Stop Using Google Trends

This article was originally published on Danny Page's Medium, and has been republished here with the author's permission.

Google Trends is a very interesting product, as it gives us real-time data on how people are using Google. Google is the Address Bar of the Internet, so if you need information on a topic, just type in "Euros" and you'll have the scores and times of every game of the UEFA Euros Championship. Google can then track that interest in a topic and we can see it. But what shouldn't you use Google Trends for? Well, until people start using it appropriately, everything.

Generally, a lot people learned how to search on the internet from Ask Jeeves. Just last week we saw a story about a woman that always searched politely.

It's not a stretch to think that many people's interaction with Google and other search providers are by asking a question, instead of putting in a topic. Here's one such example that Ben Casselman brought up:

Casselman's example gives us insight in how people use Google. With this knowledge, you can either be an asshole and assume that many people didn't know who Mitt Romney is after the election, or perhaps we can more likely conclude that it's a function of people's search habits of researching a relevant topic:

Sadly, a lot of journalists assume the worst.

"But if all of yesterday's frantic Googling is any indication, plenty of Americans don't even know who she is," Christopher Ingraham wrote for the Washington Post in an article titled "If you have no idea who Harriet Tubman is, you're not alone."

Oh. "Frantic". "don't know.". Maybe the real issue is that it's not "any indication?"

Here's why: Google Trends reports search numbers relatively within the date-range and in context of other trends. Here again are those reported numbers, but with wider date ranges.

Note how the Washington Post also said: "Plenty of Americans." What does "plenty" mean here? Remember, Trends is relative. And we can see this with the most recent Google Trends Freaking Outrage (GTFO), like this Washington Post story titled "[The British are frantically Googling what the E.U. is, hours after voting to leave it](The British are frantically Googling what the E.U. is, hours after voting to leave it)."

They note that searches about the EU tripled. But how many people is that? Are they voters? Are they eligible to vote? Were they Leave or Remain? Trends doesn't tell us, all it does is give us a nice graph with a huge peak. More likely, it's a very small number of people, based on this graph that puts it in context with other searches in the region:

But it's giving plenty of people cover to insult the entire country, when it's likely just a few people searching for something in a way that they always search for something. It makes "The British are frantically Googling what the EU is, hours after voting to leave it" absurdly disingenuous without better numbers. Remy Smith points this out: The peak was merely ~1000 people! It's ludicrous that so few people get turned into a massive story, but it underscores the need for context.

I'm disappointed that this is how data is being used, and really drives home the need for people to understand the data before they use it incorrectly. Google Trends is an interesting tool, but please do a bit more research before using it. Beware, you can look quite foolish by solely depending on it.