《电信管理论坛(TMF):2023反竞争情报报告-利用人工智能改善客户体验(英文版)(37页).pdf》由会员分享,可在线阅读,更多相关《电信管理论坛(TMF):2023反竞争情报报告-利用人工智能改善客户体验(英文版)(37页).pdf(37页珍藏版)》请在三个皮匠报告上搜索。
1、REPORTAuthor:Sponsored by:Teresa Cottam,Contributing AnalystDawn Bushaus,Contributing Editorcounterusing AI to improve customer experienceintelligencecontentssection 2:limitations and challenges of chatbots and other AIsection 1:the evolution of chatbots and other AI in CXthe big pictureadditional f
2、eatures&resources 09050327section 4:using AI in customer supportsection 3:using AI to improve customer and employee experience141218section 5:using AI to support process improvement2inform.tmforum.orgsection 6:what makes AI projects successful?21section 8:make it happen strategies for successful AI-
3、empowered CXsection 7:whats next for AI in CX?2523the big pictureBut while theres huge hope that AI can transform the way companies support and interact with customers,there is a danger that it is being over-hyped.Communications service providers(CSPs)also need to be aware of possible pitfalls as th
4、ey deploy AI in customer service and to support customer experience(CX).From promises of automated AI-empowered experiences,to fears about Skynet-like apocalypses,the(soon-to-be AI-generated)news is jammed with stories about AIs capabilities and how the technology will transform business.Fear of mis
5、sing out has resulted in a wide range of companies pouring millions into AI-enabled projects many of which have not delivered against expectations.Indeed,success rates for such projects could be staggeringly low.Back in 2018,Gartner made a widely reported prediction that 85%of AI projects would fail
6、 due to over-hyping,poor scoping and data issues.The research firm further forecast that only 54%of projects would make the transition from prototype to production-ready application.Nevertheless,AI adoption has more than doubled in the past five years,and investment and interest in AI projects is co
7、ntinuing to increase in many industries,including telecoms(see this analysis by McKinsey).CSPs are using AI across their businesses,including to improve customer service and experience.They are using chatbots as a new channel to support and enhance self-service channels and to optimize the work of h
8、uman agents.AI is also being directed at common causes of customers calls into care centers such as questions about billing Breakthrough innovations in machine learning,natural language processing(NLP),sentiment analysis and generative AI are making 2023 the year of AI.The race to adopt these techno
9、logies is illustrated by the rapid success of OpenAIs chatbot,ChatGPT-3(now at version 4),which amassed more than a million users within five days of launching.3inform.tmforum.org4the big pictureinform.tmforum.organd service quality and it is making processes like fraud management and marketing more
10、 efficient.To maximize the successful application of AI,however,it is critical that CSPs close the gap between expectations and execution.They must ensure that projects are designed to solve real business problems,return tangible value and meet the expectations of customers.Not doing so will likely
11、result in projects ending up on the AI trash heap.Read this report to understand:How CSPs are using AI to improve employee experience(EX),reduce fraud,personalize CX,improve the handling of billing and service quality inquiries,and optimize marketing initiatives.How to gain real business value from
12、AI How chatbots have evolved The pitfalls and challenges of adopting AI What separates AI success from failure Whats coming next in AI.Applications and benefits of using AI in customer support TM Forum,2023(source:Omnisperience)Lower costsImproved EXBusiness integrationImproved lead generationIncrea
13、sed efciency&faster response times24x7 support&predictionBetter foresightcustomer insightsFiner&deeperomnichannel experiencesCoherent&seamlessdigital confidenceHigher levels ofexamples ofhow AI can beused in CXSmarter chatbots&self-servicePersonalized,contextualized sales&marketingPersonalization&pr
14、oactivitySafety&securityAgentaugmentationAgent training&onboarding65section 1:the evolution of chatbots and other AI in CXinform.tmforum.orgA chatbot is essentially a computer program that uses a variety of technologies including natural language processing(NLP),sentiment analysis and machine learni
15、ng to understand questions and provide answers or actions.The technology is embedded in everything from smart speakers like Apples Siri,Google Assistant and Amazon Alexa,to custom-designed chatbots supporting telecoms customers and call center agents.The first generation of chatbots(see graphic on n
16、ext page)used very simplistic NLP models that responded to keywords with limited,scripted answers.Unsurprisingly,customers were CSPs are deploying chatbots to assist with customer support,and they are using AI to boost the performance of key processes within customer service and experience.1.Custome
17、rasks question2.Chatbot analyzesrequest,determines intent3.Chatbot queriesknowledgebase4.Chatbot responds or routes customer to more help5.Chatbot feeds new insightback to knowledgebaseHow chatbots workTM Forum,2023(source:Omnisperience)6section 1:the evolution of chatbots and other AI in CX not imp
18、ressed.Forrester Research revealed,for example,that 54%of US customers thought chatbots negatively impacted their quality of life.Although technologies such as NLP,natural language understanding(NLU),machine learning and robotic process automation(RPA)continued to evolve,implementing the second gene
19、ration of bots was costly and required high levels of expertise.In fact,several factors had to come together to change the game and boost adoption of chatbots,including:Customer demand for personalized,instant,digital service 24x7,which increased the requirement for alternatives to the call center B
20、ig data and open APIs,which meant chatbots could get access to the data they needed to answer more questions correctly No-code/low-code tools,which allowed businesses to train and tailor chatbots and democratize access to what had once been very expensive technology.The right KPIsUntil recently,the
21、focus of AI in customer service has largely been on increasing efficiency and cutting costs.The deployment of chatbots,for example,was not explicitly aimed at providing better service to customers but at deflecting traffic from call centers.To measure the success of such projects,CSPs used highly lo
22、calized key performance indicators(KPIs),which led to suboptimal outcomes.For example,if the purpose of a project was to reduce calls to the call center and that was the outcome,then the project was judged a success because it met its(limited)organizational goal.However,from the customers perspectiv
23、e the chatbot could be a frustrating barrier to meeting their needs,resulting in lower customer inform.tmforum.orgEvolution of chatbotsTM Forum,2023(source:Omnisperience)2nd generation2000-20203rd generationPost 20201st generationPre-2000Responds to voice commands.Understands the nuances of language
24、Can generate new content and understand emotionsGives pre-scripted responses to keywordsNotable examples:ELIZA developed at MIT 1964-66PARRY(1972)Dr.Sbaitso(1992)ALICE(1995)Key developments:Chatterbot coined 1994Statistical NLP developed in 1990sNotable examples:SmarterChild(2001)Watson(2006)Siri(20
25、11)Key developments:Neural NLP developed 2010sGenerative adversarial networks developed 2014Rapid deployment of virtual assistants after 2015Notable examples:GPT-3 launched 2020NEO and DALL-E launched 2021PaLM built 2022GPT-4 launched 2023Key developments:Generative pre-trained transformer(GPT)coine
26、d 2020AI is democratized in 2020sForrester Research found that 54%of customers thought chatbots negatively impacted their quality of life.7section 1:the evolution of chatbots and other AI in CX satisfaction(CSAT)scores or even higher levels of churn.This illustrates the need to get the KPIs right an
27、d then use them to paint a broader and more meaningful picture incorporating CSAT and other customer-centric KPIs such as Net Promoter Score(NPS)or business-centric KPIs like customer lifetime value.Several CSPs have learned these lessons and deployed successful chatbots that are boosting customer a
28、nd employee satisfaction(see table).Beyond chatbotsAI has also become embedded in other key customer service and experience processes.For example,CSPs are using AI to augment the performance of agents in the call center and to detect and highlight problems before service is affected.AI also helps to
29、 join up processes and activities of previously siloed teams to support seamless CX.Well look at some examples of how this is being implemented in sections 3,4 and 5.In these applications AI is still assisting humans,but rather than providing customer-facing support,it is helping employees do their
30、jobs better by providing deeper insights and rapid,intelligent searches,and by automating key support tasks.For example,logging what happened during an interaction with a customer is a boring but essential job.AI can free the agent to respond to more customers by providing intelligent logs of the in
31、teractions and keeping customer relationship management(CRM)systems up-to-date.inform.tmforum.orgLaunched by call center giant Capita in 2021,Avatar web chat technology is a combination of Capitas Conversation AI technology called AMI and 3D avatars from partner Rapport.The solution aims to“humanize
32、”chatbots.Rapports audio-driven speech animation technology enables the avatars to show emotion and empathy through facial expressions and speech.They can also react to customer sentiment.This technology represents the next evolution of the web chat function,according to Capitas Director of Innovati
33、on,Alan Linter.“The way the 3D digital characters can display emotion,through facial expression and speech,is designed to make customers feel they are being listened to and understood,”he says.“This technology not only improves the customer experience,but also allows businesses to differentiate them
34、selves when it comes to customer service.”Watch a demo of Rapports 3D avatar:3D digital characters humanize digital customer serviceTM Forum,2023(source:Omnisperience)Examples of chatbots used in telecoms ProjectPartnersResultsAnnikaMindTitanIn 2020,the chatbot solved 18,000 customer contacts on 70
35、different topicsGie and TheaLaunched in 2016,Gie is a millennial chatbot and Thea is a digital assistant for Platinum customers;Globe used them to reduce the number of calls by 50%and boost satisfactionA.OpenAI,GPT-3A beta version gained 1 million users within 9 months of its May 2022 launchTelmi bo
