Like many businesses, wages are the biggest single cost item for most telecom operators. Anything that can be done to replace humans with software falls straight to the bottom line.
A key goal of NFV is to automate back-end processes such as service fulfillment and assurance in a "closed loop," minimizing human manual intervention. In traditional OSS fields, such as network planning and fault management, operators are also looking to increase automation, drawing on techniques such as machine learning (ML). Similarly, automation opportunities abound within BSS, notably in customer care.
For example, at the recent TM Forum Live event, Telstra Corp. Ltd. (ASX: TLS; NZK: TLS) presented its estimate that 30% of inbound calls to a contact center could be resolved by artificial intelligence (AI) chat bots. To take this to the next level, Telstra is looking at using text sentiment analysis to enhance the performance of its messaging and chat agents.
Similarly, Bell Canada championed a catalyst at TM Forum Live -- Sentimental Applications -- that used sentiment analysis of a customer's speech while navigating Interactive Voice Response systems. The analysis was used to measure and track customer experience and help decide an automated next-best-action.
The role for RPA in telecom
CSPs have used IT for decades to automate business process. However, ever-changing business needs and technology has meant that many telecom organizations have ended up with a portfolio of "siloed" applications supporting a patchwork of business processes. These processes require high levels of manual intervention leading to high opex, low agility, data inconsistencies and unsatisfactory customer experience. Attempts to streamline processes through large transformation programs have often taken too long, cost too much and delivered fewer than expected benefits.
Robotic process automation (RPA) can fill the gap between big bang transformation programs and existing manual operations by automating disparate processes at a lower cost. RPA is most effective for processes that require predictable and high frequency interactions with multiple applications. Telecom operations include many mundane and repetitive but essential processes that require multiple systems to be queried and/or updated to complete the task. The tasks must be completed reliably and accurately, making the telecom industry a textbook case for robotic process automation. However, to date RPA has seen greatest traction in the financial services industry, not telecoms.
Nonetheless, telecom providers are starting to leverage RPA to reduce costs, improve data quality, boost customer service and drive significant improvements in operational efficiency. A study by the London School of Economics showed that of the 160 different processes that Telefonica's UK subsidiary had automated, the payback period of the investment through headcount reduction was just 12 months. Similarly, AT&T is using RPA as virtual assistants to boost the productivity of its customer service reps.
RPA can be a cost-effective alternative to business process outsourcing of back-office functions. It can also be used for activities that don't require much judgement (though may involve following a set logic), such as logging in to multiple network management systems in order to provision a particular service. Think of it like a macro that needs one click instead of 30 to perform a task that may need to be carried out regularly.
RPA combined with AI can significantly increase the scope for automation
But to make RPA truly valuable to telecom operators, and other enterprises, it needs to be combined with some sort of AI or cognitive learning capability. This is especially true in networking, as the skills of highly trained engineers are not easily turned into a set of simple instructions that an RPA engine can follow. A cognitive engine, however, can learn from how people take decisions on complex data sets in order to create its own rules, rather than have them explicitly programmed. Indeed, the huge data sets that network teams collect, from probes and management systems, can be difficult for humans to analyze using standard tools. AI may reveal new insights, enabling operators to run their networks more efficiently and reliably.
For example, Zhejiang Telecom has implemented an AI engine to assist with route optimization, capacity planning, traffic prediction and dynamic optimization of the network. This led to an 8% increase in routing optimization and the identification of many vulnerabilities that had previously been undetectable.
During the coming months, Heavy Reading will be researching the potential impact of RPA and AI in telecom operations. We are keen to discover which areas of the telecom operations value chain have the greatest potential for RPA -- network rollout, fulfillment, assurance or billing -- and identify those areas where there is the greatest scope for AI/ML to complement RPA -- routing optimization, resource optimization, security, or something else?
We'll also examine the obstacles to the adoption of AI-enhanced RPA in the telecom industry. For example, as human work changes from manual troubleshooting to automation development (coding), this will require retraining. However, ultimately the service operations center of the future is likely to provide more satisfying work than is typical today.
— James Crawshaw, Senior Analyst, OSS/BSS Transformation, Heavy Reading