Does Google Have a Crystal Ball That Lets You Predict the Future?

Google may indeed have a crystal ball that lets you peak into the future. It may help you predict things like whether there’s a growing (or dying) market for a product or service. Or who is going to win a political race. And who knows what else. It may be able to do that faster and cheaper than other tools you might be using.

But if you want to know if it’s going to rain on this weekend’s picnic or which pony you should put your money on at the racetrack you’ll have to look elsewhere. The financial markets or a particular stock? Well, let’s think about that later.

I’ve spent a good part of my professional life helping people plan their future. A lot of that time the last few years has been focused on retirement planning. A few years ago, Google created a tool it calls Google Trends. It is the tools behind Google Flu Trends, which Google says is able to estimate flu activity up to two weeks faster than traditional systems. Basically, Google uses its vast database of individual searches to look for trends. Here is it’s explanation of how it works with the flu.

We have found a close relationship between how many people search for flu-related topics and how many people actually have flu symptoms. Of course, not every person who searches for “flu” is actually sick, but a pattern emerges when all the flu-related search queries are added together. We compared our query counts with traditional flu surveillance systems and found that many search queries tend to be popular exactly when flu season is happening. By counting how often we see these search queries, we can estimate how much flu is circulating in different states and countries around the world.

The same idea might apply as a tool to predict other trends. And trends, my friends, is just way of saying that people are buying more or less of whatever it is we have to sell – whether it is a widget, a service or a political candidate.

Let’s see what Google Trends can tell us about the future of retirement planning. Once you see the general concept, you can apply it to consider the business or market you’re in, the things you sell or the ideas you promote.

Go to the Google Trends home page. Type in the phrase ‘retirement planning’ and you get a graph something like this.

Ignore the letters. They are for references to news items but they’re not important right now.

When I first saw the graph, I had two questions. I could see the trend was on a gradual downward slope. But what do the numbers mean? That is, what’s the difference between the graph points in early 2000 that are close to the 2.00 line and the graph points in late 2008 that drop down to about 0.34?

My second question was, “So what? What does this mean and can I use it to ‘predict’ what my clients and customers are interested in right now – and will be interested in for the next little while?”

What does the graph show?

The graph covers, by default, a specific time period and geographic area. The default time is January 2004 and the present – currently that’s the last 5½ years. The default geographic area is worldwide.

You can change the time so it covers just the last 30 days or the last 12 months, a specific year between 2004 and 2009 or even a specific months during that time.

You can limit the geographic area to a specific country and within a country a specific state or province. In some cases you can further limit it to a specific city – assuming there’s enough data. My experience is that unless you’re using a very popular search term, there’s not enough data to create graphs for individual states much less cities.

But let’s get back to our graph on retirement planning. Anyone can see the line is headed downward – though it does flatten out in 2009. To understand how big those changes are, you must understand how Google scales the graph.

I’m not a mathematician, so I found an example helpful. From what I can tell, this is how Google creates the trends graph.

Google first calculates the average number of searches for ‘retirement planning’ during the period you’ve selected. Let’s say the number is 1,000. I don’t know if they use the average number of search each hour, each day or each week. But it probably doesn’t matter as you’ll see in a moment. 

That average becomes the magic number Google uses to create the graph. It becomes the 1.00 on the graph.

Let’s see how that ’standard’ affects the rest of the graph.

During the week of February 29, 2004, let’s say that 2,000 people thought about retirement planning and went to Google to get information. The number 2,000 is, relative to our average 1,000 searches, twice as many. Therefore, we set the graph point for that week to 2.00. During the week of December 21, 2008, people only searched for ‘retirement planning’ 340 times – they were obviously more interested in Christmas and were not thinking about retirement. We set that graph point to 0.34 because it is 34% of our standard 1,000 searches.

What do the numbers on the graph mean?

If you look at the graph, it seems to tell us that, compared to the entire 5½ years, people were twice as interested in retirement planning in early 2004. Now, in 2009, they are only about half as interested. That’s quite a decline.

But can you look at that trend and conclude that the market for retirement planning is drying up? Is this just so much hocus pocus? In a moment I’ll talk about the assumptions I think you have to make to conclude this might be a useful predictive tool.

