This is an excerpt from our Trend Analysis Report on AI. We surveyed 187 executives from 76 communications service providers (CSPs) operating in 51 countries, and 115 executives from 56 supplier companies. We also conducted dozens of interviews with respondents.
Using artificial intelligence (AI) to enhance interactions with customers is the single most popular application today – and not just in telecoms. Gartner projects that more than 85 percent of all customer interactions will be managed without a human by 2020.
Close to half of respondents to our survey ranked improving customer centricity as the No. 1 driver for AI. The technology is emerging as an important tool to help customer care agents, and the customer-facing organization more broadly, transition from being reactive (handling complaints) to being proactive (cross-selling and upselling).
Amdocs classifies customer experience AI use cases as “intelligent care” or “intelligent marketing”.
Types of customer-centric AI applications
Many of these classifications are categories rather than use cases per se, but we expect to see a huge amount of innovation within the categories, particularly from CSPs that are developing their own AI capabilities. One example is Orange’s new service to alert customers when there is a problem in the network. The idea is to let customers know before they experience a problem via a message delivered by whatever medium the customer is using at the time.
In the Amdocs matrix, this could be described as a “surprise & delight” use case. One of the key drivers for this capability is mass deployment of IoT devices where a loss of connectivity can have serious consequences – a connected car, for example, that could cause an accident if connectivity were interrupted.
Make way for chatbots
The benefits of using chatbots can be clearly measured: Fewer calls into call centers mean staff can be reduced. But CSPs are also deploying bots to help customer care agents serve customers better.
Telstra’s approach is to put basic information into the chatbot but emphasize to its own staff the higher value of human interactions. The impact can be measured in improvements in Net Promoter Score (NPS) or any other key performance indicator (KPI) measuring customer experience. If customer care staff are also given a remit to upsell or cross-sell other products, impact can be measured in revenue gains.
In the survey for our inaugural Digital Transformation Tracker, which was published in September, only 6 percent of respondents said they had already implemented virtual agents or chatbots, but in the survey we conducted for this report, which was more detailed, that number jumped to 30 percent.
AI gets personal
Delivering more personalized services is an integral part of any program to improve customer experience. Personalization relies heavily on effective use of data analytics, but AI can also play an important role in augmenting analytics.
How to improve personalization
Suppliers are developing innovative tools to help CSPs improve personalization. Nuance, a leading vendor of chatbot solutuions, offers a service called Nuance Loop which continuously analyzes offer-conversion rates, subscriber profiles, content usage and network activity in order to more accurately create and place offers that are relevant to each subscriber.
IBM has a “Cognitive Marketing & Advertising” platform for CSPs that targets customers based on triggers such as location, weather, profile attributes and consumer behavior.
Many AI developers have built language translation capabilities. These can be extremely effective in helping CSPs target immigrant populations. For example, Microsoft developed a language translation service for Swedish operator Tele2 aimed at refugees who have just arrived in the country. It did this by integrating the Microsoft Translator API and Bing Speech API into the Tele2 network.
Using AI to fight fraud and cyberattacks
The use of AI is already widespread in the financial services sector to monitor fraud and security risks. McKinsey Global Institute reckons that AI-optimized fraud detection will be a $3 billion market in 2020.
TM Forum’s revenue assurance research indicates that revenue leakage is equivalent to 1.5 percent of a CSP’s revenue. Given that total CSP revenue globally is around $1.5 trillion, this means operators collectively are losing in excess of $20 billion a year. As such, it is somewhat surprising that CSPs have been slow to invest in AI-based fraud and security solutions.
Our survey reveals that less than a third of CSPs are starting to deploy AI-based fraud and security solutions or already have them in place. These findings were reinforced in our interviews, with very few CSP respondents saying they have solutions in place.
Swisscom is the exception. The company has invested in big data and AI-supported security solutions including classic IT-based security defenses against distributed denial of service and phishing attacks and malware, all of which have been operational since early 2016. The company also uses AI to enable nuisance-call blocking on both fixed and mobile networks.
Securing the cloud
As CSPs transition to cloud-based networks we expect to see increased activity in AI-enabled security. This is because in the past when operators bought hardware, they could set up restrictions surrounding the software to define who could access sensitive areas. The introduction of cloud-computing has increased accessibility, and by extension the risk of security breaches.
Furthermore, each time there is a software update (with cloud-based solutions software is updated regularly), there is risk of a security breach. Risks will be compounded when cyber criminals begin using AI, as inevitably will occur. The only way to counter this will be for security experts to develop their own predictive analytics and AI-based solutions. Given that IoT will be the first cloud-centric division of a telco business, this may be the first security-enabled AI solution for many operators.
A TM Forum Catalyst proof of concept called Empowering business assurance with artificial intelligence is using AI, machine learning and data analytics to better detect and prevent revenue leakage and ensure customers experience. The project, which is being championed by BT, Deutsche Telekom, Orange and Telia and was demonstrated at TM Forum Live! Asia in Singapore in December, is looking specifically at how these technologies can help address the growing complexity of revenue assurance and fraud management as CSPs transition into digital service providers and enablers.