Last week, a post on the Predictably Irrational blog described the differences in what boyfriends and girlfriends were looking for their beloveds to do based on Google Suggest. Google Suggest can provide hours of fun. Just spend some time at Autocomplete Me or QuestionSuggestions. But can it provide useful audience analysis for understanding our customers, building products for them, and marketing to them? We’ll take a look at that next, but first, let’s see what Google has to say about men and women are really thinking. I should warn you, not all of these search queries are safe for work, but no pictures — only text!
First, the Predictably Irrational findings:
Does Yahoo agree? Interestingly, Yahoo has similar results for what one might want their boyfriend to do, but no results at all for what one might want their girlfriend to do:
As noted in more detail below, the Google AdWords Keyword Tool may provide more accurate findings. So what do we find there?
Not exactly the same, but pretty close. Via Google Insights for Search, we find that searchers are looking for information about their boyfriends only slightly more than those looking for information about their girlfriends, and there’s no real seasonality component. Until we get to last week, once assumes as a result of all the interest drummed up by the blog post.
What about what men and women like? Amazingly, the queries are almost exactly the same, with almost exactly the same volume.
Maybe men and women understand each other more than we think.
Is Google Suggest Useful for Market Research?
But back to usefulness. A number of people in the comments to the blog post wondered:
- “A really clever use of a product for research purposes.”
- “Google is becoming a new metric of social behavior studies.”
- “A fabulous means for research”
- “There must be a way for someone to do marketing research using Google auto complete. Interesting, I may try it one day.”
Of course, those of us in search marketing have been singing the praises of search data for better customer understanding for a long time and use many keyword research tools to gain this insight.
Here at Search Engine Land, we have lots of great information on keyword research tools, including:
- Search Illustrated: Keyword Research
- Keyword Research Resources
- Google’s Search-Based Keyword Tool
- Google Insights for Search (Now With Forecasting)
- Google Sets and Squared
- Using Search Data To Measure Brand Recognition
But is Google Suggest reliable? Again, someone in the comments wonders:
has google actually said that this is a list of common queries? people assume that, but it sometimes comes up with suggestions that are clearly not common search queries. you have no idea what this data actually is or how it has been filtered. hence this is an amusing hack but a terrible research tool. there is a temptation for people to believe they can tap into the power of the oracle with hacks like this. it looks like a numerologists pretending to be mathematicians.
Yes, we actually do know quite a bit about the search data the engines provide and in fact, this data can be awesome for research. But it’s true that Google Suggest itself may not be the best source of that data. Danny Sullivan pinged Google last month about how Google Suggest lists are constructed and was told that:
Google suggestions are based on aggregate data including popular searches that have been entered on Google over time. In addition, Toolbar shows queries that a user has typed before, which are retained on the user’s machine. It is perfectly normal for suggestions to appear for a short while, stop appearing, and then start appearing again.
He concluded that other sources of search data (such as the Google AdWords Keyword Tool and Google Insights for Search) are likely more reliable.
Google’s Stance on Search Data and Prediction
Giorgio in the comments points out that “Google is very aware of the possibilities their data mining has in research”. And of course, they are. They write about this a lot. Last April, they looked at economic indicators and summarized their findings in a paper called Predicting the Present with Google Trends.
We find that Google Trends data can help improve forecasts of the current level of activity for a number of different economic time series, including automobile sales, home sales, retail sales, and travel behavior.
And in August, they noted (armed with a new paper):
“An understanding of search trends can be useful for advertisers, marketers, economists, scholars, and anyone else interested in knowing more about their world and what’s currently top-of-mind.”
A Wall Street Journal article way back in 2007 noted that:
As part of an experiment, Google analyzed search-query volumes related to movies released in 2005 and compared them with opening weekend box-office revenue for each movie. The company found that it could predict with 82% or higher accuracy based on consumer search activity as early as six weeks before the opening whether a film would top $25 million in receipts its first weekend.
Someone else wonders “These results change over time and I’m thinking what might be a really interesting research project is to collect the results over a period of ~1 year to see if world events / seasons / holidays have a significant effect.” As it turns out, you can see that topical events and seasonality do indeed impact search volumes, and you can see the results for yourself at Google Trends and Google Insights for Search.
If you’re looking for even more seasonal data, Dennis Goedegebuure just blogged about seeing this trend in web analytics.
The search engines have an amazingly amount of data about searchers that they make available to us. And while they may not be able to tell us exactly what our significant other is thinking, they can come pretty close. Especially in December.