blog_header

Google Analytics and AdWords optimisation with GNU R

9th – 11th March we were at Booster Conference in Bergen to present AdWords optimisation using controlled experiments with GNU R.

In the session we:

  • showed how standard analytics reports can bring misleading results
  • showed that controlled experiments can help to identify the true ROI
  • reviewed a study (PDF) by a team of economists from eBay Research
  • provided examples with real life data
  • and showed how to implement these methods using GNU R and the Google Analytics API – Big Data without Big Code

In the presentation we focused on brand keyword bidding. In this case, paid clicks potentially substitute organic traffic.  The advertisement effect we observe is the total AdWords traffic  which includes not only the truly gained traffic but also the clicks of individuals that would have visited the website even in absence of SEM. The latter is what economists call the missing counterfactual. Missing as we do not observe it. Without this data it is impossible to calculate the true ROI.

The way to find the missing counterfactual is through controlled random experiments. The complete AdWords optimisation toolbox necessary to run controlled experiments can be found in this slideshare presentation.

We are excited to share our methods. Feel free to share your opinion.

Read More

data2day: Big Data Conference in Karlsruhe (2015)

Meet us at the data2day conference!

The data2day conference, organised by Heise and dpunkt publishing companies, will take place in Karlsruhe, Germany from 29 September to 1 October. Hinnerk will be presenting a session there on how to use GNU R and the Google Analytics API to optimise AdWords bids – and show to conduct big data analysis without requiring big IT projects and do a case study on brand keyword bidding.

In the session, I will be looking at

  • when standard analytics reports can be misleading and how controlled experiments help
  • review a study (PDF) by a team of economists from eBay Research
  • and show how to implement these methods using GNU R and the Google Analytics API

Case Study: Brand Keyword Bidding

These methods are especially important for sites that combine organic search and paid search acquisition channels. In this case, it is important to know how many sales are truly gained by a paid campaign – and where paid clicks potentially substitute organic traffic. A great example for this is bidding on your own brand keyword – typically, the brand will be the first paid result and the first organic result.

But often, others are also bidding on the brand keyword; in the case of travel, this may be OTAs and their affiliates. Finding out precisely how brand keyword bidding, by oneself and others, influences sales – and not just clicks – is possible with a controlled experiment approach. More on that at the conference!

If you are German-speaking and interested in big data technology, this promises to be an exciting event. Early bird discounts are available until Friday this week.

Image Credit: The cover image is based on https://flic.kr/p/w7fE3h, published under a Creative Commons share-alike licence.

Read More