What do the numbers mean on Google Trends?

How to use Google Trends when your problem is too big for their public-facing website

Carrie Fowle

Table of Contents Show

• Using Google Trends at Scale
• How to use Google Trends when your problem is too big for their public-facing website
• What does Google Trends measure anyway?
• What makes working with Google Trends tricky?
• 1. Five Trend Limit
• 2. Relative Metric
• How can I look at more than five trends at a time?
• Can I do this programmatically?
• In summary:

May 25, 2020·4 min read

Google Trends is a powerful measure of public interest in a topic, in other words, its hype; however, the way it is set up can make it difficult to use outside the simplest of applications. In this article, we dive into exactly what Google Trends is measuring and explore how to use it when your question exceeds the five-topic limit of the public facing tool.

What does Google Trends measure anyway?

Google Trends gives you a normalized measure of search volume for a given search term over a selected period of time. From their FAQs:

Each data point is divided by the total searches of the geography and time range it represents to compare relative popularity. Otherwise, places with the most search volume would always be ranked highest.

The resulting numbers are then scaled on a range of 0 to 100 based on a topics proportion to all searches on all topics. Google Trends FAQs

That is to say, for each word in your search Google finds how much search volume in each region and time period your term had relative to all the searches in that region and time period. It combines all of these measures into a single measure of popularity, and then it scales the values across your topics, so the largest measure is set to 100.

In short: Google Trends doesnt exactly tell you how many searches occurred for your topic, but it does give you a nice proxy.

What makes working with Google Trends tricky?

As handy as Google Trends is for quickly taking the pulse of the internet, the structure of the service itself makes larger scale application difficult for two reasons:

1. Five Trend Limit

Currently, the public-facing Google Trends site will not allow a query with more than five terms. This is obviously limiting if you want to explore any real world question. For example, early in the democratic primary you wouldnt have been able to easily compare the popularity of all the major candidates.

2. Relative Metric

The obvious reaction to this limit is, well, Ill just use multiple queries. But as we mentioned earlier, all searches are scaled to the highest volume topic in your query, so two different queries arent comparable if they dont have the same largest topic.

How can I look at more than five trends at a time?

The last statement is the key:

if they dont have the same largest topic

So long as two queries share the same most popular topic, they will be scaled in the same way, so the trends will be comparable. It follows then that if you want to compare more than five topics in Google Trends, you just need to include a control topic in each search.

For example, in a recent project, my team and I wanted to compare the popularity of Netflixs and Amazons content. We went through the list of both platforms best content and pulled each titles Google Trends measure relative to that of the word France.

Exactly how did we decide to use the word France?

Its a bit like a recipe: you want something stable and safely higher than all of the terms you are interested in.

Using domain knowledge, I knew which titles would see particularly high search volume, so I looked for a word that was safely more popular than our most popular titles, but not so popular that Id lose any real information on the less popular titles. For example, Bitcoin has such high search volume that many titles were normalized to zero (see what I mean here).

Can I do this programmatically?

Yes.

For the streaming content project, I used the Python library pytrends, an unofficial API for Google Trends. Below you can find a quick snippet to get you started:

You should note that if you are trying to do this for more than a handful of queries, you will hit Googles DoS (Denial of Service) limits relatively quickly, so you will want to pause your loop for a few seconds between each query.

In summary:

1. Google Trends is a good proxy for public interest
2. To make multiple queries comparable, include a control term that will be the largest in every search
3. Use pytrends to automate the process

With this method, you can now answer big questions like: how popular are my products compared to those of my competitors, did a recent ad campaign impact interest in different product categories differently

In short: the five word limit is no longer yours.