car accident lawyer or how to cut one’s hair.
We find that users click on Google Adwords links or paid listings 19% of the time overall but depending on the query this can go from 0% to 30%.
We measure how often users click on video links. Our results are very polarized here because for a single question around 95% of users clicked on a video result while in all other cases virtually no one did.
We find that users click on Google Maps or local listings 40% of the time when the query is about finding a car accident lawyer, and very few times otherwise.
Additionally, our data indicates that when the user clicked on Google Maps or local listings, their chance of clicking on one of the 3 first listings is 97%.
Almost 20% of users click on Google Shopping results when the question is about finding a car phone holder, and 5% of them click them when looking for back pain supplements. In other cases virtually no one clicks on Google Shopping results.
Almost 15% of users click the “people also ask” box when the question is about finding supplements forback pain. For other queries this varies from 0% to 3%.
We find that very few users (0.5%) click on the second google search page or further. Looking at the distinct questions, we see that users looking for supplemenetation to treat back pain are the most likely to reach the second page or further (1.5%).
The average search is computed in 76 seconds, and we find that half of the searches are completed in 30 seconds to 90 seconds.
During an average search a user clicks 7 times on average, in our dataset is computed in 76 seconds, and we find that half of the searches are completed in 30 seconds to 90 seconds.
We observe some important correlations between some variables of our study. In particular the number of modified queries, the number of bounces, and the number of seconds to complete a search show correlations from 0.7 to 0.9.
This may indicate that users faced with a challenging Google search try different strategies.