CSP data driven Analytics trends for 2014
By : Flytxt
The digital shift is inevitable for consumers with half the global population carrying a smartphone and next generation digital services like M2M finally coming of age. Proliferation of digital services and explosion in connected devices will surge Big Data in CSP world to the order of Zettabytes per month. IDC predicts that CSPs on an average will be scanning 100 billion records a day by end of 2014 as again 40 billion records a day now. CSP worldwide are in the pursuit of finding the perfect Big Data recipe for leveraging this huge information asset to solve their immediate concerns of decline in revenue and margin as well as increase in customer churn.
2013 was a year of experimentation and trials. We have seen a number of use cases applying Big Data Analytics to make an impact across functions like network optimization, contextual marketing, customer care, location based services, etc. 2014 is when Big Data is expected to go mainstream in Telecom with almost two-third of the CSPs planning to deploy Big Data Analytics capabilities to generate sustained economic value.
1. Integrate any and all data:
CSPs realize the importance to integrate any and all data for decisioning due to its multi-dimensional nature. Decisions cannot be made looking at one data point alone or at one dimension of the data, it could go horribly wrong. The whole relevant data cluster has to be sliced and diced and then analyzed from multiple dimensions. Recommending more sales outlets in a region going only by subscriber count could erode margins if the majority of subscribers turn out to be in Low value segment. A Product Manager coming up with a flat data pack without assessing the usage trend and cost of service could again bleed. There is increasing need for cross functional and external data for improving decisioning. Some examples are Network team considering Subscriber distribution by CLV to determine the profitable locations/routes for QoS optimization and BTL marketing campaigns team looking at Call drops and consumers sentiment analysis from customer care while running retention campaigns or acquisition teams looking at integrated social media data to enrich customer profiles for targeted acquisition programs. The focus for CSPs is going to be around how to integrate each and every data in a useful and readily accessible form for contextual decisioning.
2. New stakeholders emerging to own Big Data projects in CSPs
CIOs used to own majority of the Big Data projects as it was mostly considered as a technology transformation and enterprise-wide data integration initiative. However, Big Data use cases have proved its potential to generate business value across different operating teams like Marketing, Finance, Network, Customer Service, so on. We will see more and more CMOs, CFOs, CTOs and CCOs demanding and owning Big Data projects. Unlike other CIO initiatives like OSS/BSS, the ROI of Big Data projects is not just dependent on IT efficiency, but more on how data is consumed and analyzed to derive insights for decisioning and value creation. Functional leaders realize this and will look forward to play a bigger role in planning and executing Big Data projects.
3. Focus to shift from experimentation to sustainable economic value generation
Last few years, we are seeing many use cases involving CSPs in both developing and matured markets leveraging Big Data to optimize contextual marketing, margin management, network optimization and new business models for monetizing data. While many of these use cases have resulted in 5% or more incremental revenue and margin for CSPs, the key lies in how they can sustain it and do it with ease. There would be renewed focus on sustained economic value creation from the information assets.
4. Value of speed will be acclaimed
Data is there in plenty but to derive value out of it, we need to transform this raw data to insights which in turn should be leveraged for decisioning and instant action. Contextual use of data increases its potential value and a timely good decision or action is always better than a delayed best decision or action. A top tier CSP last year implemented Flytxt demonstrating that real-time event triggered marketing evokes 40 to 300% higher favorable response from consumers. It is just not the data which has to be captured and analyzed in real-time, but how real-time analytics could be leveraged to trigger a real-time action too which will make it effective. CSPs will attach highest priority to maximize the potential of fast data as they see a faster ROI in doing so.
5. Unstructured data driven use cases to flourish
CSPs have lot more data sources coming in now with the onset of social media and the advancement of technologies like DPI, mobile payments and M2M. Since the data doesn’t have a standard form, it gets all the more challenging to interpret and derive insights in time. CSPs on the other hand realize the benefit of tapping this huge ‘unstructured and uncertain’ information asset. DPI logs if properly analyzed is a mine of information on consumer’s browsing patterns that can unveil his contextual needs and preferences too; a consumer’s state of mind and his personal sentiment is better expressed in social media interactions. The other worthy consumer of unstructured data is location based services. The next year will see many of these use cases emerging from a mere experimental to more commercial deployment.
6. Predictive and behavioral analytics maturing
Analytics has definitely moved beyond query based reporting and data mining. CSPs are experimenting with predictive and behavior analytical models especially for use cases related to churn prediction, price modeling and mobile advertising. We will see more of these models coming into limelight and being extended into areas like network optimization, margin management, self-care as well as sales & distribution. Academic and research bodies are partnering with vendor community to develop and mature these models. 2014 will also see predictive and prescriptive analysis coming together. CSPs believe that predictions alone cannot drive economic value, for example predicting the churners is not good enough, the real benefit is when causal churn analysis can be done and next best action can be suggested to mitigate the churn.
7. Man-machine collaborated decisioning key to maximize value
The blockbuster Hollywood movie ‘Terminator’ told a story which is relevant in Big Data world too. Machine can crunch billions of data records and also feed in intelligence and throw options and recommendations in real-time, but you still need the intuition, reasoning and judgment power of a human being to factor in aspects like domain understanding, local market knowledge and competitive differentiation for decisioning. 4 pillars are evolving out – data analysts to choose and integrate desired data, data scientists to build scalable and interpretable data models, decision scientists to apply inductive or deductive reasoning to derive decisions and operation analysts to make sure that whole ‘data to insight to decision’ process flow is smooth.
8. Significance of roles of data and decision scientist to increase
Big data technology capabilities, Data Analytics techniques and value generation use cases are all evolving at fast pace and needs specialized skillset to cop up and maximize the value of data. These are new job profiles and new skill sets to CSPs. Though many CSPs attempted to bring these skills in-house, better sense prevailed and latest Gartner survey showed that more and more CSPs are looking for vendor community to step in and assume the roles of data and decision scientists to ensure sustained economic value creation leveraging Big Data Analytics. In addition to technical and statistical skill sets these profile demand a lot of multi-disciplinary and intuition skills too which explain why it should be left to specialists to handle.
9. Visualization to focus on faster decisioning
New visualization techniques are evolving as Big Data technologies and tools are maturing in analyzing multi-dimensional data. This year we will see CSPs and vendor communities giving more focus on how to visualize the data better to enable business users to make instant and better informed decisions. Flytxt is working with CSPs in this direction and we evolved out a visual map for each customer segment to detail out their usage behavior classified by using primary and then secondary KPIs which in turn can deliver insights to associate churn behavior as well as up-selling/cross-selling opportunities with each segment. Visualization of multi-dimensional data clusters in CSP world is going to be a focal point for R&D effort in the coming year.