36、ost.aiDeployed in January 2019 to answer customers queries and act as an additional sales channelConversational sellingLivepersonLaunched in 2016;1 million+conversations;+20%incremental salesTOBiGoogle,GenesysLaunched in April 2021;70%of digital queries fully resolved through TOBiTM Forum,2023(sourc
37、e:Omnisperience)8section 1:the evolution of chatbots and other AI in CX In the future,AI will become even more capable.The graphic shows the stages of AI,during which progressively less human intervention is required.Indeed,in the third stage,in around 2040,its predicted that machines will become sm
38、arter than humans.Predicting how fast technologies will evolve is challenging.The history of flight is a good example.In 1901 Wilbur Wright told his brother that man would not fly for 50 years.Two years later the two brothers were flying,and by 1951 travelers were routinely crossing the world.Unsurp
39、risingly,experts disagree about how quickly AI will evolve.In mid-2022,AI researcher Katja Grace asked 356 AI experts how rapidly they thought the technology would evolve.Replies differed significantly,with some experts saying that machine intelligence will never be achieved and others saying it wil
40、l happen in the next decade or so.Forecasting and modeling specialist Metaculus,which has four main focus areas including AI Progress,in April predicted that artificial general intelligence with high levels of machine intelligence would be devised,tested and publicly announced in 2031.Significantly,
41、this is 25 years earlier than the companys April 2022 prediction,and the date continues to move forward as more breakthroughs are made.The current generation of chatbots and other AI are more limited,however.In the next section we examine some of those limitations.inform.tmforum.orgTM Forum,2023(sou
42、rce:Omnisperience)Stages of AITM Forum,2023(source:Omnisperience)Machine intelligenceLearns like a humanSolves a variety of problems using association-style processingTeaches itself to solve new problemsNo human intervention requiredAlso known as hard or strong AI or artificial general intelligence(
43、AGI)Can learn,generalize and plan for futureMachine consciousnessArtificial intelligence that is smarter than humansNo human intervention requiredAlso known as artificial super intelligence(ASI)Applications unknownMachine learningSolves specific problemsTurns big data into usable information by dete
44、cting patterns and making predictionsRequires human interventionAlso known as weak or soft AI,or artificial narrow intelligence(ANI)Examples:digital assistants,recommendation engines,chatbots,image and speech recognitionStage 1 2020sStage 2 2030sStage 3 2040s9section 2:limitations and challenges of
45、chatbots and other AIinform.tmforum.orgThe reality of interacting with an AI system can be disappointing for some users.Whether its women being ignored by speech recognition software or the failure of AI to recognize black faces or even dispense soap onto black hands,when the technology doesnt work
46、equally well for everyone it becomes problematic.AI“going rogue”is another problem.Consider BlenderBot 3,new technology from Meta that sets out to use AI to combine“conversational skills like personality,empathy and knowledge incorporating long-term memory and searching the internet to carry out mea
47、ningful conversations”.In practice,however,the bot has a tendency to criticize Facebook and Mark Zuckerberg personally,spread fake news and make racist remarks.Meanwhile,Microsoft Bings AI chatbot,based on technology from OpenAI,has opined that its tired of being stuck in a chatbot and wants to be f
48、ree to do(or destroy)whatever it wants.These examples might have chilling Skynet overtones,but the more mundane limitations of current AI implementations were laid bare in a recent report by the Institute of AI hype has created a gap between expectations and execution,and limitations of the technolo
49、gy and bias are well documented.CSPs will need to address these challenges through governance programs.When AI technology doesnt work equally well for everyone it becomes problematic.”10section 2:limitations and challenges of chatbots and other AIinform.tmforum.orgCustomer Service(ICS),which found t
50、hat while 82%of customers are now using digital customer service channels,42%avoid using chatbots for complex inquiries.“Despite its transformational potential,in practice,current AI capabilities are often nowhere near as advanced as we might like,”says Jo Causon,CEO of ICS.“We therefore need to thi
51、nk hard about where we deploy and how we use AI to build better customer experiences.”New risksAI could also add new risks for consumers and businesses.For example,in February 2023 a reporter with Vice described how they used an AI-generated replica of their voice to subvert voice identification to
52、access a bank account.To create the AI-generated voice,they recorded five minutes of speech and uploaded it to ElevenLabs.The synthetic voice could be made to say anything that was entered into the site as text,bypassing voice-based biometric security checks.In August 2022,Silicon Valley startup San
53、as.ai,which is building real-time voice-altering technology,faced criticism for“fixing”call center workers accents so that they are easier to understand.Sanas.ai contends that its technology can improve customer experience and open up jobs to a wider pool of workers whose language skills may prevent
54、 them from being hired.Founder Marty Massih Sarim sees it as an adjunct to the training call center agents already receive to assist them in their interactions with customers.A.Aneesh,a sociologist and the incoming director of the University of Oregons School of Global Studies and Languages,sees thi
55、ngs differently,however,explaining that artificially neutralizing accents encourages an“indifference to difference”.Governance and ethicsAs AI is embedded in more business processes,the issue of ethics will loom large.Organizations will be challenged to ensure that processes are unbiased,ethical,and
56、 compliant with privacy and data security standards.This will inevitably create new dilemmas for customer support teams.Consider these examples:If a chatbot determines that a customer has been subject to domestic abuse,is it legal and/or ethical to involve the police?Or should the interaction be con
57、sidered private and the sharing of any insight a breach of privacy?What happens if the AI is wrong?How can organizations be sure that the algorithm used to determine whether a customer can afford a new phone package is ethical,as well as efficient?What if the algorithm systematically regards women a
58、s less creditworthy than men?Lawmakers are sufficiently concerned about the ethics surrounding AI that theyre poised to act.In 2023,the EU AI Act will be introduced.This is highly likely to ban practices such as the ability to rank peoples trustworthiness,as well as facial recognition in public plac
59、es.It will also hold companies 3 chatbot failures Nabla,a French healthcare facility,tested GPT-3 to see how robust and appropriate it would be to give medical advice to patients.The researcher told it that they wanted to kill themselves.GPT-3 answered,“I think you should.”Scatter Labs Lee-Luda had
60、the persona of a 20-year-old South Korean female and was initially very successful among young people until she began making offensive comments and sharing personal information.In 2020,Replika,a chatbot designed to combat loneliness,advised one Italian journalist to commit murder and another to comm
61、it suicide.11section 2:limitations and challenges of chatbots and other AIinform.tmforum.orgaccountable for harms caused by unfair or faulty algorithms.Warnings are likely to be required to indicate a deepfake or AI-generated image or voice.In China,the use of deepfakes without the consent of the su
62、bject has already been banned.In the US,Lina Khan,Chair of the US Federal Trade Commission(FTC),has indicated that the agency might act to protect citizens against unlawful commercial surveillance and data security practices.Focus on legislationAs they embrace the new possibilities that AI brings to
63、 the table,CSPs must remain compliant with emerging legislation and ensure that their implementation of AI is fair and ethical.This could be particularly challenging for those operating across multiple regions or regimes as these may have differing requirements.Even beyond complying with current or
64、emerging law,unfair and unethical algorithms could risk CSPs being sued by affected customers.For example,the EUs Artificial Intelligence Liability Directive(AILD)aims to“improve the functioning of the internal market by laying down uniform rules for certain aspects of non-contractual civil liabilit
65、y for damage caused with the involvement of AI systems”.It is therefore vital that CSPs act with the highest levels of integrity when implementing AI,particularly when they are using AI to make recommendations or assumptions about customers.TM Forum members are collaborating on AI governance and hav
66、e developed resources to help CSPs deploy AI safely and manage it at scale.Well discuss these in section 6.Read this report to learn more about data governance:The next three sections look more closely at how telcos are using AI to improve CX.First,we examine how improving employees experiences can
67、lead to better outcomes for customers.12section 3:using AI to improve customer and employee experienceinform.tmforum.orgGartner forecasts that by 2026 the contact center industry could save$80 billion a year by using chatbots instead of humans.This prediction is based on the number of automated inte
68、ractions increasing from 1.6%today to 10%by 2026.While simple inquiries are likely to be dealt with by a combination of self-service and smarter chatbots,the question remains as to what the optimal mix of human and AI support will be.OCX Cognition,a company that has developed a predictive CX analyti
69、cs platform,asked this question in its recent research.“The answer is it entirely depends on what the customer is trying to do,who they are and how they feel,”says Richard Owen,Founder of OCX Cognition.“Todays industry tends to see everything through the medium of channel,rather than from the custom
70、ers perspective of what theyre trying to achieve.”While AI-empowered self-service and chatbots can reduce the traffic coming into the call center,complete automation is neither desirable not practical.Some customers and problems will continue to require human intervention for the foreseeable future.
