How Google predicts house prices

Search term data gleaned from Google can now provide an early warning of movements in house prices...

The online oracle of Google is now so firmly planted in public life its verb form is even in the Oxford English Dictionary. If you don’t believe me – Google it.

It’s hard nowadays to imagine life before Google. Where would you have looked when you needed to track down an obscure song lyric, or movie or restaurant? Or when you wanted to take a virtual stroll down a friend’s street and take a virtual look at their house?

In fact, such is the dominance of Google that data concerning what we are all looking for on the search engine can provide an instant and accurate snapshot of what is high in the public mind. And if you know what’s high in the public mind, you can begin to predict a whole host of other things – including house price movements.

Word surges

Research carried out by the Bank of England found that searches for certain words could provide an indication of shifts in wider society. Specifically, the bank found that searches for words such as ‘mortgages’ and ‘estate agents’ tended to peak around a month before rises in house prices, as people began to look at options for buying property.

The research was carried out using Google’s Insights for Search tool – a publicly available program that shows the number and location of searches for certain words at any given time.

In fact, if Google data stretching back to 2004 is considered, the researchers found that search term data was actually more accurate than traditional house price indicators. A key reason for this was the timely nature of search term information as well as the vast size of respondents sampled.

And house prices aren’t the only changes Google can predict...

Unemployment figures

The Bank of England also found that Google search data could help predict changes in levels of unemployment. Searches for words like ‘Jobseeker’s Allowance’, ‘JSA’ and ‘unemployment benefit’ tended to rise with levels of unemployment. This was especially noticeable after the 2008 recession hit and many began to fear for their jobs and so looked into what help was available if they were made redundant.

However researchers did say that regular indicators such as Jobseeker's Allowance claimant counts provided a more accurate picture of the labour market. And indeed, despite the timely and comprehensive nature of search term data, it does still have its flaws. The Bank of England stated that the information was bound to be slightly skewed as most internet users were of a certain age and income group.

Random searches rooted in curiosity, along with the fact that the information only went back a few years could also add a considerable amount of distorting ‘noise’ to the results.

But this research isn’t the first time that Google search data has been used to provide a snapshot of real life.

A snapshot of the economy

Back in 2009 a study into Google search term data concluded that there was a significant relationship between increases in searches for certain terms and surges in certain retail sectors. Specifically, the research found that searches for a particular make of car would spike as sales in the car rose. Searches for various travel destinations would also move in line with tourist trends to the same cities and countries.

For the retailer, this search term data represents the Holy Grail of marketing, providing an instant and wide-ranging snapshot of a customer base’s preferences and attitudes. Using Google’s Insights for Search tool, data concerning seasonal and location-based interest in certain goods and services can accurately be obtained. Retailers can literally ascertain mass data concerning what you're Googling, when you're Googling it and where you're Googling it from.

And this isn’t the only predictive information the search engine stores about you...

Making guesses about you

From 2009 Google began using 57 different types of information obtained from your computer to personalise search results. Everything from your location to what browser you are using is now used to make guesses at what websites you would like to look at.

If you are logged into your Google account, the concise web history of every site you visit and every word you search for will also be used to construct your own personalised results page.

More recently, the search engine giant has combined this technology with Google Instant – the application that shows you website results as you’re typing in your search. The result is a new feature that begins to contact and load pages it thinks you will choose from the results list, before you’ve even clicked on the link.

The development is estimated to shave off between two and five seconds from the overall time it takes to make a search.

Creepy or convenient?

Now, I’ve written before about various ways that technology, and specifically smartphones invade our privacy. And while anonymous mass data concerning search term volumes seems pretty harmless, the more personalised web history and location-based data stored by Google does make me feel a little uneasy.

Yes, I’m sure that by storing this data Google can make our online experience even slicker and user-friendly. But is sacrificing an element of our privacy a reasonable price to pay for a slightly quicker online search? Could seemingly innocent developments like this start to push us down a slippery slope to a Big Brother-esque state?

Let us know what you think in the comment box below.

More: House price indices are a waste of time | Why house prices will rise over the next five years | Why property is profitable