《BCG&Etis：2024年電信IT基準(TeBIT)-高管調查報告：AI賦能電信智能化 vs 運營商AI智用困局（英文版）（18頁）.pdf》由會員分享，可在線閱讀，更多相關《BCG&Etis：2024年電信IT基準(TeBIT)-高管調查報告：AI賦能電信智能化 vs 運營商AI智用困局（英文版）（18頁）.pdf（18頁珍藏版）》請在三個皮匠報告上搜索。

1、JANUARY 2025AI Makes Telcos Smart.But Are Telcos Smart About AI?TeBIT 2024 Executive ReportAsk a telecom company the big questionshow do you spark growth,boost efficiency,or improve customer satisfactionand chances are,AI is their go-to answer.And increasingly,artificial intelligence,including gener

2、ative AI(GenAI),has become telcos answer to other questions.How,for instance,do you fight back against telephone scammers preying on your customers?For one operator,its by developing an AI-generated“granny”that wastes fraudsters time with rambling talk about knitting.Thats the beauty of AI:Its a mul

3、ti-purpose enabler.But are telcos truly poised to unleash the technology?Are they making the investments,developing the strategies,and reaching the maturity levels that will let them take AI in new directions?Findings from this years telco IT benchmark(TeBIT)study,jointly developed by ETISThe Commun

4、ity for Telecom Professionalsand Boston Consulting Group,raise concerns that operators may not be realizing,or be poised to realize,the full value of AI and GenAI.(See“About the TeBIT Benchmark.”)About the TeBIT BenchmarkTeBIT,a survey of European operators IT spending and performance,was completed

5、in the fall of 2024 and reflects their 2023 financials and operations.The study sheds light on how participants are adapting to new trends and challengesand zeros in on where telcos may want to focus their efforts.Each year,TeBIT takes a deep dive into a topic of particular interest to telcos.In pre

6、vious reports,weve looked at sustainability and process efficiency,the role of cloud services,and the impact of COVID-19 on operators,among other areas.Collaboration is a key component of TeBIT.The surveys goallike that of ETIS working groups and community gatheringsis to identify,and even shape,the

7、 best practices that can help telcos better serve customers in a rapidly changing world.In return for allowing their organizations to be compared with other telcos,TeBIT participants can access a full set of benchmark results along with further trend analysis.The TeBIT study looked at participants A

8、I standing from four vantage points:maturity,investments,use cases,and expected and realized benefits.While the study found that as a group,participants surpassed other industry averages on AI maturity(topping scores from the media,software,and utility and power industries,for example),it also revea

9、led that scaling is uneven,with telcos faring better in some functions,like marketing and sales,than othersnotably,network and operations.The distribution pattern suggests that telcos are struggling to deploy AI in areas where out-of-the-box solutions are less prevalent and the necessary data is har

10、der to access.This challenge spotlights the importance of investment in the talent,processes,and technologies that help telcos scale AI use cases efficiently and across functions.It also underscores the need for a strategic approach to AIone where telcos are not building isolated use cases,but resha

11、ping critical functions end to end and inventing new experiences,offerings,and business models.Our analysis suggests,however,that telcos may not be investing or even,crucially,planning to invest sufficiently.In addition,their approach may be more ad hoc than strategic.Tipping Point or Topping Out?Th

12、e TeBIT study finds IT spending up slightly,from 6.6%of revenues in 2022 to 6.9%in 2023.The more significant trend is a shift from capex to opex.This wasnt unexpected:Moving from on-premise data centers and BSS and OSS solutions to third-party cloud servicesa core pillar of telco digital transformat

13、ionmeans less capex,more opex.But it does make an even stronger case for leveraging the efficiencies of AI.So how,exactly,are telcos positioned on the technology?The TeBIT survey asked respondents to rate their maturity on two dozen capabilities relevant to AI and GenAI.These capabilities covered ei

14、ght areas:strategy,innovation,customer experience,operations,technology,data,operating model,and people.In preparing an October 2024 report on AI,BCG asked these same questions of companies across industries.The results from that study enabled us to see where TeBIT participants sit on a cross-indust