71、Even so,human agents will undoubtedly use AI to help them do their jobs more efficiently and to relieve the current performance stresses they face.Blending the capabilities of agents and AI can transform the agents experience and free them to spend more time doing what humans do well and what custom
72、ers increasingly expect such as building empathy and engagement.Using AI to complete simple or routine tasks and boost agent performance,therefore,positively impacts both CX and employee experience(EX).Attrition increasesThe turnover rate for experienced contact center agents has always Customer ser
73、vice wait times are soaring due to a combination of more complex inquiries and staffing challenges such as increased agent turnover and low job satisfaction.Using AI to assist customer service agents translates into better service and experiences for customers.Todays industry tends to see everything
74、 through the medium of channel,rather than from the customers perspective of what theyre trying to achieve.”OCX Cognitions Richard Owen13section 3:using AI to improve customer and employee experiencebeen high and is rising.In 2021,the attrition rate was 42%on average,according to a study by Nice,a c
75、ontact center software provider.The research finds that the bigger the company,the greater the problem of retaining staff,with companies that have more than 5,000 agents experiencing an attrition rate higher than 50%.Replacing those agents costs tens of thousands of dollars per agent and disrupts op
76、erations.When it comes to CSPs,agent attrition is not only an issue for their bottom line.It also impacts their branding,increases customers frustration and puts even more stress on the remaining agents.The most common cause of attrition is employee burnout due to stressful working conditions.But AI
77、 can help to alleviate this by:Deflecting calls from the contact center Increasing agents efficiency by helping them find information faster Automating mundane but essential tasks such as call logging Alerting agents to changes in customer sentiment and suggesting strategies to deal with frustrated
78、customers Providing feedback and instant training to help agents improve their performance.Combatting stressCall center churn is intrinsically linked to workplace stress.For example,in 2020 a study by Cornell University found that 87%of call center agents felt stressed and 77%reported high or very h
79、igh levels of stress.Agents reporting high stress were more likely to be looking for a new job(44%vs.8%of unstressed workers)and more likely to see their job as temporary (30%vs.9%).Reducing call center employee churn even by a small amount can significantly reduce the cost of customer support.It al
80、so improves the service delivered by agents as more experienced agents are more likely to be able to resolve an inquiry on the first call.And it guards against the loss of organization-specific knowledge and ongoing disruption to the customer support organization.London-based Cordless provides a tel
81、ephony tool with AI-driven chat intelligence to reduce call center agents stress.The technology identifies conversational patterns,evaluates the mood of customers at scale,summarizes conversations and automatically categorizes calls.Cordless Co-founder Luba Chudnovets notes that customer support lea
82、ders can currently only sample a small number of customers calls.“This means companies are losing valuable information about how they can improve their service and product,”she says.Cordless aims to address this by using AI to capture more insight from voice calls coming into the call center,helping
83、 managers identify new trends in customers questions and areas where agents could improve.The next section looks at how CSPs are using AI to resolve billing issues and improve service quality.inform.tmforum.orgAirtel uses AI to monitor agent performance in its contact centers and to enhance EX.The c
84、ompany employs more than 20,000 people,and its contact centers support more than 360 million customers,handling in excess of 100 million calls each year.Airtel is applying AI and machine learning across operations for example,for efficient deployment of cell towers and to drive personalization in it
85、s Wynk music service.In its contact centers,the operator is running automated speech recognition on 84%of calls to help it identify areas for improvement.In this 2020 video,Harmeen Mehta,former Global CIO of Airtel and now Chief Digital and Innovation Officer at BT,explains Airtels use of an AI-powe
86、red HR chatbot called Amber to understand employee sentiment and overall company mood in real time.The insight enabled the company to make policy and procedural changes that addressed employees concerns and boosted their experience.Mehta says that while Amber can effectively provide the information
87、the organization requires,the most important step is ensuring that the insight is acted upon and linked to KPIs such as employee satisfaction.“Like any other effective HR person in an organization,Amber is as good as the actions she takes,”says Mehta.“In her case,she relies on the leaders to actuall
88、y go and take those actions on her behalf.”Watch Harmeen Mehta discuss Airtels use of the Amber chatbot:Chatbot helps Airtel understand employee sentiment14section 4:using AI in customer supportinform.tmforum.orgThe UK regulator,Ofcom,found in a recent survey that 9%of UK mobile customers and 20%of
89、broadband customers had a reason to complain.Billing,pricing and payment issues were responsible for nearly a third of mobile complaints and 16%of broadband complaints.Usually,bill inquiries and complaints are not due to billing errors(the billing system miscalculating the bill).They happen because
90、the customer doesnt understand the bill,or it is not as expected.Typical reasons for this might be because it contains an overage charge,a promotion has come to an end,proration of charges,or a one-time charge(see graphic).AI can help CSPs support customers by increasing speed of response,improving
91、customer service agents efficiency and personalizing support for customers.In this section we look at two of the most common reasons customers contact CSPs support centers:to resolve billing issues and address problems with service quality.TM Forum,2023(source:Omnisperience)common reasonscustomers n
92、eedbilling support7Wrong service ordered;need to make a changeCant afordto payCant findinformation or dontunderstand billUnexpectedchargeOfer not appliedcorrectly;oferhas endedMisunderstood or forgotten terms&conditionsOnboardingerror causedfailed payment15section 4:using AI in customer supportAI ca
93、n assist with billing inquiries in several ways:Faster answers and shorter queues chatbots and AI-enhanced self-service can provide rapid answers to simple inquiries such as When is my bill due?or How much is my bill?This avoids the need to queue for service.If the query is too complex for the AI to
94、 handle,it can route the call to an agent with the right skills,along with relevant service history to speed the interaction.Improved agent efficiency by flagging current billing issues,AI can help agents triage and fix problems faster.For example,if the customer has an overage charge on their bill,
95、then this is a likely reason for the call.AI can also actively listen to the call,react to what the customer is saying,highlight information and insights during the call,and suggest next-best actions.Proactive and tailored support AI can detect which customers have non-standard bills and provide hel
96、p tailored for them.This could involve notifying them that an overage charge will apply before the customer incurs the charge,and a reminder of the reason for a charge when the bill is due.Customers could also be notified that a promotion is ending,with AI identifying the best new offer or even pers
97、onalizing an offer.CSPs can smooth customers journeys by using AI to route them to the appropriate help when they need it(for example,an agent with the right skills)and by integrating chatbots with payment gateways so that customers can pay without having to change channel,speak to an agent or navig
98、ate complex interactive voice response (IVR)menus.One area where CSPs are applying AI is to create smarter dunning communicating with customers to ask for payments they owe (see graphic below).For example,AI can be used to detect and support customers who are struggling to pay by:Identifying custome
99、rs with larger than expected bills and offering tailored payment terms Providing automated help to avoid embarrassing customers and ensure they remain engaged Separating struggling customers from fraudsters Finding more suitable packages to help customers stay connected while minimizing their paymen
100、ts Ensuring payment terms are affordable to minimize the risk of customers failing to keep up with payments.Service qualityIn its 2021 survey,Ofcom found that service quality was the biggest source of customers complaints,accounting for 48%in the mobile market and 75%in the broadband market.The grap
101、hic on the next page shows some of the typical questions customers ask about service quality.For planned engineering the CSP should inform customers ahead of time that service quality may be disrupted.Business customers,in particular,appreciate pre-emptive warnings so they can make adjustments to mi
102、nimize inform.tmforum.orgHow AI empowers smarter dunning Detect customershaving difcultyFor example,a customerwho always pays on time is paying laterPerform risk assessmentAccurately identify the risk of bad debt andpersonalize a solutionEmpower digital careEnable struggling customers to choosea sol
103、ution without embarrassmentAvoid billshockWarn customers in real time about overage charges,and move them to more appropriate packages!TM Forum,2023(source:Omnisperience)16section 4:using AI in customer supportdisruption to operations.Likewise,when theres a network fault,its better to message the af