You may want to test the concept yourself. I thought it would be interesting to look at a couple political elections.

Here’s the graph showing Google searches for Obama and McCain in 2009 in the United States. Obama is the blue line. McCain is the red one.

Notice how the gap between the two started increasing in early October and just blew through the roof about election time? Of course, maybe people were just looking for dirt on Obama so they could justify not voting for him. Uh huh.

Just yesterday Virginia had its primary election for governor. The democratic race was hotly contested. Yesterday, the morning of the election, the newspaper headlines said “Governor’s race goes to the wire.” This morning, the day after the election, they said “Deeds wins in rout.”

Here’s the Google Trends graph for the last 30 days for the three democratic contestants in the 2009 Virginia governor’s race.

Hmm. Maybe there is something to this.

Can this be a useful tool?

Before you give too much weight to Google Trends, you probably should carefully consider the assumptions needed to conclude that search behavior on the internet can be used as a crystal ball to predict the future.

Here’s how I see that analysis. I’m not sure these are all the necessary assumptions, but I am sure (well, pretty sure at least) that if any of these assumptions are wrong, the usefulness of Google Trends as a predictive tool is significantly weakened.

• Most people think before they act. In other words, most people don’t just do something out of the blue.
• Most people spend their time doing things they find interesting or that they are concerned about. Likewise, they don’t spend time on things that are not interesting or that they are not concerned about.
• Most people like to have information before they make a decision. I’m not saying they get the right information or analyze it correctly. I’m just suggesting that people tend to look for information before they do something about whatever is on their mind. Likewise, people do not go searching for information on subjects they are not thinking about.
• The more interest people have on a particular subject, the more likely they are to do something. The less interest they show, the less likely they will do anything related to the subject or topic.

Now the big assumption.

• Google has become the primary way people look for information. Notice I’ve specifically not limited that statement to “look for information on the Internet.” I think people are now using Google to get information more than reference books, magazines, newspapers or any other source of information.

If that’s true then the use of Google Trends will have less value in predicting trends that shape markets, political races or anything else. The more true the statement is, the more value Google Trends has.

And the big conclusion.

• If you can see an increasing or waning interest in a particular topic or subject or concept or idea or product as shown by how often people turn to Google to get information about it, you can also predict the market for it. You can predict trends and you can predict consumer behavior of groups of people (but not necessarily individuals in those groups).

The Future of Retirement Planning

If all that is true, consider what Google Trends tells us about the future of retirement planning.

• People are doing fewer and fewer searches for information about retirement planning.
• Therefore, that suggests they are thinking about retirement planning less often.
• Therefore, they are less likely to do anything about retirement planning.

Conclusion? People are less interested in retirement planning than they’ve been over the last several years. Overall, they are less likely to do anything related to retirement planning. While the decline has leveled off, it is still much less than what you’d expect given the financial crisis of the last year. Wouldn’t you think people would be more concerned after seeing their retirement accounts decimated? But that doesn’t seem to be what’s really happening.

Final Thoughts – And Some Questions for You

Google Trends doesn’t explain the ‘why’ behind what it is seeing – it only tells us the ‘what’. Why has the interest in retirement planning declined so much the last few years? Why did it not jump back up significantly in the last year? What are people thinking? Are they so numbed what has happened that they’ve given up hope? Is there other evidence that contradicts the trend Google suggests? Is ‘retirement planning’ the wrong term to analyze? Are there other terms that might suggest a different conclusion? And, generally, are their particular topics or subjects or products that Google Trends would be more useful in predicting behavior? Others where it would seem to be less useful?

We’ll hold on to those questions and consider them for a future article.

Meanwhile, I’d love to hear your thoughts. Leave a comment. Or email me. What do you think?

Related articles from WalterBristow.com:

1. Did Laker Fans Use the Internet to Find Out Where to Join the Riot?
2. Just-in-Time Mind Reading with Google Hot Trends
3. Keeping Track of Yourself on the Internet – The Starting Point of Personal Branding
4. Google Voice – It Will Change How You Use Your Phone
5. Copernic Tracker: The Lazy Man’s Way to Track Changes in Web Pages