Putting Predictive Analytics to Work

Originally published 7 October 2009

I recently participated in a webcast for Smart Data Collective titled, “Putting Customer Value to Work: What Predictive Analytics Can Do for Your Bottom Line," with James Taylor from Decision Management Solutions, Anne Milley from SAS Institute and Mike Rote from Teradata Corporation.  We discussed how to best use predictive analytics in the decision-making process. In this article, I would like to share what we discussed during the webinar.

Predictive Insight in Telecommunications

It will be very beneficial to examine the use of predictive analytics from the customer life cycle point of view in the telecommunications industry. In my previous article, “What it Takes to be a Real CVM Player,” I mentioned the stages that the customer passes through during his/her life cycle with the company.

In the acquisition phase, you need to target the right customers with the right products at the right rates. The most common acquisition techniques where you can use predictive analytics include:

  • Profiling prospects who visit your website to personalize their experiences,

  • Analysing prospects who are most likely to respond to the offers,

  • Using response modelling to predict which marketing programs will generate the highest response.

Marketers interactively refine these segments, select optimal target lists and deploy the resulting predictive insight into front-line applications.

When we look at the value enhancement phase, telecommunications provides various examples about how predictive analytics can be used:

  • Targeted campaigns that predict which customers are likely to respond to a new offer.

  • Segment-migration models which are primarily used in base management activities that predict which customers are likely to migrate to more or less valuable segments.

  • Cross-sell models which predicts the likelihood of buying additional product or service.

  • Up-sell models that determine how likely it is that customer will make an additional investment in a product or service, such as upgrading to a more expensive phone plan.

  • Profitability modelling and optimization techniques which enable operators to understand which customers drive profitability by allocating costs to the processes.

In the retention phase, the most common use of predictive analytics is the churn prevention programs. In the churn management process, you actually develop three models:

  • One model to identify likely churners,

  • The next model to pick the profitable potential churners worth keeping, and

  • The third model to match the potential churners with the most appropriate offer.

And finally, in reactivation phase where you try to win your good customers back, you simply calculate new potential lifetime value (LTV) of candidates, then target high value lost customers and build focused campaigns for the highest return.

Trends in Analytics

There are many trends in the analytics space. Let’s look at a few of them.

Predictive campaign analytics: One of the emerging trends in the analytics area is predictive campaign analytics, which involves the analysis of customer behaviour in campaign management. It enables a company to target customers more effectively, such as the ability to stage an offer during a customer interaction.

Newer marketing techniques, such as inbound marketing, are really taking advantage of predictive analytics with offer management applications that put recommendations in front of marketing analysts and campaign managers. Segmenting customers and prospects based on their propensity to churn or purchase can significantly increase response and conversation rates.

Real-time decisioning: Real-time decisioning is the next wave of business which combines predictive analytic and decisioning capabilities to identify the optimal next action to take in a real-time process, such as a customer service interaction or website visit.

This technology is really moving from a niche capability, and we see many projects in the telco industry emerging. Some early adoption has primarily focused on the contact centre, turning a purely service-oriented interaction into a blend of marketing and sales activities.

Content analytics: Another area is content analytics which includes text analytics (photo captions, blogs, news sites, social networks) as a subset, but which can analyze many more data types, including multimedia, photographs, speech and faces.

Content analytics is used to support a broad range of functions. It can identify high-priority customers, product problems, customer sentiment and service problems. You can even analyze competitors' activities and consumers' responses to a new product It also has a vertical focus, such as "voice of the customer," to analyze call centre data.

Social network analysis (SNA): And finally, SNA, which is simply a technique for analyzing patterns of relationships among people in groups, is really useful to learn the social structure of individuals or organizations.

In telecommunications, SNA uses information from call distribution records such as number dialled, incoming number, call count and types of call to find out information about the individual consumers and their calling circles. This basic information can then be added to other information to provide specific data sets for different activities such as preventing churn or planning marketing expenditures.

Reduce Time to Answer Business Questions

Think about call centre agents who make yes/no decisions everyday about how to best treat and retain customers. They handle hundreds of operational decisions in their day-to-day activities. Most are not able to convey the same message to the same kind of customer consistently. That means different messages, activities and decisions are passed to the customer which, in turn, leads to inefficient management of the customer.

If you consider the total impact of those decisions taken everyday, the result could have a big impact on the company’s overall effectiveness. Thus, any improvement in that process will reduce the time to answer a specific question. If proper predictive analytics techniques can be applied along with the integrated decisioning capabilities, the customer experience would be much better. When a customer calls, the agent automatically selects the best type of offer (i.e., retention, cross-selling) along with sales arguments and other relevant information during the conversation and presents it to the client in real time. So if your call centre agent knows how to deal with specific questions ahead of time and is backed up with many insights regarding solving a particular problem, the call is likely to have a great customer experience. The value of analytics becomes substantial when you use it to improve your day-to-day operations in addition to the strategic decisions.

Predictive Analytics in Personalization

Personalization has always been a focus area in telecommunications in a search for understanding the single question of “What do customers really want?” Personalization has two meanings. On the one hand, it is the ability for the customer to modify the process to benefit themselves. On the other hand, it is the ability of the company to differentiate its process on the customer's behalf.

For a personalization to be really successful, you need to have consistent multichannel customer service interactions including call centre applications, IVR systems, retail stores, web portals and online billing applications, mobile decks, storefronts, bill inserts and direct marketing systems. You also need to build a platform that allows automated decision management capabilities ­ such as advising and prompting the agent with the recommended next and personalized action to take.

There is a great example here that I would like to share. One mobile operator analysed its subscriber’s usage, primarily the failed MMS attempts along with the additional device data. Then, they sent the required MMS configuration for that particular mobile phone to each subscriber who faces the similar problem. You can guess the result, not only an increase in MMS usage but also significant customer experience with extreme personalization.

Choose to Lead or Lag Behind

Leading organizations are using predictive insights to improve their decision-making process. They invest in their valuable data assets, build predictive analytics techniques and spread analytical tools and mind-set across the organization which turns insight into actions. Organization can either to choose to lead by putting predictive analytics into the workflow or they lag behind.

  • Korhan YunakKorhan Yunak

    Korhan Yunak is a Global CRM Business Analyst at Vodafone Group, based in Dusseldorf, Germany, where he is responsible for business intelligence strategy, data warehousing best practices as well as the customer value management program within the group. He currently works on customer value management practices, BI strategy and customer analytics focusing more on the CRM area to cover the big picture around business intelligence. He has more than 6 years of practical project experience helping companies establishing enterprise data warehouse solutions and building BI vision.

    Early in his career, Korhan was involved in various enterprise performance management (EPM) and BI engagements in wide variety of industries including telco, banking and retail as a consultant. Korhan has extensive experience, combining technical and business acumen, especially in telecommunication industry.  Korhan holds a degree in Management Information Systems from Bosphorus University.

    Korhan can be reached at +491732620423 or kyunak@gmail.com.  

Recent articles by Korhan Yunak



Want to post a comment? Login or become a member today!

Be the first to comment!