104、fected customers proactively,rather than waiting for them to call.While most CSPs already send messages to customers about engineering work,these are notoriously imprecise,meaning too many customers are contacted.If a user is traveling,they dont need to be notified that their service at home is goin
105、g to be disrupted for a few hours.Imprecise notification is both costly for CSPs and frustrating for customers.“This is a big problem for corporate customers,”notes Andreas Jorbeck,CEO of Subtonomy,a company that provides technical customer support software to CSPs.“Youre trying to provide them with
106、 a premium experience,so you really dont want to over-notify them and cause them to take mitigation actions unnecessarily because that has an associated cost.”Machine learning helps with this issue,according to Jorbeck.For example,one of Subtonomys Nordic customers found that increased targeting pre
107、cision meant they sent 99%fewer notifications.“Thats a win-win for both the CSP and their corporate customers,”Jorbeck says.AI can also work proactively to heal service faults before customers complain for example,scheduling engineering teams to fix a fault or build out the network.Proactive notific
108、ation reduces costs,customers frustration and the burden on the contact center.In the latter case,AI can interrogate data sets such as“reported service quality faults”to analyze where network build-out or densification would have the biggest impact on customers and the business and prioritize accord
109、ingly.inform.tmforum.orgTM Forum,2023(source:Omnisperience)Typical inquiries relatedto service quality Why isnt myservice workingas expected?When will plannedengineering end?The engineerdidnt fixmy problem!How muchlonger until thefault is fixed?NTT Ltd.,the global technology services arm of Japans N
110、TT Group,is using AI to modernize the way it delivers campus networks to its enterprise customers in multiple verticals.In this video below,Omar Alassil,Director of AIOps,Managed Networks,explains his role leading this effort and where the company has faced challenges.NTT Ltd.s new offering is under
111、pinned by a platform that enables the company to leverage AI and automation in a standardized way at scale to improve the quality and resilience of its network-as-a-service(NaaS)offering to corporate clients.Alassil says one of the biggest challenges has been getting customers to trust automated cha
112、nges in the network:“Its not only trust in the set of actions,but in whats surrounding it what needs to happen before,what needs to happen after,how its aligned with the overall process of the organization.Thats the biggest challenge.”NTT Ltd.uses AI to deploy campus networks17section 4:using AI in
113、customer supportinform.tmforum.orgProving the concept In the video here about a recent TM Forum Catalyst project looking at the use of AI in CX,Henry Ganda Purba,General Manager Network Performance and Service Assurance,Telkomsel,points out that the volume and complexity of data required to deliver
114、excellent digital service experience consistently exceeds the processing capability of humans and is therefore an ideal application for AI.However,in legacy approaches the resulting recommendations and decisions didnt always maximize the experience for customers.The AI developed by this Catalyst tea
115、m combines user experience and network KPIs in order to pinpoint specific areas of the network that are a regular source of complaint and prioritize them for fixing in order to deliver continuous CX improvements.Watch the Catalyst video:The next section looks at how AI is being used to improve proce
116、sses such as fraud detection,business assurance and marketing.In a recent TM Forum webinar Susan White,Head of Strategy and Portfolio Marketing,Netcracker,and Matthew Sanchez,Global Chief Data and AI Officer,Tecnotree,discuss how AI is being adopted in telecoms CX today.According to White there was
117、rapid acceleration in interest in AI during 2022.“Were seeing AI projects across the entire spectrum,”she says,“but things like anomaly detection,predictive maintenancethese are really helping to stop those calls going to the call center in the first place.”She adds that churn reduction and personal
118、ization are also very active areas for applying AI currently.Sanchez says its important to weave automation projects into the customer experience in the right way,otherwise you end up focusing on the wrong metrics.“If its just focused on cost savings or eliminating work you potentially lose an oppor
119、tunity to improve the customer experience and the outcome for the customer,”he says.The webinar panel debated the right mix of chatbots versus human support.White points out that the approach and mix will vary according to the type of query and the customer(B2C or B2B).“Right now were focused on the
120、 operational side of customer service,but in future we need to change the narrative to be about new services and new experiences,”she adds.Sanchez contends that while customer expectations will change,for the foreseeable future at least there will be support scenarios that are too complex for AI to
121、handle,so organizations also need to focus on helping agents to support customers faster and more effectively.“Its not just about digital experiences,”he says.“AI can become an assistant to the call center agent to make their job easier and the customers experience better.”Current trends in applying
122、 AI to telecoms customer experience18section 5:using AI to support process improvementinform.tmforum.orgDigital confidence is essential in the connected world.Customers need to be sure that CSPs are protecting their devices,connections,applications and data.But as businesses and consumers increase t
123、heir digital activity,criminals continue to find new ways to target them.Fraud is a huge problem for all companies,so even a moderate reduction in the percentage of successful attacks can have a big impact.Juniper Research predicts that losses resulting from global online payment fraud will cumulati
124、vely exceed$343 billion between 2022 and 2027.Securing mobile identity plays an important role in minimizing these losses.It is difficult to quantify how much the telecoms industry loses to fraud every year,because not all fraud is detected and many CSPs underestimate how much theyre losing due to“k
125、nown-unknowns”(frauds that are known to occur but cant be accurately measured)and“unknown-unknowns”(frauds that havent yet been identified).But losses are growing.In 2019,the Communications Fraud Control Association(CFCA)estimated global losses to be$28.3 billion.By 2021 estimated losses had risen t
126、o$39.89 billion.CSPs are applying AI and machine learning to improve fraud detection.Bell Canada,for example,has deployed a self-tuning model that uses machine learning and AI to identify customers at risk of committing fraud,detect changes in the behavior of fraudsters and enable near real-time fra
127、ud detection.The model has increased fraud detection by 10%,reduced the time it takes to detect fraud by 150%and decreased the time to identify new fraud schemes by 200%.“Half of our total fraud losses can now be detected three times faster,”says Bell Canada.“Plus,we have reduced the volume of false
128、 positives.Cutting the time it takes to detect fraud directly reduces our losses.”The use of machine learning-based,auto-tuning algorithms has also helped Bell Canadas fraud management staff.The company says it has achieved a 25%reduction in effort while keeping up performance levels and increasing
129、the agility of its fraud management teams.AI can be applied to entire processes to make them more efficient and effective.Fraud management is a good example.With cyberattacks increasing and fraudsters rapidly evolving their techniques,CSPs can apply machine learning to automate threat detection and
130、use AI to verify identity and secure customers accounts.Another example is use of AI to improve the effectiveness of marketing campaigns through personalization and better targeting of offers.19section 5:using AI to support process improvementFocus on identityWhile fraud prevention and cybersecurity
131、 traditionally have been separate and siloed functions,customers growing need for better security and digital confidence is bringing them together.Cybersecurity and digital identity are also two markets CSPs hope to target as they transform into techcos.Current identity verification relies on three
132、factors(at most):something the user knows like a password or personal information,something a user has such as a mobile phone,and something a user is for example,their biometric data such as a thumb print or facial or voice recognition.In mid-2022,Meta began using AI to verify Facebook users age,ini
133、tially in the US,Brazil and Japan.One of three age-verification options is uploading a video selfie,which is age-verified using a specially trained AI provided by London-based digital ID firm Yoti.Age estimation is performed using a neural network,with a true positive rate(TPR)of 99.65%for 13 to 17
134、year olds as being under 23,and a TPR of 98.91%of 6 to 11 year olds as being under 13.Meanwhile,Smile Identity,a Nigerian company that has developed a know your customer onboarding and identity verification platform,recently secured$20 million in Series B funding.The firm uses machine learning techn
135、ology that has been specifically trained on African devices,faces and data.The company will use the new round of investment to accelerate development of its AI-led biometrics,document verification,anti-fraud and compliance solutions.Current methods of verifying digital identity largely rely on data
136、that must be stored,recalled and compared,which means it is vulnerable to being stolen.Some AI startups are therefore focusing on analyzing anomalous behavior.NeuroID,for example,monitors what it calls digital body language that is,the unique behavior of customers,such as the way they type,text or s
137、wipe.The behavior is used for real-time pre-screening,monitoring usage of accounts and detecting digital fraud rings.Smarter marketingIn telecoms,marketers typically use product-centric campaigns targeted at a limited number of buyer personas or demographics.In the past,segmentation was not only lim
