The Data Scientist’s Role in eCommerce Marketing

For the last few years, big data has been gaining momentum as an industry-wide hot topic. This is where data scientists come in. Though people have had no qualms with overusing the term, even experts have had a difficult time pinning down exactly what big data means. For the purposes of this post, let’s define big data as massive amounts of collected data that are difficult to analyze or organize using traditional database management methods.

You’ve heard of Amazon’s data scientist army; the Harvard Business Review named data scientists as “the sexiest job of the 21st century.” But what exactly do they do? In the Plainly Speakingseries, we’ll give you the information you need to arm yourself as a savvy marketer – but without the buzzwords and confusing jargon. In this post, we’ll cover what data scientists do, and why their work is so important for effective eCommerce marketing.

Data Scientists: Part Analyst, Part Artist

Do you remember the Magic Eye books of yesteryear? You hold the image of repeating patterns close to your face, and then slowly move it away. As the focus of your vision changes, you’re able to see an image within the image.

hand

A Magic Eye stereogram – can you see the shape?

Data scientists do something similar, but with information. They look at data sets and identify patterns and trends. These patterns and trends can then be interpreted to reach specific conclusions, until suddenly you can see the whole picture. Just as your eyes adjust to recognize shapes in the chaos of a Magic Eye stereogram, data scientists have better-trained eyes and techniques to do the same with big data. Here is a quick overview:

Data Cleansing

Big data encompasses huge amounts of data from different sources, so it’s often messy and in mismatched formats. Data must be cleaned and organized before trends can be identified, at which point data scientists start building models with the information they glean.

Modeling

In the most basic sense, a model can be defined as your belief about how something works. Modeling is a somewhat abstract term for how data scientists build mathematical representations based on those beliefs. Don’t let the “math” part confuse you – we’re not talking pure numbers, but rather the formulaic and logical nature of the representation.

An extremely simplified version of modeling would be defining your belief in a set of if-then statements, based on patterns you recognized in the data. For example, say you notice that when you include a song title in your email subject lines, your open rate consistently spikes the next day by 15%. You can create an if-then statement based on this – “if song title, then 15% spike” – and start making predictions. If you include a song title in your email subject line tomorrow, you can predict that there will be a 15% increase in open rates the day after tomorrow.

But by definition, variables change, so you must monitor and adjust your model when needed. You might realize your prediction doesn’t apply on Fridays, because people don’t check their emails as regularly on weekends. Your model would then change to “if song title AND not Friday, then 15% spike.” As you collect more data, you refine your model.

data-model-ex
A visual representation of a data model.

Translating

IBM once defined the data scientist as “part analyst, part artist.” And this is true. Analysts look at past and current data to understand the movement and interpretation of data; to a certain extent, data scientists do too. But data scientists take it a step further by building on those interpretations to forecast for the future. They take identifiable patterns from all different kinds of data and help translate the knowledge into easily understandable terms. They reveal the shapes within the images, but they also help us grasp what those shapes mean and will mean in the future. That’s an art.

Why Marketers Should Care

The more fully you understand the role of data scientists, the easier it is to grasp why what they do is so important to eCommerce marketing. Marketers interpret data, too, from open rates to click-through frequency. They take on the analyst role by looking at past data, identifying trends, and coming up with reports. Decisions are then made based on those reports.

But compared to the vast amount of data generated through online commerce, marketers are only looking at a fraction of the information available. Every interaction from an online shopper leaves a cyber-footprint. Information is a natural byproduct of eCommerce: the problem is that we don’t know how to read that data. Data scientists take that information overload and distill it down into insights on customers – and for marketers, that’s pretty much invaluable.

Screen Shot 2017-07-19 at 1.42.20 PM