Delivering value from analytics can be a mix of both art and science, as every organisation has its own requirements, practices and goals.
It’s imperative that solutions are built by professionals familiar with the business ecosystem that they are modelling. If not, there’s too much science and not enough art. A lack of understanding creates models unable to maximise the value of applying analytics to the specific needs - and increases risk.
The missing element that enhances high performance outcomes is the pre-modelling process that takes the data and transforms it to truly reflect how the system works.
For instance in marketing, you need to understand that advertising has a residual or memory effect in the minds of the consumers. Processing TV Gross-Rating-Points (GRPs), so that they reflect this behaviour, is called an ad-stock. In retail, the effects of TV advertising are almost immediate. An increase in sales can typically be seen within a week of an ad campaign running.
In automotives however, this ad-stock effect normally has a three month delay before making itself felt in sales. This is referred to as a ‘lagged effect’. If you simply look at the car sales in the same month as the advertising, then you miss the strong relationship between advertising and sales.
The art of analytics is mathematically transforming the input data to reflect the system, in this case the TV GRPS and 3-month lagged effect. Once data is transformed so that it reflects the way the system works, running the statistics is straightforward. You must consider functional form, and different analytical approaches, but at that point you have transitioned from art to a largely scientific exercise.
From statistics to recommendations
The second portion of the art comes when you are translating scientific results back into actionable recommendations. By understanding the business environment and situation the information will be used in, you need to translate mathematical results into actionable recommendations.
While these recommendations are underpinned by scientific data the formulation requires the involvement of professionals with relevant expertise and experience. The use of this experience and expertise is art.
With art and science paired correctly, our automotive advertiser would know whether they should put a $500 million (£311 million) budget into TV advertising or how much of that budget they should move over to online search due to higher marginal returns. We know from the art that we cannot spend $500 million in online search realistically, so we constrain our mathematical results to reflect what we know from the art.
This example of how to deliver better business outcomes with analytics is solely from marketing. There are however many more cases from other industries and business functions. To continue the conversation on how analytics is adding value to your organisation, I invite you to contact me to discuss how you are bringing the art of analytics to your business.
By Michael Svilar, managing director, marketing analytics delivery, Accenture Analytics