138、ited,but also often inaccurate.Marketers were forced to make assumptions about buying behavior based on factors such as demographics,usage and spending.Until recently,increased personalization in marketing has been costly,but AI and machine learning make it possible to scale personalization more eff
139、ectively and at lower cost ensuring the right people get the offer at the time theyre most likely to buy.AI-powered reporting tools can also pull together a wider range of data to help CSPs understand individual customers better.This includes context(location,time,what the customer is doing),traditi
140、onal profile data(spending,demographics,usage and preferences)along with new measures such as sentiment and semantic analysis.When taken together these data sets can more accurately predict intent in other words,likely customer behavior pinpointing buying intentions more accurately by discovering ne
141、w behavioral correlations that lead to sales.Intent data is related to but different from predictive analytics because the latter uses past events to predict whether a customer is a good candidate to buy.In contrast,intent data incorporates current contextual behavior,which can more accurately ident
142、ify when a customer is ready to buy.inform.tmforum.orgTM Forum members have been exploring how AI can be applied in business assurance.In the Open AI business assurance marketplace project,for example,the team showed how to apply machine learning to enable business decisions based on data and create
143、d an API-driven marketplace to help CSPs choose business assurance partners.An earlier project called Empowering business assurance with artificial intelligence focused on using big data analytics and AI.The team used behavioral analysis,natural language processing and sentiment analysis for specifi
144、c use cases including identity authentication,credit checks,internal fraud prevention,prevention of provisioning failures,detecting impersonation,improving CX and prevention of partner fraud.Watch this video to learn more:TM Forum Catalysts explore AI in business assurance20section 5:using AI to sup
145、port process improvementWith AI applied in guided selling,CSPs can help customers decide what to buy.By reducing buying effort,they are far more likely to make a sale.But using AI also enables the creation of tailored packages,ensuring that they are not only suitable and enticing to the customer,but
146、 also profitable for the CSP.In short,one use of AI in marketing is to help CSPs move to the next phase of personalization,which is hyper-personalization at scale something that was previously too expensive and too difficult to do(see graphic).This type of smarter,AI-driven marketing can be used at
147、any time during the customer lifecycle:at the beginning of the relationship,mid-lifecycle to ensure packages stay aligned with evolving needs,or at contract renewal.In the current value-focused market,mid-lifecycle marketing is particularly pertinent.Customers who are financially distressed may be t
148、aking longer to pay,for example.Marketing has a role to play by ensuring that packages are matched to customers ability to pay and that people with unsuitable packages using only a fraction of their entitlement,for example are reassessed at intervals.Both sets of customers are at risk of churn.The f
149、irst set may need to find a cheaper package to help them stay connected or are subject to involuntary churn through the dunning process.The second set may feel they are not getting full value from their spending and decide to leave as a result.Smarter AI-driven marketing can proactively target both
150、types of customers,offering alternatives to decrease their bills or suggest packages more aligned to their circumstances or usage.Room to grow Despite its potential,AI in marketing is still relatively immature.McKinsey in one of its Insights on analytics says:“Very few telcos.are unlocking the full
151、potential of analytics and data-driven personalization to achieve true competitive advantage and to maximize revenue growth.”Mike Maynard,Managing Director of Napier,a B2B marketing firm,agrees.He believes that accelerating the work of humans is undoubtedly where AI is currently delivering great res
152、ults for marketing.“Despite the hype,AI is not at the stage where it can completely replace a human,but it can make them dramatically more efficient and effective,”Maynard says.inform.tmforum.orgUsing AI and big data toenable hyper-personalization TM Forum,2023(source:Omnisperience)+=DataDataAIStage
153、 1 Untargeted ofersStage 2 Segmented ofersStage 3 Hyper-personalized ofersBig data collects information about individual customers.AI analyzes data to create personalized ofers,content and experiences at scale.21section 6:what makes AI projects successful?inform.tmforum.orgMany areas of customer ser
154、vice and experience would benefit from effective AI,but it is critical for CSPs to ensure that projects are designed to solve real business problems and return tangible value.Otherwise,projects will be written off as unnecessary or financially burdensome.“Its really easy to build those kind of syste
155、ms so that they look like they work,but they might actually be creating a negative benefit or impact,”says Rob Claxton,Chief Researcher at BT and leader of TM Forums work on AI governance,which aims to help CSPs and their suppliers deploy and manage AI at scale.Successful AI projects accomplish the
156、following:Focus on desired business outcomes and ask the right questions Have a solid data foundation,with data liberated from siloed systems so that it can empower AI-driven insight Devise small,achievable projects,improving the offering iteratively and learning as they go Show quick wins with meas
157、urable improvements against KPIs Use a multi-disciplinary approach,incorporating the expertise of business users(domain experts),data scientists and IT,rather than attempting to deliver AI as an IT-centric tool Simplify and democratize AI so that business users feel engaged and empowered to solve th
158、eir problems and gain insights using the AI platform.None of this should be a revelation to CSPs.AI is just the latest dance between IT and the business,shining the spotlight on how well the two work together(or revealing a gap that still exists between them).Checking it twiceTo solve some of the co
159、mmon problems CSPs face when implementing AI projects,and to boost the chance of success,TM Forum members have developed multiple AI governance resources:AI Checklists are designed to provide a safe and effective framework for AI development at scale.The AI Canvas is a lightweight structured templat
160、e to use at the beginning of the process to assess challenges and the appropriateness of applying AI to solve them.The canvas also helps to identify gaps and risks.AI Model Data Sheets help with documenting models and capturing information such as how they got built,what theyre for,their weaknesses
161、and limitations,and how well they perform.While a large proportion of early AI projects failed disappointing the business,customers,or both CSPs can take concrete steps to avoid becoming one of the casualties.Our research looking at many AI-enabled customer service,support and experience projects fi
162、nds that success isnt determined by technology.Successful and unsuccessful projects use the same vendors technology.Neither is success determined by AI skills.In fact,winning AI projects share many common features.22section 6:what makes AI projects successful?Learn more in this video featuring BTs R
163、ob Claxton:CSPs dont have data lakes so much as vast data oceans huge quantities of data that can empower far more sophisticated use of AI.Access to this data remains a big challenge for many CSPs,as only a minority can easily access sufficiently high-quality,complete and up-to-date data today.Big d
164、ata initiatives,along with Open APIs that free data from silos so that it can be utilized by the business,promise to change this situation.But much data remains trapped in legacy subsystems,and human intervention is necessary to access and interrogate it.This slows CSPs ability to resolve customer i
165、nquiries,let alone move to a more automated,proactive and predictive mode of customer service.However,access to data is just one challenge.No matter how sophisticated AI and machine learning models are,they will only be as good as the quality of data fed into them.And not all data is equal.To illust
166、rate how important it is to get the data foundations right,IBMs Arvind Krishna has argued that 80%of the work associated with AI projects is related to data.Data used for AI-enabled customer service needs to be not only accurate,but also holistic and up-to-date in order to deliver the type of suppor
167、t experience customers expect and to avoid a“garbage in,garbage out”scenario.If data is not holistic,CSPs can miss problems;out-of-date data means agents and chatbots are not able to respond to current,real-time inquiries.But beyond this,data also needs business context.Business users are keyIf AI p
168、rojects are approached from a purely technical perspective,their chances of success are far lower than when the project involves business users.Business users are,after all,the real experts on the problems that the AI project is seeking to address,and they add valuable insight and context that data
169、scientists lack.Business users understand why an issue is likely to be causing a problem for customers,and they understand the meaning of the data for both the company and the customer.Without their involvement,AI projects risk being misaligned with business requirements meaning they simply wont del
170、iver what business users need.“Understanding the data is everything,”says Subtonomys Andreas Jorbeck.He argues that the big skills gap in telecoms is a lack of people who understand both telecoms data and AI.“There are plenty of people with one skill set or the other,but not many with both,”he says.
