Home / Expert Answers / Operations Management / need-a-case-study-analysis-of-the-vodafone-managing-advanced-technologies-artifical-intelligence-and-pa649

(Solved): Need a case study analysis of the Vodafone Managing Advanced Technologies Artifical Intelligence and ...



Need a case study analysis of the Vodafone Managing Advanced Technologies Artifical Intelligence and provide a short description about ChatBot
student submitted image, transcription available below
student submitted image, transcription available below
student submitted image, transcription available below
Chatbots The center's automation teasn provided services to operations manager where they indicated pressing needs. Chatbots wete one of the most prevalent automation requests. This reflected a global trend, as Oracle reported that 805: of busincsies wanted chatbots by 2020, while Orbis Research predicted that the globul chatbot market would grow at CACR of 37\% during 2017-2021. According to Juniper Research, chatbots would cut business costs by billion by 2002 , to When the automation center received a request, the first step was to understand the operation and the key use cases the operations lead wished to automate. Brunet explained, People needed to understand why they wanted to do this. Was it because of poor quality of service, or because of a cost issue, or both? Then we applied some templates that we shared in order to experiment with varying automation scenarios; one journey could require a high development workload for a low return while others although cost eifictive could negatively impact customer satisfaction. This technology has a cost, business leaders need to understand the potential outcomes," Once the use case was agreed upon, the automation team built a high-level technology design outlining where it needed to interface with the rest of the information systems. Then a two-site agile team worked in sprints of two weeks to co-develop the chatbot. Brunet said, "My teams were coding. but the business was driving the transformation in the meantime. We put it on a platform for it to go through iferative festing. It required a lot of tweaking because it's machine learning -it's like a child's brain in many ways, it learns faster but you need to take it through logical learning steps." The business identified a Subject Matter Expert (SME) as a digital coach to fine-tune the robot. Brunet explained, 'When the robot can't answer, it hands over to a human agent who answers the request and the robot learns from it, But we want somebody validating that what the agent did was correct and that the robot can learn from the right approach." The business then decided how to take it into production. Brunet illustrated, "For example, did they want to do it with a small perimeter and a number of limited users, 7 Vedufone: Mandging Advamsed Tochnelogies and Artificial Intelligence or take it to a wider audience to better measure potential outcomes?" (See Exhibit 9 for workfiow diagram.) Anelia In early 2016, the center completed its fins implementation with the introduction of chatbot 'Amelia' to deal with the support requests of its internal IT service heipdesk. Wibergh explained, "We basically started testing and experimenting on ourselves. We looked at requests that our IT service helpdesk receives such as ' 7 have forgotten my password' or 'T am locked out of my account' and thought if we could replace IT service agents with a chatbot, we could reduce the number of people in the internal helpdesk in India. From an efficiency and cost point of view, it is an interesting case to look at because the labor cost is still low in lndis, so how can you do it better? It was good to start internally; only internal people can get mad and not customers " Vodafone performed this development in-house. Brunet shid, "As soon as we deploy, scale is an issue, because Vodafone very quickly became the biggest platform in the world Instead of giving the work to outside vendors with limited experience in this emerging area, we recruited our own cognitive engineers, humanizers, etc. and I worked directly with the softwafe editor, IPsoft. We agreed that IPsoft and Vodafone would combine resources in each agile sprint," Vodafone started experimenting with some of the user journeys before taking the product live in December 2016. It further expanded and integrated the chatbot into seven of its back-office systems. By early 2018 , Amelia solved about half of the problems it encountered, equivalent to about 160 IT service helpdesk agents. (See Exhibit 10 for screenshots of Amelia.) TOBi Wibergh continued, "Then we asked how we can apply this to our customers. If you take a step back and look at us as a company, we have around 500 million customers and around 70,000 call center agents, both employees and external contractors. We have to continually make customer interaction better. One thing was bringing customer service online and making sure it was a great experience in itself. Then if we could replace some of our call center agents with an intelligent chatbot agent, we would increase quality and reduce costs." Vodafone pursued three pilots in parallel in three countries with three different vendors: IBM's Watson in the U.K., Microsoft in Italy, and IPsoft in Ireland. Brunet said, "The purpose of the pilot is to learn and begin benchmarking." Wibergh added, "Part of the experiment was picking which one to work with given the immature technology. With these