Communications service providers (CSPs) have understood that digital transformation is critical to their long-term success. It’s the only way to manage the scope, scale and complexity of 5G networks and services, especially with flat headcounts and flat budgets the reality for the foreseeable future.
While many CSPs have actively pursued initiatives to adopt automation, artificial intelligence (AI), machine learning (ML) and — more recently — public cloud platforms, few have met with the results they hoped for. The road to becoming a true digital service provider (DSP) has proven harder than expected.
What’s the best way to avoid the pitfalls that have caused digital transformation efforts to stumble up to now? Based on real-world experience in virtually every region of the globe, Nokia has identified some clear principles of digital transformation success. All depend on viewing the transformation effort holistically across three interconnected areas — the network, processes and people — with a rigorous approach to balancing complexity and cost.
The need to adapt quickly to new traffic patterns and usage demands spurred many CSPs to accelerate aspects of digital transformation with the outbreak of COVID-19. At the same time, resource constraints — human and financial — have made it even harder for CSPs to enact their plans. Getting the equation right in terms of setting priorities and managing costs is more important than ever.
Why digital transformation fails
According to McKinsey, seven out of 10 digital transformations fail. In our experience at Nokia, there are some common reasons for this — pitfalls that even the best-intentioned CSPs fall into.
The first is trying to change everything at once. While digital transformation can ultimately touch every dimension of the business, mammoth initiatives are hard to propel and often grind to a halt. This is especially so now, after years of cost-containment measures have eroded the resources for large-scale, sweeping transformation efforts.
A second pitfall is trying to apply digital transformation to an architecture that simply isn’t ready for it. Legacy environments are siloed and segmented across the stages of the network lifecycle — design/deploy, support, operate. It’s hard to scale up automation and streamline workflows in a structure where there’s no integrated, system-wide view of overall network performance.
Equally challenging are attempts to automate obsolete processes or ones that are poorly defined or poorly understood, out of context of the end-to-end view of the network and services. The siloed structure of legacy CSP organizations often leads teams to develop inconsistent ways of diagnosing, defining and resolving network issues. While each fix works on its own, the result is a patchwork that further prevents system-wide performance improvements. Further, traditional waterfall decision-making processes are inherently slow, linear and inflexible. They don’t provide the agility to test, experiment, discard and adapt, making it tough for transformation initiatives to advance.
Finally, there’s the inescapable reality of human factors. Even with the right technology and processes, momentum is hard to achieve if leadership does not communicate goals clearly, consistently and often, or if activities don’t sufficiently engage affected teams. Engineers focused on their own specific areas may not appreciate the big-picture benefits of automation and so stick to their ways of doing things. Others may distrust automation and AI — and even overtly resist them unless they have a role in defining how they’ll be used.
In general, change creates uncertainty. Teams may resist transformation outright if they feel the results such projects deliver put their own jobs at risk.
Lessons learned from real-world transformations
Nokia has helped more than 75 CSPs implement digital transformation projects in the past few years. That experience has revealed some key ways to avoid typical transformation pitfalls that occur in relation to the network, processes and people.
Network: Implement a digital network architecture
Digitalizing the network architecture is a fundamental step on the path to transformation because it supports and facilitates so much else. It should be done with open standards and the aim of breaking down operational silos — making possible integrated, end-to-end control over the entire service lifecycle with a single view of network performance, tightly orchestrated updates and common business rules.
Modern tools and pervasive, open APIs enable the automation of routine tasks easily and at scale. It’s critical that CSPs look across their entire network lifecycle to identify which tools need to be modernized and which processes are now obsolete, because incorporating legacy tools or outmoded processes into new digital architectures will inevitably create roadblocks as networks grow more complex.
Through machine learning, data from automated tasks can drive an ever-growing and continuously learning AI that engineering teams can use to automate a broad range of critical network functions. Over time, CSPs can build a shared library of standard operating procedures for future transformation projects, each with its own repeatable workflows, technologies and related resource requirements.
In Nokia’s experience, CSPs that embrace the use of digitized methods-of-procedures (“DigiMops”) experience significant improvements: up to 50 percent faster site build times, 40 percent faster software upgrades, and similar gains in overall operational efficiency. Injecting AI use cases via a common cognitive framework brings even more benefits: 50 percent faster average data speeds, as much as 60 percent less video streaming buffering, and 20 percent lower energy consumption through machine learning.
The cognitive framework can also be used to diagnose network performance problems. Using consistent definitions and automating the diagnostic process makes it possible to compile a set of repeatable solutions to solve future issues in less time and at a lower cost.
Process: Act like a startup
By replacing waterfall decision-making with an agile DevOps approach, CSPs can extract the full value of digital transformation more quickly. DevOps processes are adaptive and take advantage of programmable software so teams can make adjustments mid-course instead of having to restart the entire process. Cross-functional collaboration, programming and testing are iterative, occurring in the same cycle and lasting a few weeks – rather than months or years. This way of working is not only key to transforming successfully but also to handling the complexity of 5G and hybrid network environments.
CSPs can further increase their agility by extending network operations and services into the public cloud. While there’s historically been reluctance to extend even non-core functions beyond CSPs’ own walls, this is changing. In 2019, Vodafone chose Google Cloud to host its real-time data analytics for customer service enhancement, network planning and optimization, and personalization activities. Earlier that year, Microsoft and AT&T launched a multi-year collaboration to create new services using their combined infrastructures as part of AT&T’s “public cloud first” strategy.
