Last Update: 03.07.2020
Google Trend Data can show the interest of a specific word over time on Google. The official website of Google Trend is https://trends.google.com/trends/?geo=US. This article will provide you a basic understanding of how Google calculates this data and how to use it with Python.
Researchers can use it to study the investor’s interest in the company (Drake, Roulstone, and Thornock (2012)), its products/brands (Peng-Chia Chiu, Siew Hong Teoh, Yinglei Zhang and Xuan Huang(2018)).
Understand the data
We now use the keyword “Microsoft” as an example. From the picture above, we can see how Google Trend represents people’s interest in the word “Microsoft” for the past 12 months. Google provides a search volume index rather than raw search volume.
“Interest over time” represent search interest relative to the highest point on the chart for the given region and time. A value of 100 is the peak popularity for the term. A value of 50 means that the term is half as popular. A score of 0 means there was not enough data for this term.
From the picture above, we can find that “Microsoft” attracted more attention in September 2019 (Interest=100). There was no significant upward or downward trend for the past 12 months.
“Interest by region” helps you see in which location “Microsoft” was most popular during the specified time frame. A higher value means a higher proportion of all queries, not a higher absolute query count. So a tiny country where 80% of the queries are for “Microsoft” will get twice the score of a giant country where only 40% of the queries are for “Microsoft”.
As we can see, “Microsoft” was not very popular in Africa for the past 12 months. But that does not necessarily mean that Microsoft was not successful in that area. To understand the reasons, one needs to obtain more information about the culture of the region, the technology available, the economy and the events going on.
“Related Queries”: Users searching for “Microsoft” also searched for these queries. “Top” stands for the most popular search queries, whereas “Rising” stands for Queries with the biggest increase in search frequency since the last time period. Results marked “Breakout” had a tremendous increase, probably because these queries are new and had few (if any) prior searches.
The most popular search queries for “Microsoft” is “Microsoft forms”, a new member of office which can be used to make the evaluation forms, online exams, and customer feedback.
“Related Topics”: Users searching for “Microsoft” also searched for these topics. A topic can contain multiple queries.
A rising related topic for the search of “Microsoft” is “Playstation”, which is an online game store developed by Sony. Microsoft X-box and Sony Playstation are competitors in the game industry.
An interesting function of Google trend data is the comparison. By adding comparison, you can find which term is more popular. We added “Google” in our example. The maximum number of comparisons on Google’s official website is four.
Note: For users who need more than four comparisons, they can use one of the objects as an index, then compare it one by one. By combining the results, the user can obtain the relative popularity of all objects.
It seems Google is much more popular on Google Trends. But no surprise: the fans of Microsoft might use Bing instead!
In addition to the function mentioned above, people can also customize their search by editing the following settings:
- Location ( Worldwide, countries…)
- Time ( Hours, Days, Weeks, Months, Years…)
- Categories (Finance, Game…)
- Search Type (Web Search, Image Search, News Search, Google Shopping, Youtube Search)
Use the Data by Python
A very popular python API of Google Search Data is pytrend, which is built by John Hogue and Burton DeWilde.
Although the data can be obtained more easily by pytrend, Google sets a limitation rate for the download. According to the discussion on varieties of forms, the limitation is approximately 360-400 requests per hour. Therefore, if you are requesting a large volume of data, such as S&P 1500 from 2000-2015, you might need additional adjustment.
A sample python code for downloading data in larger volume can be found in the CoLab link below:
Sales Count: [mycred_content_sale_count]
- Peng-Chia Chiu, Siew Hong Teoh, Yinglei Zhang and Xuan Huang. 2018. Using Google Searches of Firm Products to Assess Revenue Quality and Detect Revenue Management, https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3252314
- Drake, Roulstone, and Thornock. 2012. Investor Information Demand: Evidence from Google Searches Around Earnings Announcements, https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1475-679X.2012.00443.x
 All the explanations are obtained from Google Trend website. https://trends.google.com/trends/explore?q=Mircosoft
 If you need access to the code, or have problems in using the code, please email firstname.lastname@example.org