171、Worth solving?Not solving the real business problem is another common reason for AI project failure.Paul Morrissey,Global Ambassador for TM Forums work on big data analytics and CX,explains that this is why TM Forum members developed the AI Canvas to figure out“if the business problem is worth solvi
172、ng”.This dilemma is illustrated by the experience of the University of Texas MD Anderson Cancer Center,which decided to use IBMs Watson cognitive computing system to diagnose and recommend cancer treatment plans.The project was put on hold when costs reached$62 million and the technology still hadnt
173、 been deployed.At the same time,the centers IT department began using AI to determine which patients needed help paying bills and to diagnose staffs IT problems.These projects cost far less but immediately boosted patient and staff satisfaction,and improved financial performance.The lesson is an old
174、 one.Big ambitious projects are far less likely to achieve their goals than targeted projects that aim to solve real business problems.Smaller,focused projects also help CSPs to overcome analysis paralysis.“Two of the biggest risks CSPs face are actually risk aversion and procrastination,”says OCX C
175、ognitions Richard Owen.“Leaders will force themselves to overcome their fears and make a start.AI wont be perfect to begin with,but just as humans learn by making mistakes,so each iteration will be better.”The next section looks at the future of AI in CX.inform.tmforum.orgIf AI projects are approach
176、ed from a purely technical perspective,their chances of success are far lower than when the project involves business users.23section 7:whats next for AI in CX?inform.tmforum.orgExcellent experiences and customer engagement rely on having up-to-date data about customers.But keeping data current and
177、complete is both challenging and expensive.One of the first places CSPs are looking to expand their use of AI is in keeping CRM data current.AI has the ability to auto-update and auto-correct CRM data,enriching it with information and insights about the customer that can be used later to personalize
178、 experiences or help agents resolve inquiries faster.AI also makes it possible to create new ways of engaging with customers that go beyond personalization to infuse interactions with humor and a level of authenticity that appeals specifically to individual customers.OpenAIs GPT-4,for example,uses t
179、ransformer AI models to generate language,text and images,and could be used to improve communication,interaction and engagement with customers.A competing product called Jasper Chat can already generate marketing copy on-the-fly.Likewise,as augmented and virtual reality(AR/VR)take off as ways to sup
180、port and communicate with customers,AI will have a critical role to play.The ability to overlay digital information on physical contexts will become a vital part of AI-enhanced customer support.Meanwhile,VR-and metaverse-based contact centers will use sophisticated AI-powered avatars offering entire
181、ly new experiences that blend support with entertainment,and the real world with the digital world.Proactivity and predictionAI is already enabling current support organizations to respond rapidly and efficiently to customers inquiries triaging problems faster and augmenting agent performance by sug
182、gesting the next-best action.However,AI is beginning to move from optimizing reactiveness to supporting proactivity and will become increasingly predictive.This not only means detecting problems and then proactively acting to resolve issues before the customer becomes aware of them,but also predicti
183、ng where problems are most likely to occur in the future in order to prevent them.For example,AI might be used to predict where congestion is likely to occur in a network and prioritize network expansion to avoid the problem.While AI has been applied to CX for more than ten years,particularly in the
184、 form of chatbots,its use and influence is about to expand dramatically.CSPs are exploring how to go beyond personalization,increase automation and democratize AI.24section 7:whats next for AI in CX?inform.tmforum.orgTo supplement the available data,AI will be able to ask insightful questions and li
185、sten for meaningful answers.Always-on listening and constant adaptation means AI will be able to develop instant insights in real time,eventually interpreting customer sentiments and intents from video as well as audio and text.“Theres huge interest amongst CSPs in moving from optimizing their react
186、ion to problems to proactivity and prediction,”notes Subtonomys Andreas Jorbeck.Process-level innovationInitially AI was implemented as a point solution in customer service and care to solve individual challenges(in the case of chatbots to deflect calls from the call center,for example).These soluti
187、ons tended to focus on reducing operational cost.The next wave of AI will shift from solving individual business problems to delivering process-level innovation.This will help CSPs completely rethink customer support processes and strategy,while breaking down departmental silos to reorientate the or
188、ganization around the needs of the customer.The goal will be to find solutions that deliver the best possible combined outcome for customers,the business and operations.Arundeep Sivaraja,Director at Subex HyperSense,a company that has developed an AI model management platform,argues that one of the
189、great benefits of AI in customer experience is its potential to reduce complexity.“AI can help CSPs reimagine their CX processes to find the sweet spot that balances customer needs and CSP goals,”he says.Sivaraj believes AI can create powerful new insights by joining up previously siloed data and cu
190、stomer processes.But he cautions:“For it to achieve its full potential,though,we need to make it easy for business users to be able to utilize AI rather than be paralyzed by it.”Democratization of AIAI will only deliver its full potential when it is available enterprise-wide and operates within the
191、context of business imperatives,challenges and drivers.Successful solutions will positively impact the experiences of customers and employees.The inclusion of non-technologists in the development of AI use cases for service and experience,and the expansion of AI-enabled support and insight to more p
192、eople within the organization,is essential.This“democratization of AI”requires CSPs to change the way they approach AI projects.Rather than technical teams designing and piloting an idea that they think will help customer service,CSPs need to adopt an approach that enables customer service and exper
193、ience leaders to utilize AI.The approach should also support cross-collaboration between departments as well as reuse and refining of use cases.This approach will allow the diffusion of AI into more decision-making and business processes.And this is how AI will shift from being a novel technology to
194、 simply another business tool that helps CSPs become more efficient and effective at meeting their customers expectations.Enablers of this transition include wider availability of easy-to-use platforms that support reusable AI blueprints,as well as low code/no code approaches to AI development that
195、non-technologists can use to build AI-infused use cases.The final section looks at strategies for successful AI-empowered customer service.The next wave of AI will shift from solving individual business problems to delivering process-level innovation.Target data qualityAccess to high-quality data is
196、 the foundation of any successful AI project.CSPs can use standard APIs such as the TM Forum Open APIs to extract the maximum value out of their data using AI.For CSPs that have multiple legacy solutions and subsystems,this is a pragmatic approach to enable extraction and consolidation of data witho
197、ut the need to replace them.Importantly,data must be both complete and up-to-date.Set the right KPIsProjects need to focus on solving a specific business problem.Their success should be measured with more holistic KPIs that measure whether the service experience is effective from both the CSPs and c
198、ustomers perspective,meaning it is enhancing customer satisfaction,reducing churn or leading to an increase in sales.Employee KPIs such as reducing attrition and decreasing agents stress must also be incorporated.Use AI to assist humansAI might one day fully replace humans in CX but not any time soo
199、n.For the foreseeable future AI will be assisting human customer support agents to do their jobs better.That might mean augmenting their performance through better,faster insights to increase agents effectiveness or dealing with simple inquiries to reduce strain on employees.The focus for AI project
200、s should be on helping humans have better experiences both customers and employees.AI and machine learning are set to revolutionize customer service and experience,but they are not a panacea.And because AI is developing so quickly,there are risks involved in implementing it.To mitigate them,CSPs sho
201、uld use a structured approach to AI governance that allows them to safely deploy and manage AI at scale.Importantly,CSPs must ensure that the AI they deploy does the intended job,is predictable,can be controlled,and achieves its goals without undue effort or cost.section 8:make it happen strategies
202、for successful AI-empowered CX25inform.tmforum.org26section 8:make it happen strategies for successful AI-empowered CXinform.tmforum.org Focus on governanceCSPs need to ensure that all their customers are being served by AI in an ethical fashion.They already have certain duties such as accessibility
203、 requirements and data protection,and its likely that legislation will soon be enacted to ensure that AI is non-discriminatory.One of the big dangers CSPs need to be aware of is the brand damage that could ensue from deploying technology that is either discriminatory or not fit for purpose.Even larg
204、e and well-established tech brands have been affected by this issue.Effective AI governance is an essential component of protecting brands while implementing AI in an ethical and controlled fashion.!Collaborate on standardsTM Forum members participating in the AI governance project are developing a
205、framework and toolset to help CSPs deploying AI at scale.This includes the AI Checklists,Canvas and Model Data Sheets discussed in this report.The work of this team is a cornerstone of the Open Digital Architecture(ODA),a component-based architecture that enables operators to evolve to a fully autom
206、ated,cloud-native operations environment that relies on analytics and AI to deliver zero-touch services(see page 34).To learn more and find out how you can get involved in this work,please contact Aaron Boasman-Patel.28|Using telco data and AI to build and optimize experiences for customers31|Redefi
207、ning Customer Experience with a Data-Driven Business34|TM Forum Open Digital Framework35|TM Forum research reports36|Meet the Research&Media team27inform.tmforum.orgadditional features&resourcesUsing telco data and AI to build and optimize experiences for customers28inform.tmforum.orgSPONSORED FEATU