By embracing a public cloud strategy, CSPs can experience the same benefits that other industries have been enjoying for years — specifically, shorter cycle times, greater scalability and lower costs. It’s also a first step toward developing a robust IoT ecosystem that CSPs can monetize through open APIs. In an open 5G environment, third parties can use the assets of the network to create solutions for consumers and enterprises that the CSP can co-generate revenues from. The more valuable the services, the greater the potential new revenues. Emerging technologies such as blockchain can provide a secure and shared ledger to track logistics or drive monetization.
With their 5G networks poised to power new data-driven solutions for new enterprise customers across a broad range of industries, CSPs can position themselves to increase their strategic value to broad range of third parties and end customers.
Digital transformation drives continuous innovation and learning
People: Engage and empower everyone
Effective change management is essential to drive lasting transformation. CSPs can mitigate uncertainty and boost employee engagement using a combined top-down/bottom-up approach, with senior leadership communicating new goals and related key performance indicators (KPIs) clearly, honestly and frequently. Engineering teams should have a hand in guiding the development of new projects and deciding what to transform first. Employee buy-in and trust are both more likely when affected teams have a voice on the transformation team.
At Nokia, we’ve seen this firsthand. A European CSP we work with enabled its radio engineers to select, test and implement a new automation solution themselves and the outcome had significant uptake. Our own engineering teams have also had a hand in training Nokia’s supervised-learning AI algorithms.
The majority of operations tasks still need human experts to provide input and make decisions. Moving to fully autonomous, closed-loop systems is a big jump. But CSPs can build trust in new solutions if they proceed step-by-step from the start and provide clearly articulated goals and control mechanisms that allow for human intervention along the way.
Where to start?
CSPs should aim to focus digital transformation efforts where they stand to make the greatest gains soonest. Network expansion, operational streamlining and performance optimization offer promising starting points. But where exactly within those domains should activities focus? A structured approach helps answer that question and set priorities while avoiding the pitfall of “too much at once” and provides a means for weighing the relative cost and complexity of transformation decisions (which are related, with costs climbing as network complexity grows).
Employing a structured program/methodology that utilizes cognitive use cases with AI and machine learning capabilities will help CSPs determine where to concentrate their digital transformation efforts for the best possible returns. This kind of analysis gives CSPs empirical, objective direction, clarifying the business case and avoiding assumptions.
Some CSPs are surprised to learn that the tools of digitalization can deliver benefits not only in the inherently digital parts of the business (e.g., operational and business support systems), but also where physical assets and infrastructure are concerned. This matters because network planning, deployment and the acquisition of spectrum assets are key CAPEX (capital expense) drivers for CSPs. Some CSPs have used drone footage and data from geo-spatial, network and environmental systems to build digital twins of their base stations that allow them to plan upgrades without having to re-survey the same site, saving time and money. And by applying AI to network planning and multi-layer performance optimization, it’s possible to improve the returns on CAPEX investments by up to 30 percent.
Power consumption and field maintenance are among the top contributors to CSPs’ OPEX (operating expenses) — with power consumption growing at an eight percent compound annual growth rate. Cognitive applications that control off-peak power consumption can deliver up to 20 percent cost savings on energy. Predictive algorithms for field maintenance can cut OPEX by five to 20 percent.
AI and automation can also boost the efficiency of Network and Service Operations Centers (NOCs and SOCs). For example, intelligent alarm clearing can reduce the number of alarms engineers need to manage — from several million to a few hundred or thousand each month. This can also speed up the resolution of network issues, shrinking mean time to repair (MTTR) by more than 80 percent. By applying AI and automation across their NOCs and SOCs, CSPs have been able to improve overall operational efficiency by up to 40 percent.
Optimizing network and service performance increases subscriber satisfaction, reduces churn, and can also lead to increased revenues. For example, through AI we can provide an end-to-end view on the Quality of Experience (QoE) for video streaming or cloud gaming services. Machine learning is then used to create a library of ‘signatures’ that link poor QoE to underlying network issues, and provide automated recommendations to guide engineers on the best resolution paths. Such an approach can increase annual revenues by up to 5 percent for the service in question.
Automated cognitive applications can also give CSPs more detailed and more accurate insights into network coverage, capacity and quality than are possible through manual drive tests, while reducing their greenhouse gas emissions at the same time. Instead of a mere snapshot of system performance along a drive test route, automated applications can reveal details of subscribers’ experiences inside buildings, and contrast network performance in the same location at different times of day — among many other factors — with no trucks required. These new insights make it possible to predict and resolve issues before they become major problems.
Every CSP must build its own path to digital transformation — one that includes the network, processes and people. Network systems must be digitalized through a common reference architecture, prescriptive and repeatable workflows and smart automation. Processes must be more agile and responsive to keep pace with changes in consumer demand and business conditions. And people must feel engaged in making these changes happen.
By applying a structured approach to transformation and weighing the relative cost and complexity of various options, CSPs can stop trying to do everything at once and instead do the right things to realize their full potential. There’s no doubt COVID-19 has made digital transformation both more urgently needed and more challenging. It’s also created the opportunity for CSPs to reset their projects, take a new approach and truly set themselves up for 5G success.