DSPs can now complement streaming data-driven insights with real-time actions
Nikos Tsantanis, Senior Product Marketing Manager, Intracom Telecom
Spyros Sakellariou, Telco Software Marketing Section Manager, Intracom Telecom
As the transformation of Communication Service Providers (CSPs) into Digital Service Providers (DSPs) is well under way, operators concur, among others, on a main pursuit: how to optimally leverage all interactions –physical or electronic – that their subscribers have with them in order to improve customer experience, maximize loyalty and, eventually, achieve better commercial results.
This dominant objective may sound compelling enough, until the concept is applied in practice. It is only then that DSPs may realize that a key limiting factor is their ability to continuously figure out and carry out the ‘Next Best Action’ toward every one of their millions of subscribers, in a manner that is simultaneously:
- Fast. Initiate actions almost in real-time, while the customer still remains engaged.
- Omnichannel. Be able to include all available interaction and service delivery channels, recognizing those that individual customers favor.
- Comprehensive. Take into account as many aspects of the subscriber as possible, including current ‘situation’ (geo-location, time, device, language, etc.) and historical interaction patterns.
Welcome to Contextual Marketing, the art of determining, in real-time, the optimal action based on the context of each individual customer. Evidently, DSPs are ideal industry players to implement this notion as they are involved, more than anyone else, with systems, traffic and subscribers and everything in-between.
However, when thinking of Contextual Marketing, many DSPs overemphasize the importance of the Campaign Manager, the system serving as the ‘last mile’ in delivering marketing propositions to end-users. Unfortunately, even the best Campaign Manager does not have by itself the ability to intelligently distinguish those events –among myriads– that will help it reveal who to target, with what messages and when exactly. Without a doubt, the ambition for personalized Contextual Marketing often degenerates into just conducting batch campaigns, aiming at broadly-defined market segments, an old-school technique of limited effectiveness.
Harnessing the terabytes of streaming data to trigger in real-time the Campaign Manager is unfeasible with conventional methods because of a number of complications, such as:
- DSPs’ typical infrastructure landscape is quite extended and encompasses channels and systems as diverse as CRM, website, web portal, POS, Billing, customer care, voice and data probes, fixed/mobile networks and 3G/4G antennas, to name just few of them.
- Raw or transient data volumes generated daily from all interactions are enormous, easily reaching practically unmanageable levels.
- The path from data sources to the Campaign Manager may be either missing or involving too many steps, resulting in data transmission delay.
The combination of these factors raises two challenges:
- Focusing only on data and events that are interesting, valuable or actionable
- Discovering and interpreting subscriber’s context via the ‘distilled’ data.
To overcome them, two massively scalable functional components need to mediate between data sources and the Campaign Manager.
Event noise reduction
Compacting data to the absolutely necessary subset, involves two parallel actions:
- Filtering. Select only the data necessary for recognizing the subscriber’s context. For instance, a DSP may only need to consider that part of total data volume that is associated with social media, while discarding data traffic due to, say, visits to generic websites.
- Aggregation. Sum up piecemeal indicators into cumulative values, yielding big picture business event overview. For example, the essence of a 30-minute voice call, generating lots of CDRs, may adequately be summarized in just few bytes indicating, say, calling parties ids, call duration and QoE metrics, such as dropped calls.
The scope of this part of the process is to forward to the next component only relevant, noise-free data and events, which are much more compact and manageable.
Context & trigger discovery
This is a business-oriented component which enables marketers to define rules that narrow down the filtered and aggregated events received by Event Noise Reduction into meaningful, marketing contexts pertinent to the type of promotional campaigns configured in the Campaign Manager. Such contexts may include, for example: churn alert, sports event, customer complaint, customer satisfaction below threshold, prepaid account recharge, etc.
Use Case: IPTV contextual marketing
Let us see how all parts described so far fit together in a real-life scenario illustrated on the diagram.
- Anne maintains a basic triple-play subscription package with her DSP, allowing her to watch her favorite content, mainly films and series that are best viewed on HD. Tonight, Anne unpacked her new 4K UHD TV set and downloaded the DSP’s IPTV app.
- While waiting for her TV to setup, Anne searches through her tablet the DSP’s portal to find out how she could upgrade her ADSL line to VDSL.
- The DSP’s Context & Trigger Discovery system captures the IPTV app download event and the search event and classifies them as relevant to Contextual Marketing.
- Through business rules, predefined by the DSP’s Marketing Department, Context & Trigger Discovery system discovers that the search event is in the “on-demand TV series and movies media” context and passes it on to the Campaign Manager.
- The Campaign Manager identifies three alternative promotional campaigns because Anne is currently visiting a specific webpage:
a. Upgrade to VDSL for a free 3-month trial subscription
b. Upgrade to a VDSL subscription and get a free sport channel bundle
c. Upgrade to a VDSL subscription and get unlimited 4K UHD content for free
The system automatically determines that, for Anne’s particular context, the Next Best Action is option (c).
- Anne is pleasantly surprised by an automatically-created banner proposing this offer and decides to buy, as it perfectly matches her needs.
Concluding, DSPs can complement streaming data-driven insights with real-time actions via Intracom Telecom’s Big Data platform, BigStreamer. BigStreamer comprises a Complex Event Processor, capable of handling the massive filtering and aggregation requirements of Event Noise Reduction, a Distributed Decision Engine supporting Context & Trigger Discovery in real-time, as well as an advanced Campaign Manager. DSPs can choose ‘a la carte’ one or all of these modules.