With my curiosity piqued, I saw that the tweeter was denouncing an article which explained why marketers should use data science.

The post itself was a bit hard to follow, granted, but nothing it said would lead one to believe that data science was a corrupting influence on the venerable art of marketing.

On the contrary, the article simply stated that data science is a way marketers should explore data to become more confident in their proposed strategies. And, it continued, marketers would benefit by using data science to improve their performance. Hardly controversial.

I, however, would go further in support of data science. Besides just improving performance, learning how data science works is a transferrable skill and so could also help your career. And, finally, it’s hard to conceive why anyone, even marketers, would suffer by learning how other disciplines collaborate using data. Good enough for NASA, good enough for us.

The reason why the term ‘data science’ rub marketers up the wrong way, possibly, is that they prefer for marketing to remain an art, something creative through which we express ourselves. But this still doesn’t mean a bit of rigour in data analysis is worthless. In fact, data science has the potential to validate creativity as much as expose bad practices.

For those who are interested in learning about data science in detail and how you can apply it to marketing, Econsultancy is running a course ‘Data Science for Marketers’ in Singapore on December 12. Click below to find out more and book your spot!

But for those who are content with a general understanding of data science concepts, below are four key aspects of the discipline which hopefully put to rest any scepticism about data science in marketing.

1) Data science is a process

Perhaps the first thing to point out is that data science is not only about algorithms and high-performance computers. At its core, data science is about following a set of steps to ensure that the practitioner or team is thorough in their data analysis.

Understanding the data science process is first and foremost a matter of becoming familiar with the general scientific procedure, itself: Gather observations, build a hypothesis, test it out, revise accordingly, rinse and repeat.

datascience flowchart
Data science flowchart (via Upxacademy)


Upxacademy helpfully translates those steps into a flowchart (above) which accurately represents the typical data science workflow.

It’s surprising how much can be gained from following these simple steps methodically. And, you might wonder, what were you trying to accomplish with data before following them?

2) But it’s also a set of skills

Data science, however, requires certain skills beyond just a familiarity with the scientific method.

Microsoft’s diagram of the “Data Science Lifecycle” from their Azure blog offers some insight into data science skills and how they all fit together. In brief, data scientists typically have:

  • A solid understanding of the business area being researched,
  • Data management skills (databases and the like),
  • Modelling skills (discussed below), and
  • The ability to deploy the model, be it through a data analysis platform, web app or even just an excel spreadsheet.
data science lifecycle
Data science lifecycle (via Microsoft Azure blog)


So, while following the data science process is required, it is not enough to be able to get the most out of the practice.

3) Data science does not require advanced-level maths

The ‘modelling’ skill is often a stumbling block for those just becoming familiar with data science as the term seems to imply that you need to be able to come up with some funky-looking 3D graph using mathematical equations filled with Greek letters.

But modelling is just correctly preparing historical or ‘training’ data, choosing the best algorithm for the outcome you’re looking for and training the algorithm using your data so that it provides useful information from future data outside of your training data set.

training data
Training machine learning (via Nvidia)


And most of data science’s most commonly-used techniques, the algorithms require little, if any, math. Linear regression, a cornerstone of data science, can be done with a few clicks on a spreadsheet and many other techniques are now built-in to various, easily-accessible data science platforms.

So, while, admittedly, data science algorithms are written by experts and are often complicated ‘under the hood’, virtually anyone can use them to analyze data.

3) But it does have a learning curve

So far, so good, but there is a catch. The difficulty in using data science as part of your business decision-making process is in deciding which algorithm to use for your model. What works for your initial explorations of unfamiliar data may not serve you well when you are trying to optimize using well-understood data.



It’s not easy and the blogosphere is full of posts to help you navigate the murky waters of choosing the right algorithm. But the most important part of the process is to understand the data you have and what outcomes you are looking for. Without a clear plan, you will not know enough to choose wisely and risk wasting a lot of time testing algorithms which a poor fit for your data.

4) And it is already being used in marketing

Finally, for those more practically-minded folk, yes, data science is already being used in marketing.

To start, basic regression analysis can answer all sorts of fundamental marketing quandaries, such as how ad rank affects CPC on Google Ads.

Chart via Search Engine Land


Unsupervised learning, used correctly, can identify patterns in consumer behaviors which may help with segmenting or bundling. And, with a bit of expert help, you may even be able to leverage data science to build your own real-time bidding robot.

But before going into what seems to enter the realms of science fiction, marketers need to first acknowledge that using data, and using it correctly, can help with marketing a great deal – and that data science offers useful methods and algorithms which anyone, with a bit of learning and careful planning, can harness to improve their productivity and performance.

So, while there are a lot of ‘everythings’ wrong with marketing these days, clear thinking and a methodical approach to the practice isn’t currently one of them!

Econsultancy is offering a ‘Data Science for Marketers’ course which covers these topics and many more on December 12 in Singapore.