15、ry spectrum.Based on their overall score,respondents fell into one of four buckets:AI stagnating,AI emerging,AI scaling,and AI future-built(in order of increasing AI maturity).On average,TeBIT participants ranked in the third-highest category,AI scalinghigher than global telcos ranked as a group(AI

16、emerging).But no participant ranked in the highest category,AI future-built.And drilling down,we saw that the TeBIT group lagged other industries in adopting AI in two key areas:customer experience and operations(though they were better positioned on AI enablers like technology and data).The underpe

17、rformance in customer experience and operations was somewhat surprising,as these are typically focal points for companies that embrace AI and GenAI.On the other hand,telcos have unique areas where these technologies have high potential,such as network automation.So their focus may have a wider sprea

18、d.Either way,the takeaway is clear:Telcos would benefit from developing AI maturity and use cases in customer experience and operations.This raises the issue that developing,building,and scaling use cases requires investment.The TeBIT study found that on average,participants AI and GenAI spending am

19、ounted to 0.16%of revenues in 2023,less than both the global telco industry average(0.49%)and the cross-sector average(0.42%).Further,we see evidence that investment may be leveling off.For 2024,TeBIT participants expect their AI and GenAI spending to amount to 0.24%of revenues before dropping to 0.

20、2%for 2025.This,too,is surprising.We would expect investment levels to increase for at least the next few years.This“topping off”could be temporary,but if it does signal an investment peak,that begs the question:Are telcos deploying and planning to deploy AI and GenAI as they should?Of course,invest

21、ment is about people,too.On average,participants anticipate that by the end of 2024,1.27%of their total workforce would be dedicated to AI and GenAI through activities such as developing AI use cases,training LLMs,and integrating AI tools into the IT landscape.This is up from 1.0%in 2023.And telcos

22、expect that,all told,they will have upskilled 10%of their employees in AI and GenAI knowledge and use during 2024.How many of these employees will go on to work with AI and GenAI?Thats an open question.But the upskilling does demonstrate that telcos are committed to addressing the impact of the tech

23、nology on their people.Deploying AI Technology at ScaleTo see where telcos stand on putting AI and GenAI into action,we compiled a list of common use cases,based on BCG experience,discussions in the ETIS community,and out-of-the-box solutions.The use cases covered areas including customer service(co

24、nversational chatbots,automated email generation,real-time guidance for agents);marketing and sales(next-gen sales assistants,among other use cases);revenue management and finance(churn management,personalized cross-selling and upselling);and digital,data,and IT(code generation,cybersecurity,and so

25、on).On average,TeBIT participants have implemented at scale 26%of the predefined use cases.But across functions,it wasnt an even spread.Marketing and sales led the pack,with telcos deploying 40%of these use cases at scale.Digital,data,and IT followed at 38%.Other areas lagged far behind.In product a

26、nd offer management,telcos deployed just 17%of use cases at scale,but perhaps more tellingly,they hadnt even considered 67%of our list of use cases.In network and operations,where only a quarter of use cases were running at scale,telcos had tried and abandoned 5%.While marketing and sales is a key l

27、ever for stabilizing revenues,and thus would seem a logical focus for AI and GenAI,theres another perspective worth considering.The data that these use cases need tends to be independent from operational data,which is often spread across the enterprise.There is also a growing portfolio of out-of-the

28、-box(or nearly so)AI and GenAI solutions.These factors often make marketing and sales use cases easier to scale than,say,network use cases,which typically require customized solutions and much broader data.So,are telcos scaling the use cases that they need to scale or those that they can scale?If th

29、e answer is the latter,thats another signal that telcos may want to redouble investments.We also found that telcos often go their own way on use cases.We asked participants to describe use cases they have implementedor have in the worksthat werent on the predefined list.Telcos reported that they hav

