Data analysis is an art as well as a science

By: Mark Patron | October 13th, 2011

It sounds easy – right customer, right offer, right time and more recently right channel. As the meerkat would say… simples! So what’s the problem? Well, in the spirit of answers are easy, it’s the questions that are difficult, let’s start with some questions and challenges.

Challenges

First customer analytics has become much more complex over the last ten years with many more degrees of freedom, for example channels. Digital and telephone channels have also created a need for real time analytics. Previously campaigns typically had their own discreet time slot, whereas now many campaigns will run simultaneously. Gone are the days when you could analyse that TV campaign in glorious isolation. Different channels have different data sets. Sometimes these data sets are incompatible, for example offline and online data. Lots of data exists in disjointed systems.

Computer processing power has led to an explosion of data volumes. But more data does not necessarily lead to better decisions; for example the credit crunch was a car crash with the banks seemingly being blind to the now obvious subprime mortgage lending problems in the US.

The recent confusion with the new cookie laws is a good example of the challenges created by privacy. Balancing the consumer’s right to privacy with the marketer’s wish to target is always a fine line.

Often there is no one simple objective or dependent variable. For example sales are important but customer engagement and online registrations also need to be taken into account. Last but not least it is always a challenge to sell it to the board. Return on investment is not about algorithms, it is about accounting.

How to address these challenges

First recognize that good analytics is both a science and an art. Improve your emotional intelligence when it comes to data analytics. You can do this using some very specific techniques. Develop multi-functional teams to get more pluralistic input. Use complimentary data and methods. At my own business we have found incorporating usability research techniques with web analytics has led to much better results when trying to improve website conversion. Most data analysts are left brain people who see the World as something that can be targeted, segmented and put neatly into boxes. We also need to engage right brain thinking for better creativity and big picture possibilities.

While it is important to use best of breed tools such as SPSS or SAS it’s even more important to worry about who will drive them. Avinash Kaushik has a very apposite rule of thumb for web analytics, that is for every $100 spent on web analytics spend 10% on the analytics tools and 90% on the people.

Use a scorecard approach to handle many different objectives at once. Identify a small number of measures and attach targets to them. Don’t try to eat the elephant all at once. Break complex analytics questions into their constituent parts. Marketing Directors don’t tend to like neural networks, not just because they are complex and unwieldy, but because they give a black box solution which cannot be debugged in a transparent way when something goes wrong. So make sure your analysis is transparent and robust.

Strive to become a more data driven business by making your data more actionable. Explain to senior management at every opportunity why data is an asset in accounting terms. To find meaning in data it’s now as much an art as it is a science. Having enough data is no longer the issue; it’s what you do with it that counts.