Monday, 28 October 2013

Google Trends

As any investor with a casual knowledge of the ASX would know, internet stocks such as (CRZ), Seek (SEK) and REA Group (REA) have been great stocks to own. If you owned REA for the last 10 years, you would have had a total shareholder return of 55.2% per annum according to Commsec, or in other words, a dollar invested in 2003 would be worth 81 dollars today. These phenomenal results are largely due to the network effect: buyers want to go to the website with most amount of X being sold, and sellers want to be on the site with the most buyers, thus creating a self reinforcing cycle for the most popular website. It seems that because of the power of the network effect, these kinds of markets gravitate towards being a monopoly or duopoly over time, the end result being that the winner takes all.

For these internet based stocks, getting pageviews is crucial to riding the network effect, and it is my hypothesis that since a major source of traffic comes from Google, having access to Google's information should provide a good indication of the general trend in website traffic, and therefore, provides clues to the future performance of the business. However, it must be noted that there are other important factors that influence whether views will translate into revenue such as the amount of time each visitor stays on the website, and that many of these businesses such as Carsales branching into different domain names in foreign countries so they are not solely reliant on their flagship domain anymore. In addition, mobile apps are increasingly playing a larger role in driving consumer traffic, so perhaps the importance of Google to these business is declining. With those disclaimers aside, I thought it would be an interesting exercise to compare some of the changes in website interest through Google Trends, with some more in depth information from Alexa.

According to Alexa, 23.8% of the traffic going to comes from Google, and of this, the top keywords searched for were 'carsales', 'car sales', and ''. It is ranked the 53rd most popular website in Australia and visitors spend an average of 9 minutes and 52 seconds on the site. The two largest competitors to Carsales in Australia are Trading Post (which also sells caravans and boats, amongst other things) and Carsguide. To keep things simple, I've limited the keywords to 'carsales', 'carsguide' and 'trading post'.

As the above graph clearly shows, Carsales has experienced a huge increase in interest since 2004, which is reflected in its business fundamentals, while Trading Post has suffered a continuing decline starting from around 2009. Poor old Carsguide is a very distant third place. Further compounding the woes of Carsguide and Trading Post is the fact that visitors only spend 3 minutes, 28 seconds and 3 minutes, 40 seconds respectively on each website, which would significantly reduce the number of sales produced for each visit compared to Carsales. However it seems that even interest in Carsales has plateaued and started to decline - I wonder whether this waning interest will see their domestic revenue growth start to diminish, or maybe it suggests a broader slowdown in the car industry.

Taking a look at the job market, Seek is the dominant player in the space and the 26th most popular website in Australia. 24.1% of traffic originates from Google, and visitors average 8 minutes, 8 seconds on the site. The rise and dominance of Seek when viewed against its competitors looks quite similar to Carsales:

Again, Google search interest in Seek hasn't increased since the start of 2011, but unlike Carsales, Seek is facing the threat of LinkedIn which is gaining traction in Australia. Although LinkedIn is a social network more focused on people in professional occupations, the size of this global business and the more social approach has got the Seek CEO, Andrew Bassat, quite wary. Even with the competitive advantage of the network effect, leaders such as Seek can't afford to become complacent lest they end up with the same fate as the newspapers (who've failed to gain a foothold with Fairfax Media Limited's MyCareer and News Limited's CareerOne joint venture with US company Monster).

While I can't find a breakdown of revenues for the branch of Carsales, Seek is nice enough to show 'look through' revenue for each of its segments. The Seek domestic revenue from FY2011 to FY2013 provides confirming evidence for the link between flat Google search interest and flat revenues.

The usefulness of tools such as Google Trends and Alexa don't just apply to these kinds of online businesses. In this day and age, driving visitors to your website is an important goal for most companies, and investors can gain insights as to what is gaining popularity and what is not, long before it is reflected in any financial statements or recognised by the market. I recall reading about an American who noticed that a particular Hollywood film was very popular, and after researching the production company behind it, concluded that the financial benefit from the movie would be very significant compared to the small size of the company. He made a ton of money out of it when the market eventually realised this, and has continued to do the same in similar situations when he recognises a growing trend that others have yet to discover. One could gather similar insights through Google Trends about movies, toys, restaurants etc. Or, you can just play around with it out of interest - contrary to popular belief, dogs are actually searched more than cats!


  1. It's worth considering that at some point people don't need to search on a term like "car sales" because they are already aware of the URL name. So it could be said that a drop off in search on "car sales" is an indication of URL awareness, and not an indication of website traffic.

    1. That's an excellent thought, however, it is probably difficult to verify unless you have access to more in depth data. In the case of, I note that according to Alexa, the percentage of its visitors coming from a search engine has stayed at approximately the same levels during the past two years. This suggests that the trend of Google search popularity may still be a good indicator of general website traffic. It is of course still possible, and I would say likely, that URL awareness has increased as the proportion of visits from sources excluding Google may have been cannibalised to some extent by direct URL visits.