30、e implemented at scale 43%of this bespoke list.Given that nearly 40%of the additional use cases were in marketing and sales,which may be easier to scale,the higher proportion of successesuse cases that have moved beyond pilots or otherwise limited deploymentmight not be surprising.The majority of th

31、ese company-specific use cases were essentially more specialized or customized versions of the base list applications.Still,it speaks to the experimentation telcos are conducting in AI and GenAI as they try different tools to learn what works best for their needs.These patterns of investment,deploym

32、ent,and experimentation suggest that telcos are still taking a segmented,ad hoc approach to AI and GenAI and developing use cases opportunistically rather than strategically.Companies in other industries that are considered to be leading in AI arent deploying use cases one at a time,when and where p

33、ossible,but in a structured way,from a well-defined blueprint.While telcos talk about the transformative impact of AI,theyre not yet using it to activate a transformation.Doing so requires telcos to invest strategically.Its not enough to bump up the funding(though thats important).Telcos need to kno

34、w where to invest,to prioritize use cases in a way that aligns with business value and organizational goals.They shouldnt treat AI as an add-on,but as a core driver of efficiency and growth.And they should plan their investment volume and priorities accordingly.At the same time,telcos need the flexi

35、bility to seize new possibilities as both AI technologies and operators AI maturity evolve.To this end,telcos should adjust their tech stacks to enable and accelerate the development and integration of AI and GenAI.They should optimize processes and spark an AI innovation culture.By taking these ste

36、ps,telcos can quickly identify,build,and scale use casesand AI can live up to its billing.On overall AI maturity,TeBIT participants top a range of cross-industry averages but trail sector leadersSources:TeBIT 2024;BCG Build for the Future 2024 Global Study(merged with Digital Acceleration Index);BCG

37、 analysis.Note:Maturity scores were grouped into four categories:AI stagnating(score of 025);AI emerging(score of 2550);AI scaling(score of 5075);and AI future-built(score of 75100).AI stagnatingAI emergingAI scalingAI future-builtAre taking minimal or no AI action,lack foundational capabilities,and

38、 are not generating valueHave developed foundational capabilities and started initial experimentation but are struggling to scale and generate valueHave developed an AI strategy and advanced capabilities,and are scaling them effectively while starting to generate valueAre at the forefront of AI inno

39、vation,systematically building cutting-edge AI capabilities across functions and consistently generating substantial valueCross-industry AI maturity scoresTeBIT participants maturity scoresLowest/highest performerAverageMediaPower,utilities,and renewablesSoftwareDevices,components,and semiconductors

40、Telecommunications(global)Industry leaderIndustry averageTelcos lag in AI maturity in two key areas:customer experience and operationsSources:TeBIT 2024;BCG Build for the Future 2024 Global Study(merged with Digital Acceleration Index);BCG analysis.Note:Maturity scores were grouped into four categor

41、ies:AI stagnating(score of 025);AI emerging(score of 2550);AI scaling(score of 5075);and AI future-built(score of 75100).(AVERAGE MATURITY SCORE IN EACH OF EIGHT KEY AREAS)AI and GenAI adoption in business functionsAI and GenAI enablersMediaPower,utilities,and renewablesSoftwareDevices,components,an

42、d semiconductorsTeBIT participantsAI and GenAI strategyInnovationCustomer experienceOperationsTechnologyDataOperating modelPeople255075100AI stagnating AI emergingAI scalingAI future-builtAI and GenAI adoption in business functionsAI and GenAI enablersAI and GenAI strategyInnovationCustomer experien

43、ceOperationsTechnologyDataOperating modelPeopleCompared to cross-industry top performers,participants are on par only in strategy,technology,and operating modelSources:TeBIT 2024;BCG Build for the Future 2024 Global Study(merged with Digital Acceleration Index);BCG analysis.Note:Maturity scores were

44、 grouped into four categories:AI stagnating(score of 025);AI emerging(score of 2550);AI scaling(score of 5075);and AI future-built(score of 75100).(MATURITY SCORE IN EACH OF EIGHT KEY AREAS)Software(highest performer)Devices,components,and semiconductors(highest performer)Media(highest performer)Pow

