Predicting company revenue using Google Insights For Search

Chief Nut By Chief Nut, 16th Jan 2010 | Follow this author | RSS Feed
Posted in Wikinut>Money>Investing

The ability to query Google's massive store of historical search query information offers many possibilities, one of which is accurate prediction of online company revenues.

The basics of Insights For Search

Readers unfamiliar with Google's Insights tool should first read my Insight guide for beginners.

Essentially it is a tool that allows you lookup specific search terms and see their search volumes over time. By entering multiple terms you can see their comparative volumes.

Be aware that Insights does not return actual volumes, just relative values. So if you want to know the exact number of searches for term you’ll not get an answer. But if you need to know which of 2 terms is more popular, or compare the volume of a specific term over time, Insights is a great tool.

My prediction theory

I've previously articulated my theory that in certain circumstances search volumes can be directly related to product usage. On that page I outlined the ideas, and gave one practical example to demonstrate.

On this page I'll extend the theory to link the search volumes not only to usage but also to revenue. I'll give some more examples, each with more detail around them, and also show how the figures can be used to predict future revenues.

The experiment

For this to work best, we need to pick an online operator with a significant brand and a stable product. A pure online business will have a higher correlation between online search and product usage, compared to a major offline brand.

We also need a firm with a brand name that is unique enough to avoid cross over with unrelated searches. For example, would be a poor candidate as they have such a generic name.

I'm going to choose Bwin, a pan-European betting and gaming firm. Bwin has expanded rapidly through high profile branding and sponsorship deals. As they are publicly traded we can compare their actual revenue to their search volumes.

Two key points before we begin; first, this process is only expected to work for gross revenues i.e. before costs such as marketing, technology, staff etc. Search volumes indicate usage, not expenditure and hence profit margins.

Second, we must ensure we capture all the key brand search terms; bwin is an slightly usual example in that they were initially called Betandwin before changing to Bwin. However, both are unique enough to ensure that any searches for these terms indicate the users intention to transact with the firm.

So we use plug “bwin” and “betandwin” into Insights For Search. Note the dramatic switch between brand names in mid-2006 (which in itself is an interesting study, but for another day).

To get an annual total Insights scores for the Bwin brand, we take the sum of the two search terms, and then add for the yearly total.

We can obtain the Bwin revenue figures from their annual reports.

The results

The results are collated in the table above, alongside reported revenue in thousands of euros.

Things to note here:

  • Bwin have grown rapdily over the past few years, and then stablised. This is a good test of the system, as they have not simply grown steadily each period.
  • The revenue estimates are simply the Insight growth multiplied by the previous year's revenue.
  • The 2009 actual figures have yet to be released.

For 2006 and 2008, the Insight driven predictions are remarkably accurate. And in 2005 the prediction is relatively close given the huge growth.

It is only the 2007 estimate that really disappoints, predicting 18% growth against an actual decline of 7%.

However, this implementation is incredibly basic and I'm sure there are many improvements that could be made to the model.

The 2009 prediction is showing -7%, indicating a tough year for Bwin, which is a bold prediction given at the third quarter mark they were 3.5% ahead of 2008. We'll be able to check this once the full year results are announced in April 2010...


Analysis, Company, Google, Insights For Search, Investing, Prediction, Revenue

Meet the author

author avatar Chief Nut
Overly excited co-founder, keen to share the buzz so please get Wikinutting and give us some feedback! I'll be writing about new features as we add them to the site, and a few other topics as well.

Share this page

moderator Sam Wormleighton moderated this page.
If you have any complaints about this content, please let us know


author avatar elkapons
18th Jan 2010 (#)

thanx 4 the article

Reply to this comment

author avatar Sam
26th Feb 2010 (#)

Great ideas, really smart thinking. Thx bud.

Reply to this comment

author avatar Retired
29th Mar 2013 (#)

wow:) hope i will be successful here, from Philippines

Reply to this comment

author avatar Iyke dyro
3rd Dec 2016 (#)

I hope I can make some money here

Reply to this comment

Add a comment
Can't login?