Communications service providers are constantly looking to improve business efficiency and performance. However, today they face a number of challenges including competition from other operators and over-the-top players, the need to create new service offerings to better diversify their business, and ongoing operational management cost challenges and inefficiencies. Against this backdrop, CSPs are now looking to big data technologies and advanced analytics tools to exploit the untapped reservoirs of data that exist within their networks and IT systems.
The fine-grained data that service providers can collect from network probes, intelligent sensors, and other data sources are the fuel that powers big data analytics and provides new insights into ways to improve customer experience, optimize networks, drive greater operational efficiency, and enable broader service offerings. By deploying big data solutions – including traditional data warehouses – CSPs can analyze broader types of data to ensure that the business becomes more agile. However, a key obstacle that remains is making sense of this growing mountain of data in meaningful ways that will impact the business.
CSPs must address the following technical challenges before they can truly tackle next generation analytics use cases.
Creating a Standard Enterprise Data Model
Before they can deploy next generation analytics solutions, CSPs must utilize their existing data warehousing solutions more effectively and also understand the data they are collecting. While many CSPs have existing data warehouses, these warehousing solutions are nearing the end of their useful lifecycle and are barely able to manage daily operations, let alone meet tomorrow’s challenges. This problem is compounded with the complexity of CSPs’ system portfolios, which typically consist of hundreds of systems supporting multiple lines of business. To make matters worse, network and customer data is often managed by different organizations and retained in multiple incompatible systems with no way to perform a cohesive analysis. Once providers get a handle on existing data warehouses, next generation enterprise data models should feature:
- standardized metric definitions across the business – allowing for all stakeholders to speak the same language;
- model alignment with industry standards such as the TM Forum (News - Alert) Information Framework to support CSP business processes and time to market for use case development by providing standard definitions for all of the information that flows through the enterprise and between service providers and their business partners;
- transaction-level access; and
- comprehensive metadata for business intelligence reporting and ad hoc query along with industry-specific advanced and predictive analytics.
Big Data Adoption
How should CSPs build a big data platform? It often starts with a data reservoir, typically based on Hadoop, which can economically capture, store, and process a wide range of data – from web logs and social media to network information. While these sources are typically unstructured and difficult to understand, this data provides valuable insight for businesses beyond typical data sources. Big data platforms provide serious advantages to providers, including:
- a highly flexible and distributed architecture that supports parallel processing to parse large data sets;
- high-volume data acquisition;
- less expensive data storage and analysis;
- analysis of structured and unstructured data; and
- real-time analytics and event processing support.
All of these big data benefits also come with an array of challenges. CSPs must address several key issues including:
- consistently managing data in an open storage architecture;
- custom/ad-hoc development on big data sets can be impractical;
- bridging the metadata gap for downstream applications across myriad data formats, network elements, and applications;
- lack of tools to support data retention strategies;
- diverse needs of resource skillsets to implement and manage all the individual components; and
- correlating (KPI on KPI) multiple sources of data across the batch and streaming analytics tools within a big data architecture.
So how can services providers address these challenges to experience the benefits that big data can provide?
Integrating Big Data and the Enterprise Data Warehouse
It’s one thing for Hadoop experts in a corner of an organization to work with this new data, but to really create value and innovate, it needs to become part of the everyday business. This means integrating big data with existing data, processes, and tools – like the data warehouse, business analytics, and reporting. It’s not as simple as moving all data into one location, but rather minimizing data movement and working with all data in the same place as much as possible. Both the reservoir and the warehouse must work together seamlessly, securely, and quickly. Basically, all data should work together as a well-oiled machine – or data factory – using both the data reservoir and warehouse to organize and transform data assets to make them more useful and easier to analyze. For example, combining customer information from the data warehouse, with website activity from the reservoir, might yield new insights about buying habits. CSPs can exploit this insight by feeding it back into their business to improve marketing programs. The data factory enables providers to act upon real-time data streams, perhaps from industrial equipment or mobile devices, and respond to inquiries or new information as it happens.
Once CSPs address the technical challenges behind next-generation analytics infrastructures, they can look to more innovative ways to monetize this data through revenue growth or operational savings. Key areas impacted include:
- Network Operational Improvement: As next generation services such as VoLTE, Wi-Fi calling, and high-definition video become ubiquitous, CSPs will need to collect all relevant network data to ensure seamless delivery. Traditionally, they have stored network data such as signaling, policy and charging, and deep packet inspection in silos. End-to-end analytics will require correlation of this data and the resulting analytics delivered to each stakeholder in the manner in which they need to consume it. For example, the customer care team needs a near real-time view as they troubleshoot a customer complaint; while marketing will want to understand how Diameter congestion or policy failures impact specific customers. To process this mountain of data, CSPs will have to implement scalable analytics platforms capable of handling millions of transactions per second.
- Customer 360: CSPs worldwide are searching for better ways to hang onto existing customers while finding and securing new ones. They are seeking ways to transform their business to a customer-centric business model and know they need to deliver more customer-centric solutions. They need a 360-degree view of customers so they can more effectively personalize new products and services in both the short and long term. However, the truth is that it is difficult for carriers to get actionable information fast enough to make on-the-spot decisions. While CSPs often have a wealth of basic data on customers and prospects, the CSPs’ current systems may lack an effective way to leverage this data for rich analytics. The systems may also lack historical data about the customer's interactions over the lifespan of the relationship. To better understand the customer going forward, CSPs must be able to transform raw customer data into actionable knowledge. Real-time metrics at the point-of-customer interaction – such as the customer calling customer care or accessing self-care portals – is critical in understanding appropriate retention, up-sell, and cross-sell actions. In addition, traditional churn influencers such as bill shock, dropped calls, and customer relationship management data can now be augmented with data from deep packet inspection systems, network signaling, network policy, network usage, and social networks
- Data Monetization: Today’s chief marketing officer is looking for new ways to monetize CSPs’ data. Part of this monetization comes from extracting value from data previously trapped in silos and using it to more precisely segment the customer base to support retention and promotion actions. The more real time the data is, the more accurate and valuable the data is to internal teams. Additionally, providers are looking to sell this customer data (detailed or anonymized based on local regulations or opt-in selections) to third parties for a variety of use cases such as targeted advertising. Often, CSPs will work with analytics partners to analyze and sell the data on their behalf in exchange for a share of the revenue. Sponsored data has also become more mainstream lately, with providers announcing programs where content providers are billed for each gigabyte they serve to consumers. Utilizing near real-time analytics capabilities, a CSP (News - Alert) will have an accurate view to customer metrics that are important to their partners.
As the telecommunications market maintains its rapid pace of change, CSPs need to ensure they are building consistent and coherent analytics solutions that are designed to solve their business problems in real time. However, they often face challenges when it comes to adopting new analytics technologies and integrating them with existing data warehouse. Providers require next-generation analytics platform and solutions to provide customers with the speed and level of service that they have come to expect.
Tracy Melton is analytics product marketing director at Oracle Communications.
Edited by Maurice Nagle