45、er,utilities,and renewables(highest performer)TeBIT participants(average)255075100AI stagnating AI emergingAI scalingAI future-builtAI and GenAI adoption in business functionsAI and GenAI enablersAI and GenAI strategyInnovationCustomer experienceOperationsTechnologyDataOperating modelPeopleSources:T

46、eBIT 2024;BCG Build for the Future 2024 Global Study(merged with Digital Acceleration Index);BCG analysis.Note:Maturity scores were grouped into four categories:AI stagnating(score of 025);AI emerging(score of 2550);AI scaling(score of 5075);and AI future-built(score of 75100).AI maturity diverged m

47、ost in operations and innovation(MATURITY SCORE IN EACH OF EIGHT KEY AREAS)Highest performerLowest performerTeBIT participants average255075100AI stagnating AI emergingAI scalingAI future-builtSources:TeBIT 2024;BCG Build for the Future 2024 Global Study(merged with Digital Acceleration Index);BCG a

48、nalysis.Note:Figures for 2024 and 2025 are based on estimates provided by TeBIT participants.GenAI investments seem poised to level offand then drop.Whether that represents a blip or a trend remains to be seen.AI AND GENAI INVESTMENTS(%OF REVENUES)0.02023202420250.10.20.30.40.50.060.320.160.490.420.

49、100.420.240.100.300.2Cross-industry averageGlobal telecom industry averageHighAverageLowSource:TeBIT 2024;BCG analysis.Note:Figures for 2024 and 2025 are based on estimates provided by TeBIT participants.Along with AI technology,telcos are investing in AI talent and skillsPeople dedicated to AI and

50、GenAIAI and GenAI upskilling(%OF TOTAL WORKFORCE UNDERGOING TRAINING)(%OF TOTAL WORKFORCE)2023202420252023202420250.911.000.960.971.271.081.010.05.51.04.52.8HighAverageLowSources:BCG experience;ETIS community discussions;offerings from vendors specializing in telco business support systems and opera

51、tions support systems.Note:NOC=network operations center.To gauge how telcos put AI into action,we asked participants where they stood on a range of common use casesMarketing and SalesRevenue Management and FinanceCustomer ServiceNetwork and OperationsSupport FunctionsDigital,Data,and ITProduct and

52、OfferManagementPersonalized service and offer builder Product and service designProduct lifecycle managementPromotion effectivenessNext-gen sales assistantPersonalized recommendationsSocial media sentiment analysisDynamic pricing modelNext best offerPaid media and marketing optimizationConsumer pref

53、erence-based productsDemand forecasting for sales and inventory optimization Transport network planning and optimizationIntelligent knowledge management and document processingAutomated processes for HR,finance,and other functionsCustom white papers,blogs,commercials,and onboarding contentPreventive

54、 churn management Personalized cross-and up-sellingVoice assistant/voicebotReal-time guidance for agentsRoot cause analysis and resolutionProactive service messagesGenerate onboarding contentCollections managementCreate draft financial statements/budgetsConversational chatbotAutomated emailingDigita

55、l assistantCybersecuritySmart energy savingsIncident managementDigital twinZero-touch NOCIncident managementDigital twinSmart energy savingsNetwork automationCredit-rating modelsPayment matchingPersonalized billingNetwork anomaly and fault detectionPreventive/predictive maintenanceField force labor

56、optimizationNetwork rollout and optimizationNetwork configuration and asset optimizationBuild dashboards and reporting toolsSummarize legal documentsField force labor optimizationGenerate preliminary codeAutomate DevOps and ML OpsRoot cause analysis and resolutionTranscribe meetings and create summa

57、riesConvert legalese to plain languagePersonalized email scriptsFraud detection modelsSources:TeBIT 2024;BCG analysis.Note:POC=proof of concept;figures may not add to 100 due to rounding.Telcos have had the most success implementing marketing and sales use cases at scaleImplemented at scaleImplement