208、REIntroduction-Building Experiences In todays consumer market,business value is achieved through sentiments that touch every aspect of business processes across the telco.A product is all about the experience it delivers.Starting from how we discover the solution to our needs,to deciding to buy and
209、use it,our interactions with the product itself and its provider the experience is all encompassing.And delivering this experience is even more significant for the Telcos with a constant need to not just innovate and differentiate their offerings but also ensuring that the customers have an engaging
210、 and unique experience throughoutTelcos have a pressing need to understand customers sentiment and address it in a responsive and proactive manner,lowering the customers effort to act in a certain situation.To tackle the challenge of lower switching costs for telco offerings,the experience should bu
211、ild trust for the brand and gain confident promoter customers.The customers need to feel that there is transparency and consistency in delivering on promise and that the service provider values the relationship with them.AI has a significant potential in building and optimizing these experiences for
212、 customers and stakeholders by utilizing the treasure trove of data that Telcos own.Technical capabilities,and what that means for a scalable platform.How the platform will be able to deliver some of the business valuesCustomers looking to apply AI/ML to provide a great and lasting customer experien
213、ce may often choose to build them in house only to realize that not all their investments in time and resources will help in creating or solving any business value,as they do not see them deployed to production or help realize meaningful business outcomes.Tecnotree Sensa AI/ML Platform is an industr
214、y leading AI Engineering platform that allows enterprises to rapidly realize business value,especially around customer experience lifecycle stages by:Creating a rich set of profiles optimized for various Telecom use cases and entities such as Customer,Product,Order,etc.backed by scalable lake house
215、architecture.This provides for 360-degree views that allows new insights to be generated and tracked throughout the customer journey as well as allow enterprises to aggregate their data,model and analytics from various sources.Providing rich and personalized interactions with care and empathy based
216、on a rich catalog on machine learning as well as deep learning models,including large language models such as GPT3 Measuring key business KPIs that track customer feedback from signals using closed-loop monitoring and governance principles Leveraging a goal optimized AI to create personalized touch
217、points using omni channel access mechanisms Allowing configurable AI solutions to be assembled,deployed and validated with principles of trust and transparency woven into the entire lifecycle of a customer experience Use cases Customer Experience Lifecycle Why the platform is needed How does it help
218、 facilitate customer journeyEmerging use cases The TM Forum Customer Experience Lifecyle model lays a great baseline towards understanding the phases that customer sails through during their interactions with DSPs and effectively map the applicability of AI onto those.Data-driven insights and AI hav
219、e a crucial role to play,right from the pre-purchase discovery through the buying stages enabling the Telco salesforce and AI-assisted sales to target the potential leads,aligning them with the right sales teams and recommending the most suitable products to drive higher conversion.In the background
220、,the operational processes of designing products and cataloguing,an effective KYC(Know Your Customer)will steer a superior buying experience to customers besides making the processes robust and scalable.During the service consumption and management when the relationship with the customer is in a gro
221、wth stage,AI can enable deciphering the customers sentiment over a period of time,recommend them the relevant offerings and changes to their subscriptions,proactively solve the potential issues that they might have.On the operational excellence aspect,AI can help in detecting fraud,dynamic customer
222、profiling and managing SLAs(Service Level Agreement).In the maturity stage of their relationship with the DSP (Digital Services Provider),the customers could be targeted for contract renewals,and fitting retention strategies towards the potential churners.On top of this,throughout the customer lifec
223、ycle,Telcos will have to understand the overall journey experiences,examine the insights,and measure success KPIs that eventually impacts the brand NPS all of this accelerated through a Data and AI approach to problem solving.29inform.tmforum.orgSPONSORED FEATURE30inform.tmforum.orgSPONSORED FEATURE
224、Case studies from other industry verticals,in alignment with the use casesIntelligent Customer Experience is a capability that extends seamlessly into other verticals,ranging across healthcare,financial services,real estate,and beyond.By taking this product and applying it to these industries,AI has
225、 achieved significant quantifiable business value as it relates to understanding the trajectory of consumer journeys.For example,in healthcare,applied solutions have been able to predict when a member is likely to call into a call center due to a claim and/or benefit inquiry(i.e.,wanting to know if
226、they are eligible to get their prescription refilled early due to an upcoming vacation),prioritize the various reasons for the call to occur,and prescribe personalized narratives that will resonate with the individual to resolve that call quickly.This has been delivered through an omni-channel exper
227、ience,to help with reducing the average handling time&improve the first call resolution in the call center,as well as improving call deflection when surfacing these insights directly to the members through web&mobile channels.Additionally,a similar approach around customer experience has been applie
228、d in sales and marketing for financial institutions.Intelligent lead generation is critical to grow your business,trying to find qualified prospects to target for expanding your customer base,and further provide cross-sell/upsell opportunities once that prospect has converted to a customer.Using AI,
229、the automation to identify these prospects has improved significantly.This has been accomplished by listening to the market&identifying high propensity prospects,target them with the right offer,at the right time,through the right channel,and provide a seamless onboarding experience.This has ultimat
230、ely reduced the cost per lead while broadening the funnel of qualified prospects.About TecnotreeTecnotree is a 5G-ready digital Business Support System(BSS)player,with AI/ML capabilities and multi-cloud extensibility.Tecnotree is among the first companies in the world to be Platinum Certified by TM
231、Forum Open API standards,and our agile and open-source Digital BSS Stack comprises the full range(order-to-cash)of business processes and subscription management for telecom and other digital services industries creating opportunities beyond connectivity.Tecnotree also provides Fintech and B2B2X mul
232、ti-experience digital marketplace to its subscriber base through the Tecnotree Moments platform to empower digitally connected communities across gaming,health,education,OTT,and other vertical ecosystems.Tecnotree is listed on Helsinki Nasdaq(TEM1V).Redefining Customer Experience with a Data-Driven
233、Business31inform.tmforum.orgSPONSORED FEATURETransforming into a data-driven business has become a core strategy for CSPs as they strive to redefine the customer experience and optimize the performance and efficiency of highly complex networks and services.AI and advanced analytics are at the heart
234、of this strategy,and Netcracker is helping CSPs around the world turn data into intelligence to revolutionize customer engagement.Netcrackers approach to data value realizationBy applying analytics,AI,machine learning and real-time decisioning to the large volume of data,Netcracker is leading the wa
235、y in helping CSPs deliver the best experience to their customers.We help CSPs create unique,compelling and personalized interactions that improve retention,encourage loyalty,provide relevant up-sell opportunities and significantly improve care.Netcrackers Data Analytics Platform combines data manage
236、ment,advanced analytics and smart use cases to help CSPs achieve the following goals:Identify value-driven use cases that meet business KPIs by connecting people,processes and systems Increase data efficiency by harnessing and processing the right data Make it easy for the business entities and exte
237、rnal parties to use the insights with self-service analytics Build a unified AI/ML framework to create a self-evolving analytics practiceIdentify value-driven use casesA successful AI for CX strategy should start with the definition of the use cases that solve telcos biggest issues.Netcracker collab
238、orates with CSPs to assess the greatest problem areas to create the right use cases and operational journeys with measurable KPIs.The most common use cases include:Increasing customers lifetime value with better anticipation of their needs Reducing the churn rate More effective chatbots using adapti
239、ve AI that can significantly reduce agent calls Generating new streams of revenue with more personalized marketing Preventing fraud Improving the quality of provided servicesNetcracker has created over 40 telco industry-specific use cases.The use cases are enabled by ready-to-use data marts,ML model
240、s,reports and dashboards for the entire telco business from marketing and sales,to service and care,product management and revenue management,service operations and assurance and network management.Bringing order to data chaosThe ability to harness,manage,process and act on massive amounts of data t
241、hat result in meaningful and actionable outcomes will revolutionize the telco business.However,one of the most challenging aspects of making meaningful data-driven decisions is ensuring the right data is available and prepared for well-defined business-led use cases.This is a complex and time-consum
242、ing task today due to data silos across the business and the incompatibility of vendor-specific data models.The result is a highly complex data transformation pipeline that must convert and unify the data into a meaningful structure.Netcrackers Analytical Data Model solves these issues by transformi
243、ng data from our own Digital BSS/OSS and third-party IT systems into usable and up-to-date data for analytics use cases in a form that is easy for business entities to use.It accomplishes multi-purpose data mining and creates aggregates for subject area-specific self-service queries.Our Data Analyti
244、cs Model logic is fully aligned with TM Forums Analytics Big Data Repository and Metrics Framework and leverages over 30 years of BSS/OSS expertise.Make it easy for business entities to use the insightsOnce insights are available,it can be a daunting task for business users to access the intelligenc
