How Data Science Helps Marketing

This post is the updated version of a piece I wrote last year, How Data is Fueling Marketing. I updated the content after the talk I gave at SXSWi 2016 in Austin, Texas.

Data science is rapidly changing marketing. Specifically, contextual marketing is on the rise. Contextual marketing differs from traditional marketing. As its name suggests, this type of marketing strategy is all about CONTEXT. We are no longer interested in observing and predicting how people will carry out simple conscious-decision making. Instead, with this type of marketing, we seek to observe the habits that individuals get into when buying a certain product, booking a trip, or browsing through online catalogues.

Habitual patterns shape around 45 percent of choices that we make every day. Through contextual marketing, we get to discover and understand not only who someone is, but their geo-locations, what they are doing in different areas and points of time, and what they are most likely to do next. To be more precise, with the data available today, we can observe an individual’s age, gender, past behavior, device usage, location, weather conditions, time of day, and purchase history among other relevant factors. This helps us deliver the right message at the right time, and identify the right trends. Our ability to observe and assess deeply embedded contextual characteristics changes everything.

Contextual marketing is already being used strategically by big firms. For example, the online booking company, Orbitz, shows higher priced hotels to Mac users! Travelocity, on the other hand, gives a 5% discount for mobile shoppers. American Airlines has been known to raise ticket prices as you are shopping to create a sense of urgency. The shopping retailer Target has created a “pregnancy detector,” that basically data-mines its way into your womb through observing the shopping and browsing history/cookies of women that visit its website.

How does this work? Marketers know that pregnancy is a very profitable time period. Through observing hundreds of thousands of different purchases of products, data scientists were able to identify 25 of the most salient products that significantly contribute to a “pregnancy prediction” score. This analytical model helps to predict the shopping behavior that individuals will carry out at certain time periods and within small windows of opportunity. Target then sends coupons at specific stages of pregnancy when pregnant women are likely to search and buy their products. 

How does timing work? Take the following example: lotion is a widely used and widely purchased by everyone, not just pregnant women. Pregnant women, however, were observed to buy larger quantities of unscented lotion around the beginning of their second trimester. 

Data science not only compliments traditional marketing extremely well, it has transformed it before our own eyes. An important aspect about data science is that it improves contextual marketing through inputs and outputs. Inputs such as: age, gender, income, location, search history, purchase history, time of day, etc. These inputs generate relevant outputs which can be a game changer for your service/product such as: what channel you need to use to target your customer, the right price and discount for your product, what specific products is that customer interested at a specific time of year, etc.

One of my favorite examples of how data science drives context is Stitchfix. In the Mary Meeker’s Internet Trends Report which was just released on June first, Stitchfix is the font runner in micro data-riven engagement, satisfaction, data collection and personalization. Stitchfix is applying Netflix and Spotify-like content discovery to fashion, giving each customer differentiated experience. Stitchfix has a data driven on-boarding process which mixes art and science. They collect data points on customer preferences, style and activities. Stylists use Pinterest boards and access algorithms to help improve product selection. The slide bellow describes how Stitchfix uses data science to not only drive context but as the base for their whole product.

Get the full report at Kleiner Perkins

Nestle Purina also had success with data science through using it to personalize not only customer information but also customers pets! Nestle Purina, in recent years, has created a goal to provide pet owners tailored nutritional plans for their loved pets. As the second largest global pet food company, data science has helped Purina launch a number of customer-targeted ads through article indexing, clustering article structure, and through the usage of recommendation algorithms.

All of these examples might sound great. But the question arises, how do we actually achieve and implement contextual marketing with the support of data science?

Define your strategy ? whatever your brand is really about, comfortable shoes, fun to drive cars, fast food delivery, unforgettable experiences reinforce that value. At the same time start thinking about what inputs would be the most beneficial in improving your product and also the outputs you’re hoping to generate.

Facilitate customer journeys ? figure out how to address the customer journey, before, during and after purchase. Focus on customer experience and engagement.

Content is still king ? every piece of content that you deliver during an interaction must be tailored not only to the customer but to a specific moment, based on the data you gathered from past interactions.

Work with you IT department ? to bring in the right tools to execute automation and engagement.

Today we as marketers have access to an abundant amount of data. It is you who has to be willing to dig into it. Search, mobile device usage, content browsing and Social Media activity reveals the journey a customer takes throughout their lifetime. Redefine the purpose of campaigns, invite your customers to not just purchase your product or service but also to establish a deep relationship with you.

This is just the tip of the iceberg. If you have more questions please reach out to me on Twitter.

See the presentation that goes along with this post on Slideshare.