58、ed,but not yet at scaleIn POC or experimental phasePlanned or under considerationNot consideredTried but discontinued67232917813823382530252081017154781710251317402525153825Product and offer managementMarketing and salesRevenue managementand financeCustomer service5Network andoperations4Digital,data

59、,and ITSupport functions7(%OF PREDEFINED USE CASES WITHIN DOMAIN)3Use cases in product and offer management,as well as network and operations,have proved more challenging to scale Marketing and SalesRevenue Management and FinanceCustomer ServiceNetwork and OperationsSupport FunctionsDigital,Data,and

60、 ITProduct and OfferManagementPersonalized service and offer builder Paid media and marketing optimizationPreventive churn management Conversational chatbotNetwork anomaly and fault detection*Summarize legal documentsField force labor optimizationSources:TeBIT 2024;BCG analysis.Note:NOC=network oper

61、ations center;POC=proof of concept.*=Low response rate 33%Product and service designProduct lifecycle managementPromotion effectivenessNext-gen sales assistantPersonalized recommendationsSocial media sentiment analysisDynamic pricing modelNext best offerConsumer preference-based productsDemand forec

62、asting for sales and inventory optimization Transport network planning and optimizationIntelligent knowledge management and document processingAutomated processes for HR,finance,and other functionsCustom white papers,blogs,commercials,and onboarding contentPersonalized cross-and up-sellingVoice assi

63、stant/voicebotReal-time guidance for agentsRoot cause analysis and resolutionProactive service messagesGenerate onboarding contentCollections managementCreate draft financial statements/budgetsAutomated emailingDigital assistantCybersecuritySmart energy savingsIncident managementDigital twinZero-tou

64、ch NOCDigital twinSmart energy savingsIncident managementNetwork automation*Credit-rating modelsPayment matchingPersonalized billingPreventive/predictive maintenanceField force labor optimizationNetwork rollout and optimization*Network configuration and asset optimizationBuild dashboards and reporti

65、ng toolsGenerate preliminary codeAutomate DevOps and ML OpsRoot cause analysis and resolutionTranscribe meetings and create summariesConvert legalese to plain languageNot considered/tried but discontinuedConsideredPOC or experimental stageImplemented,but not yet at scaleImplemented at scalePersonali

66、zed email scriptsFraud detection modelsSources:TeBIT 2024;BCG analysis.Note:Average score covers all seven use case domains;high and low scores apply to a single(highest/lowest performing)domain.On average,TeBIT participants have scaled 26%of the predefined use cases and have scaled 43%of the additi

67、onal use cases implemented.Base AI and GenAI use cases implementedat scaleAdditional AI and GenAI use cases implementedat scale 0102030607080(%OF USE CASES ON PREDEFINED LIST)24.571.40102030407080(%OF ADDITIONAL USE CASES LISTED BY EACH PARTICIPANT)22.266.726.0 average43.0 averageSources:TeBIT 2024;

68、BCG analysis.Note:POC=proof of concept.Additional use cases were most prevalent in marketing and sales,followed by support functions(%OF BESPOKE USE CASES)7714331738731077140331431028Implemented at scaleImplemented,but not yet at scaleIn POC or experimental phasePlanned or under considerationDomain

69、contains use cases mentioned by multiple respondentsProduct and offer managementMarketing and salesRevenue managementand financeCustomer serviceNetwork andoperationsDigital,data,and ITSupport functionsExample use cases:Routing to best agents Next best action Content creationExample use cases:Cogniti

70、ve/optimized routing to agentsExample use cases:GenAI-powered assistant or chatbot Robotic processautomationSources:TeBIT 2024;BCG analysis.Greater efficiency and productivity is the number-one benefit telcos expect or have realized from AI and GenAI(%OF RESPONSES;PARTICIPANTS COULD CHOOSE MORE THAN ONE)581786641Improve customer targetingImprove business steeringReduce risk or churnImprove the brand and telco/operator perception in the marketImprove customer satisfactionIncrease revenueIncrease efficiency and productivity