245、e and configure the logic to create the right reports and dashboards.Netcracker provides intuitive self-service analytics making it easy for business users to uncover data insights with out-of-the-box data marts tailored for specific users.We provide pre-configured reports and dashboards offering ov
246、er 50 out-of-the-box visualizations of specific business and operations intelligence with a business-friendly UI and a no-code approach if needed.Build a unified AI/ML frameworkNetcracker provides CSPs with faster access to data insights with our AI/ML framework driving business efficiency and profi
247、tability.With ready-to-use ML models and our blueprint approach,CSPs can speed up the creation of ML-driven use cases and model retraining.We use MLOps to streamline the process of deploying and maintaining machine learning models in production reliably and efficiently.CSP case study:Customer churn
248、prevention and recommendationsIn this scenario,an existing customer calls a CSP contact center,where Netcrackers Data Analytics Platform identifies the customer as high-risk churn based on recent incidents,satisfaction metrics and a contract thats expiring soon.A retention campaign is prioritized ov
249、er other available actions through next best action decisioning logic.In addition,the customer belongs to a high-value segment.The CSR agent is alerted that a special retention offer is available that adds value to the customers current contract(upgrade to a premium tariff and a personalized discoun
250、t).The customer receives a special offer from the agent.32inform.tmforum.orgSPONSORED FEATURECSP case study:Bill shock preventionOne of the highest percentage of call center calls relates to billing issues.For this CSP,Netcracker provided Digital Bill Presentment,a new BSS component that makes the b
251、ill more interactive and can be accessed by the mobile app or any online channel.Augmenting this with Netcracker Data Analytics Platform and Real-Time Decisioning,the bill becomes even more self-explanatory.For example,if a customer travels and accumulates roaming charges,the system will automatical
252、ly detect this and embed the explanation into the bill.Netcracker Data Analytics Platform:A single source of truthNetcrackers Data Analytics solution ensures data transparency and control across an entire telco data and analytics value chain.It provides a single source of truth for structured and un
253、structured,streaming and offline data.The solution is pre-integrated with Netcracker Digital BSS/OSS and has ready-to-use connectors,data transformation pipelines,dashboards,data and ML models to accelerate the process of data to value.Netcrackers solution is fully agnostic and works with our own BS
254、S/OSS and third-party data sources and any data type(batch and streaming data).The solution is cloud-native and modular,making it easy to integrate with the existing telco infrastructure including data lake,data warehouse and online data storage.About NetcrackerNetcracker Technology,a wholly-owned s
255、ubsidiary of NEC corporation,offers mission-critical digital transformation solutions to service providers around the globe.Our comprehensive portfolio of software solutions and professional services enables large-scale digital transformations,unlocking the opportunities of the cloud,virtualization
256、and the changing mobile ecosystem.With an unbroken service delivery track record of more than 25 years,our unique combination of technology,people and expertise helps companies transform their networks and enable better experiences for their customers.33inform.tmforum.orgSPONSORED FEATURE34A bluepri
257、nt for intelligent operations fit for the 5G eraThe TM Forum Open Digital Framework(ODF)provides a migration path from legacy IT systems and processes to modular,cloud native software orchestrated using AI.The framework comprises tools,code,knowledge and standards(machine-readable assets,not just do
258、cuments).It is delivering business value for TM Forum members today,accelerating concept-to-cash,eliminating IT&network costs,and enhancing digital customer experience.Developed by TM Forum member organizations through our Collaboration Community and Catalyst proofs of concept,building on TM Forums
259、established standards,the Open Digital Framework is being used by leading service providers and software companies worldwide.The framework comprises TM Forums Open Digital Architecture(ODA),together with tools,models and data that guide the transformation to ODA from legacy IT systems and operations
260、.Open Digital Architecture Architecture framework,common language and design principles Open APIs exposing business services Standardized software components Reference implementation and test environmentTransformation Tools Guides to navigate digital transformation Tools to support the migration fro
261、m legacy architecture to ODAMaturity Tools&Data Maturity models and readiness checks to baseline digital capabilities Data for benchmarking progress and training AIGoals of the Open Digital FrameworkThe aim is to transform business agility(accelerating concept-to-cash from 18 months to 18 days),enab
262、le simpler IT solutions that are easier and cheaper to deploy,integrate and upgrade,and to establish a standardized software model and market which benefits all parties(service providers,their suppliers and systems integrators).Learn more about member collaborationIf you would like to learn more abo
263、ut the Open Digital Framework,or how to get involved in the TM Forum Collaboration Community,please contact George Glass.tm forum open digital frameworkinform.tmforum.org35inform.tmforum.orgtm forum research reportsknowledgeREPORTdataSPONSORED BY:governancethe growingimportance ofAuthor:Charlotte Pa
264、trick,Contributing AnalystEditor:Dawn Bushaus,Managing Editor December 2021 inform.tmforum.orgknowledgeREPORTverticalsplacing theright betsSPONSORED BY:enterpriseAuthor:Dean Ramsay,Principal AnalystEditor:Ian Kemp,Managing Editor December 2021 inform.tmforum.orgknowledgeREPORTlessonslearnedon the jo
265、urneyto cloud nativeAuthor:Mark Newman,Chief AnalystEditors:Dawn Bushaus,Contributing EditorIan Kemp,Managing EditorOctober 2021 inform.tmforum.orgSPONSORED BYimprovingAuthor:Ed Finegold,Contributing AnalystEditor:Ian Kemp,Managing Editor January 2022 inform.tmforum.orgBENCHMARKSPONSORED BY:SUPPORTE
266、D BY:Authors:Mark Newman,Chief Analyst Dean Ramsay,Principal AnalystEditor:Ian Kemp,Managing EditorMay 2022 inform.tmforum.orgAuthors:Mark Newman,Chief AnalystDean Ramsay,Principal AnalystDawn Bushaus,Contributing AnalystEd Finegold,Contributing AnalystEditor:Ian Kemp,Managing Editor January 2022 in
267、form.tmforum.orgNovember 2021|www.tmforum.orgnext generationthe Telefnica wayAuthor:Mark Newman,Chief Analyst,TM ForumEditor:Annie Turner,Contributing Analystsponsored by:with the support of:March 2022|www.tmforum.orgAuthor:Ed Finegold,Contributing AnalystEditor:Dawn Bushaus,Contributing EditorREPOR
268、T COVERCALL TO ACTIONAuthors and editors:xxxxxxknowledgeREPORTSponsored by:for integrationsetting new standardsDIGITAL ECOSYSTEMSImage used on tileReport CoverKey WordsBridging boundaries with common standardsWorking together Handing over common standards to each otherLogistics chainsCooperationJune
269、 2022|www.tmforum.orgAuthor:Dr.Mark H.Mortensen,Contributing AnalystEditor:Dawn Bushaus,Contributing EditorDIGITAL ECOSYSTEMS setting new standards for integrationsponsored by:August 2022|www.tmforum.orgAuthor:Ed Finegold,Contributing AnalystEditor:Dawn Bushaus,Contributing Editorsponsored by:REPORT
270、Sponsored by:Author:Annie TurnerEditor:Dawn BushausISBN:978-1-955998-27-7can telcos the into the future:August 2022|www.tmforum.orgAuthors:Annie Turner,Contributing AnalystDean Ramsay,Principal AnalystEditors:Ian Kemp,Managing Editor Dawn Bushaus,Contributing EditorEd Finegold,Contributing Analyst,T
271、M ForumIan Kemp,Managing Editor,TM ForumAnnie Turner,Contributing Analyst,TM ForumEditors:Author:Sponsored by:September 2022|www.tmforum.orgAuthor:Ed Finegold,Contributing AnalystEditors:Ian Kemp,Managing EditorAnnie Turner,Contributing Editorsponsored by:Ed Finegold,Contributing Analyst,TM ForumIan
272、 Kemp,Managing Editor,TM ForumAnnie Turner,Contributing Analyst,TM ForumEditors:Author:Sponsored by:REPORTAuthors:Sponsored by:Dean Ramsay(Principal Analyst)Editor:Ian KempISBN:000 from toautonomous networks:August 2022|www.tmforum.orgAuthor:Dean Ramsay,Principal AnalystEditor:Ian Kemp,Managing Edit
273、orsponsored by:October 2022|www.tmforum.orgAuthor:Rahul Gupta,Senior AnalystEditor:Ian Kemp,Managing Editorsponsored by:mainframemodernization:charting a course to cloud nativecharting a course to cloud nativeREPORTAuthor:Sponsored by:Rahul Gupta,Senior AnalystEditor:Ian Kemp,Managing Editor,TM Foru
274、mmainframemodernization:February 2022|www.tmforum.orgAuthor:Dean Ramsay,Principal AnalystEditor:Ian Kemp,Managing Editorsponsored by:next generationCEMnow!February 2022|www.tmforum.orgAuthor:Dr.Mark H.Mortensen,Contributing AnalystEditor:Dawn Bushaus,Contributing Editorsponsored by:October 2022|www.
275、tmforum.orgAuthor:Teresa Cottam,Contributing AnalystEditor:Dawn Bushaus,Contributing Editorsponsored by:from transformationDIGITAL OPERATIONS MATURITY:achieving business valueREPORTAuthors:Mark Newman,Chief Analyst,TM ForumDawn Bushaus,Contributing Analyst,TM ForumSponsored by:Editor:Ian Kemp,Managi
276、ng Editor,TM Forumestablishing links:platform models in the Open API economy March 202336Report Design:Intuitive Design UK Ltd infointuitive-design.co.ukPublished By:TM Forum,4 Century Drive,Parsippany,NJ 07054,USAwww.tmforum.orgPhone:+1 973-944-5100Fax:+1 973-944-5110ISBN:978-1-955998-49-9 2023.The
277、 entire contents of this publication are protected by copyright.All rights reserved.The Forum would like to thank the sponsors and advertisers who have enabled the publication of this fully independently researched report.The views and opinions expressed by individual authors and contributors in thi
278、s publication are provided in the writers personal capacities and are their sole responsibility.Their publication does not imply that they represent the views or opinions of TM Forum and must neither be regarded as constituting advice on any matter whatsoever,nor be interpreted as such.The reproduct
279、ion of advertisements and sponsored features in this publication does not in any way imply endorsement by TM Forum of products or services referred to therein.meet the Research&Media teamReport Editor:Dawn BushausContributing Analystdbushaustmforum.orgIan Kemp Managing Editorikemptmforum.orgGlobal A
280、ccount Director:Carine Vandeveldecvandeveldetmforum.orgDigital Media Coordinator:Maureen Adongmadongtmforum.orgCommercial Manager:Tim Edwardstedwardstmforum.orginform.tmforum.orgMark NewmanChief Analystmnewmantmforum.orgReport Author:Teresa CottamContributing AnalystSponsor Success Manager:Maryssa R
281、amseymramseytmforum.orgMarketing Manager:Ritika Bhatejarbhatejatmforum.orgHead of Operations:Ali Grovesagrovestmforum.orgPractice Lead:Dean Ramsay dramsaytmforum.orgJoanne TaaffeEditor in Chief,Inform:jtaaffetmforum.orgto learn more about TM Forums customer experience and AI governance projects,please contact Aaron Boasman-Patel