What is data analytics?

I do numbers as well as words – analysing Big Data both for marketers and in the financial markets. (I wrote a Master’s thesis on behavioural economics and work with over a hundred methods and models.)

It gives marketers huge power, because it’s the difference between marketing to test an idea … and marketing to deliver what works. Analytics does the test-marketing for you.

While it uses technology and software, human judgements remain critical – and that’s my value-add. In deciding the dataset, formulating the questions to ask of it, and working out what to do with the results. I treat it as a three-stage process: data modelling, predictive analysis, and actionable insight.

Data modelling

Capturing data starts with the question: what matters? Correct answer: everything. But that’s not an answer you can use in a world of finite resources, so it’s necessary to make intelligent choices. How big a dataset do you need? What must it encompass? How many variables create a representative landscape? A grab-bag of statistical methods, starting with Standard Deviations and Confidence Intervals, can find the edges of the jigsaw.

But numbers are only half of it: I work with qualitative stuff too, mapping in the hundred or so cognitive biases that drive everyday human behaviour. Because what customers do depends far more on personal experience and emotional state than objective logic: ultimately, all of marketing is just exploiting WYSIATI. (“What you see is all there is.”)

Customers are human, and their actions shape the data. Data modelling is about finding the “big shape”: the landscape of the customer experience.

Predictive analytics

The 64k question: “Does X lead to Y?” But X is rarely one input and Y rarely correlates with a single X. Techniques like linear programming and regression analysis can put those extra variables in the mix and iterate until something interesting happens … but sometimes the inputs need to be outputs of other processes themselves. I design and perform that sequence of operations on your data, with a view to discovering what really affects customer behaviour.

The methods themselves aren’t hard; most are in any statistics textbook. What matters is the human factor: casting a critical eye over the dataset and deciding which methods will shake out information you can use. (Often, simple is best: the scattergram or correlation peak that makes people catch their breath. Make it visual.)

Predictive analytics is about making meaning visible. Finding the peaks and valleys in your customer experience… and working out how and when the customer will climb up or fall into them.

Actionable insights

Without action, analysis is just a bunch of pie charts. The third stage is finding something to do with them. Even simple findings (“Our cars enjoy growing popularity in the 65+ demographic”) encompass non-obvious insights. (“Advertise more to older drivers” vs “Tell consumers how easily the gearstick falls to hand.”) An insight tends to be a nudge, not a shove. I deep-dive the data to tease them out.

But the concrete course of action it leads to shouts louder. And they’ll be actions you recognise. A less-travelled path through your website may contain a hole where the right content would unlock an incremental million. An extra link as prospects traverse the sales funnel might make cross-selling jump by a third. A differently-worded outreach email might result in a doubling of your conversion rate.

Actionable insights are ideas that give a measurable return. They’re backed up by solid data and come with a set of actions for taking them forward.

To get your hands on some data analysis expertise, contact me.

…see all 3 or go back home