How data science changed marketing forever

Emilia Marius
Share this content

If we think back to the days before the Internet, print media got the lion's share of the advertising spend. The advertising reach was defined by the number of copies sold. Sometimes it was about the number of readers if it was more than the number of purchases.

In reality, there was no way of gauging the success of your advert, unless it included some feedback mechanism, like a competition entry that required the consumer to cut out the entry form from the magazine or newspaper. Digital marketing has changed all that. Not only is it possible to get feedback on the market penetration of a campaign or a company’s website - the number of clicks, pages opened and time spent are all available to the marketer. 

Reversed Roles: The Customer Reaches Out

The web has created huge changes in marketing: we have websites like virtual storefronts, where customers pause and browse what we have on offer and then either buy or finish their electronic window-shopping and move on. We create a presence in the various social media to attract followers and encourage conversation and feedback. All of these strategies yield us new and relevant data about our customers and whether we are engaging with them or need to tweak our current offerings. Even the situation where someone clicks on our landing page and leaves abruptly tells us a story about the success of our call to action - we can measure the time spent and the number of people who were not engaged.

We can even keep in touch with our virtual shoppers when they are offline via a mobile app that they download. This enables us to measure our customers’ shopping behavior as soon as they enter our brick-and-mortar stores. We can create even more interaction by using location-based marketing, where visitors to our stores receive useful information and promotions via the app that they downloaded to their mobile device. Our loyalty program is stored in their e-wallet. The word "phygital" describes this merging of the real and virtual worlds.

Every one of these activities creates volumes of new data that gives us a view of our customer as an individual that we never had before. Instead of slicing and dicing our target market into "segments", based on age, gender, income and other generalities, we can take the reverse approach, building a "persona", such as a 35-year old employed woman and mother, or a 45-year old businessman, based on their social interactions blended in with the traditional retail data of shopping basket contents.

What the New Data Unlocks for Marketing

The new data enables Marketing to determine the profitability and ROI of any marketing initiative, because all the customer touchpoints have been recorded and stored. Before and after metrics are accurate.

Heineken ran a campaign tied in to the release of the James Bond movie Spectre, where they could assess the take-up as campaign result.

It is also easy to run pilots and test market response. Even Facebook (which has a daily software release) does this from time to time, gauging market reaction of a selected sample of customers before rolling out the update.

The real strength of digital marketing data, however, is its ability to let the company view each customer as an individual and to engage with him or her across every touchpoint along the customer journey. A long-term relationship can be built based on content-based marketing that encourages loyalty. Proactive action can be taken, where hints of dissatisfaction or a decrease in engagement and purchasing can trigger a retention strategy.

Marketing can also radically improve sales conversion rate, by qualifying leads to a level never before possible with enhanced lead scoring using the digital marketing feedback, as shown in this excerpt from Marketo's Definitive Guide to Lead Generation.

However, the status quo of how BI is delivered in most businesses has to change.

Why Traditional BI Has to Leave IT

One of the dilemmas of our fast-changing world of commerce and e-commerce is deciding what to do with new things digital. The normal home for these problem children is IT. It is ironic that "Business Intelligence" has up to now been something provided by IT. Business decision-makers relied on the IT department producing reports, either from enterprise systems, such as SAP, or from ad-hoc or recurring requests. These reports came out at a scheduled time, often weeks after the request. The new data changes all this.

Much of this new data is "Big Data", unstructured and disparate data that does not fit well into traditional databases. Big data was characterised by Doug Laney in 2001 as having 3 characteristics - the "3 Vs" of volume, velocity and variety.

  • Volume - the sheer size of data acquired by customer interaction with our digital platforms. This will expand even more. Imagine the storage requirements for footfall within a large retail chain, or the data picked up by beacons in an airport like Dubai International (with 81 million passengers per year).
  • Velocity - the speed with which this data is gathered, mostly in real-time or near-real-time.
  • Variety - the complexity and diversity of the data, from text messages and emails to facial recognition images.

While variety is the deciding factor in whether to use a big data database for storage, velocity is also important. The immediacy of the data means that it cannot go through the usual verification and data cleansing before it is stored in the traditional database or warehouse. The sheer volume makes accuracy less important - ten dud records in a dataset of one million is not going to affect the results of any database query in any meaningful way.

The information from all this data is needed as soon as possible. A three-week wait could render the business intelligence extracted obsolete by the time it is received.

How to Manage the Data Explosion

Obviously, the business needs to regain ownership of business intelligence, especially marketing that generates most of the data in the first place. The problem up to fairly recently has been that data extraction from a big data database, or a combination of data sources, was very complex. This has created a massive demand for data scientists to do this work. Enter a new breed of software - the data discovery tool. This is software that enables the business user to extract meaningful and statistically accurate data without the need for a data scientist.

The number of data discovery products increases by the day. Each has specific benefits and disadvantages, but they are geared at a user who does not have a degree in statistics or quantum mechanics. Most packages offer beautiful visualization options. What any product in this genre offers though, is autonomy for the user, and removes the dependence on IT.

Taking up the Data Discovery Challenge

The change to BI is accompanied by other changes both to organizations as a whole and especially to marketing departments. Chances are that your current environment is fragmented and needs to be integrated. Areas to receive attention are marketing, which typically uses a dozen different software applications, and data storage. It may be advisable to call in a data science consulting company to assist in reorganizing your data architecture. Much of the existing data may be in silos and not generally accessible. Conversely, some of that data should have restricted access, because it is very personal. Another matter to consider is whether new resources are required in marketing, such as a data scientist. There is some debate about new technology roles in marketing, which is the subject of another article. What is critical is to start embedding the ability to extract data from your many and varied data sources in all your departments, starting with marketing.


Please login or register to join the discussion.

There are currently no replies, be the first to post a reply.