three vendors, we implemented roughly similar use cases. We also experimented lifting the one we did in the U.K. to Spain to learn the complications of moving langtages and similar. Then we did the commercial negotiations. When you have multiple companies competing, you can make a better commercial deal. We chose IBM's Watson and Microsoft Chatbot, and other countries are now able to deploy. In this instance we decided to outsource given IBM's bigger service business. At the same time we have also built up internal competencies." Vodafone's chatbot 'TOBi' was launched in April 2017 and provided a fully integrated web chat with the customer. Initially focused on popular support questions, TOBI's role was expanded to answer customer-specific questions such as data-roaming charges and phone capabilities. Upon its launch TOBi was able to answer about 112 intents; just a few months later, he could respond to around 150 or about of typical queries from customers. To avoid frustration, the technology used a 'sentiment' function to pass users seamlessly onto a human advisor if the bot could not help or if they were not satisfied. Messages from TOBi and human agents remained in the same thread, allowing agents to read the past history and avoid customer repetition. Vodafone planned to bring its customer service capabilities to Amazon's Alexa, to let customers ask Alexa about their phone bills and the amount of data they have used, for example. Brunet reflected: We learned some significant lessons. When we first started, we reproduced human dialogue, which was wrong because people don't talk to a robot the same way they talk to a person. People also try to beat the machine, asking 'What is the turnover of Vodafone?' or 'How old are you?,' and you need to handle that. I insist that the SME for the agile team be empowered on changing or tuning the process. If 20 meetings are required for approval, you don't have an agile team anymore. So that's a big responsibility for them and this means that we need to empower business owners to optimize the process for a better outcome. There is joint learning - the business needs to learn about the technology, and we need to learn about the business. Both teams need to win mutuai respect. We do daily stand ups, and our video conferencing is on constantly. Version control and change management are very important; we have had to remind the software editor that we were not in a startup mode anymore. But in the meantime the platform provided by the editor was evolving so fast that we wanted to keep our development on the latest release to benefit from the latest functionalities. This was in production at Vodafone, so they could not just decide to change something and not tell us nor could they skimp on testing and cause our business to fail. Wibergh further cautioned, "People think it's faster and easier to introduce this system than it actually is. It is very trendy, many companies are doing it, but it is complex to get it up. We have several in production, and we have done several prototypes, You can often have negative Net Promotor Scores (NPS) at launch, which surprises people. The technology still needs time to mature." A team continued to build Vodafone's consumer products and services for the local markets as a group function, with particular interest emerging for Alexa, Google Home and Microsoft Cortana. Lange-Richter explained, "We are looking all around, exploring, and sharing ideas. This topic is so exciting that people really get into it. Customers are getting used to interacting with a robot and then they say, 'Why can't I maybe call a Vodafone hotline via Alexa?' You need to be quick: It took us two weeks to launch the first skill on Alexa, and in the U.K. it is already used. Long-term, it's not clear yet if chatbots will replace or complement the hotline. We are still observing." The customer products and services team worked on a workflow library, where countries could also see what others had created as a user flow for a chatbot. Lange-Richter said, "The customer intents are a bit different from country to country, but the topics are similar. There should be some reusability across all these different flows, and our library makes them retrievable. I personally think chatbots will empower customer interaction with companies at home. Our key interest is to set up a framework so that the customer care team can easily build their own flow, looking to us for central technology. Such a scalable framework would allow different customer operation teams across all Vodafone countries to do it on their own, and also leverage and learn from each other." In the customer care team, the business case was based on the number of chats with resolved cases that did not require any contact with an agent. Lange-Richter said, "Out of the volume of chat we anticipate, what percentage are hopefully resolved within the chat? How much volume can we take away from the classical contact channels and what is the saving?" The main KPI that the customer care tean looked at was the NPS. Lange-Richter said, "The good thing about this technology is that you can configure and easily change the flows; you do not need six-month release cycles. The NPS is benefiting from the speed we can have in adapting. We can observe and act quickly."


We have an Answer from Expert

View Expert Answer

Expert Answer



We have an Answer from Expert

Buy This Answer $5

Place Order

We Provide Services Across The Globe