In part one of this two part series we discussed the evolution of the big data concept to that of "fast data". With particular reference to Communication Service Provider (CSP) marketing strategies, we outlined the major opportunity that exists for operators to harness customer data, build tailored services and interactions that are personalised, relevant and engaging, and to do so in real-time.
Rivalling the sort of real-time interaction that customers receive from Google, or a variety of OTT players and in response to the highly competitive environment in which they operate, CSPs are starting to realise the need to react within milliseconds to customer or network activity. Indeed, we say that 250 milliseconds can make the difference between retaining and losing a customer.
However, to achieve this CSPs will need to redesign their businesses to enable them to personalise the customer experience as it's taking place. They need to adopt architectures that enable them to capitalise on the vast quantity of data available to them in real-time, evolving their strategies from big data only, to also include fast data.
To do so, CSPs need to address the following challenges:
Tapping into "real" real-time data
Customer interaction and marketing systems are rarely architected to tap into real-time customer data from the network. A typical campaign management solution is integrated with a data warehouse, built with inherent latency, and many batch processes. Often, source systems are legacy platforms that were never designed for real-time; the best that can be achieved is a "near real-time" integration.
The inherent latency of data means that personalisation decisions cannot occur in real-time, hence any personalised customer interactions occur too late, and lose relevance. To achieve real real-time personalisation levels, CSPs need to evolve business and marketing infrastructures away from batch-based integration of systems to agile, real-time applications.
The end of a call, use of the mobile device in a particular location, calling to a specific destination, or a prepay recharge are all events which could trigger targeted marketing offers. When a customer initiates any one of these actions, the offer must be received in the split second that the device is still in their hand. If this notification is to be enriched with a personalised contextual message for that customer, then this decisioning and execution needs to occur within 250 milliseconds.
Such applications are architected to tap directly into the massive volume of network level data, circumnavigating the legacy batch-based systems and going direct to the network source systems. They ingest network level data, undertaking contextual decisioning, and initiating customer interactions, all in real-time. This is the world of fast data applications.
Identifying "moments of truth" in real-time
Big data is not a new phenomenon for CSPs who routinely process billions of call detail records per day. However, with the advent of smart devices and the growing volume of data available from apps, sensors, RFID tags, internet browsing and social media, the sheer volume and diversity of the contextual and signalling data available has increased the challenge significantly.
Moreover, in the next decade yet more data will be created by the Internet of Things – sensors and devices connected to the internet, encompassing everything from smartphones and tablets through to cars and electricity networks.
Data levels are becoming so vast that businesses now need the help of big data analysis and machine learning platforms to process billions of records every day. This must then be interpreted in order to enable business rules and intelligent pattern recognition algorithms to identify relevant events, trends or patterns which would trigger the need for a proactive customer interaction.
Picture a customer who cannot post their comment on Instagram because they have just burnt through their data bundle watching a video on YouTube. They need to be informed of their burn rate, and be made a personalised offer for a data bundle top-up in real-time, before they experience the frustrating "wheel of death" that indicates buffering – and attribute that frustration to their CSP.
Alternatively, if data services simply continue at high "out of bundle" rates with no intelligent customer interaction, this will inevitably lead to "bill shock", and drive customer dissatisfaction and damage to the CSP's brand.
Agility and flexibility for the business team
To date, most real-time interactions with customers are established within the network, which are designed for packet switching, or real-time rating and charging. Typically such messaging is configured within the network elements themselves, which are understandably protected at an engineering team level, who must maintain the core network.
However, business and marketing teams need to react to the market quickly, and have the flexibility and agility to configure new customer interactions in real-time. From the business point of view, a "one size fits all" message may be fast, but it is a very blunt instrument. The objective of the CSP is to identify the key "moments of truth" for each customer, and personalise the individual interaction in that moment.
Business users must then also be able to measure the business impact of their actions – whether that be on usage, churn, Average Revenue Per User (ARPU) or Net Promoter Scores (NPS). This requires a robust target versus control methodology, and the appropriate analytics and insights framework; none of which are available within traditional network elements.
To achieve this CSPs are evolving their business and marketing infrastructure to agile, real-time applications designed for this purpose.
Conclusion
CSPs are uniquely positioned to build an unparalleled understanding of their customers in real-time. They can then use that knowledge to personalise interactions with customers in the moment, unlocking the potential for a stronger competitive position, lower churn, and increased revenue and margins. But the key lies in being able to achieve the successful transition from big to fast data, and leveraging that tiny 250 millisecond window.
- David Peters is CEO and Founder of Emagine International
