Three key things Big Data allow the marketer to do with segmentation that was not possible before | DMA

Filter By

Show All

Connect to


Three key things Big Data allow the marketer to do with segmentation that was not possible before


For years, segmentation has been fundamental to engaging customers and achieving superior financial performance. The emergence of Big Data has created tremendous variety and timeliness of customer information

Thus, segmentation can now unearth greater customer insight.

The Big Data-enabled possibilities are enormous, but we focus on three areas that can bring relatively quick wins.

  1. Increasing usage of text mining

The importance of text mining has grown with Big Data because much of that data is unstructured. A common example of this is text data such as customer comments. Text mining enables conversion of this text into structured data that marketers can use in segmentation.

For example:

  • A customer enters many sentences worth of frustration with a company in an online text box.
  • Text mining this unstructured data results in categorising the customer as “unhappy” due to “service issues.”
  • This customer categorisation becomes an input to a segmentation scheme.
  • The segmentation scheme then classifies the customer into an appropriate segment, such as High-Value-Dissatisfied or Low-Value-Dissatisfied, leading to appropriate retention tactics.
  1. Leveraging social network data

Let's say, company iSell has a product BuyMe. Now assume there are three individuals — HappyBuyer, ProspectFriend, and JoeStranger — who share a similar profile based on demographics and other traditional data.

First, Big Data can enable iSell to know that HappyBuyer and ProspectFriend are “linked” in social networks, but JoeStranger is not linked to any buyer.

Second, since HappyBuyer has a long relationship with iSell and is passionate about BuyMe, he may have a positive influence on the purchasing activity of ProspectFriend.

Third, while the traditional data-based segmentation scheme would have placed ProspectFriend and JoeStranger in the same segment, ProspectFriend and JoeStranger could potentially be in different segments based on their propensity to be influenced for purchasing BuyMe.

Also, typically, value segmentation schemes have classified customers as high or low value based on past or expected future spending. Now, a portion of these low-value customers could be classified as high value based on their influence within social networks.

  1. Updating segmentation frequently

Historically, in most organisations, segmentation is “refreshed” only periodically, between monthly to once a year, driven by data refresh schedules and associated process challenges.

My colleague, Epsilon CTO Chris Harrison, notes that “Technology exists for marketers today to place individuals into segments in real time and, perhaps more important, move consumers from one segment to the next in real time.”

As customer data is being refreshed in a more timely manner than in the past, it's only logical that to benefit from such information, organisations must improve on the segmentation refresh frequency, driving it towards near real time.

As organisations adopt Big Data, there will be many challenges. Therefore, it's critical to achieve quick wins to build and sustain organisational momentum. In addition to technology solutions, key skill sets such as network analysis and text mining need to be developed to execute the strategies outlined above.

These skills can be built internally or leveraged through partnerships. Regardless of the approach used, the path forward is exciting and one that will result in superior customer experience, as well as strong financial performance when done right.

Learn more about personalisation in Epsilon's Do Take it Personally webinar on Wednesday 9 December.

Hear more from the DMA

Please login to comment.