1、1Ericsson|AI business potential understanding the value of AI for telecom operationsAI business potentialunderstanding the value of AI for telecom operations2Ericsson|AI business potential understanding the value of AI for telecom operationsConsumer proofpointsAI will deliver value across industries

2、,including the telecommunication sector.For communication service providers(CSPs),AI will provide opportunities to optimize network operations,improve customer experience,reduce costs,contribute to sustainability,enable new revenue streams,and more.However,AI implementation is a practice more than a

3、 technology;it is a more efficient way of working that takes automation to the next level.As such,it is challenging to discern and convey its value since it is often indirect.To help CSPs understand AI value creation,we surveyed industry professionals about the role of AI in their organizations and

4、developed an industry-unique standardized measurement framework to help convey the value.01 Telecom operations are increasingly complex 02 AI carries benefits across CSP organizations 03 Measure AI impact in business value 04 ConclusionPreface3Ericsson|AI business potential understanding the value o

5、f AI for telecom operationsTelecom operations are increasingly complex 01CSPs will face a multitude of operational complexities going forward,and those that outperform in operations are likely to be the most successful.Telecom operation in the future Telecom operations are complex to run,requiring s

6、imultaneous,coordinated,and dynamic ways of working throughout all business units.Complexities in product portfolios and across operations drive up communication service providers(CSP)costs between 20 and 50 percent.Therefore,CSPs that outperform within operations are often more successful.Going for

7、ward,we see multiple trends driving complexity:More use cases Instead of supporting a few straightforward use cases such as voice,data,and text,networks will have to deal with dozens of different use cases,all delivered at the same time.Data growth Many use cases will be dependent on high-capacity n

8、eeds.It could be remote controlling a vehicle in a mine that navigates with the help of multiple 8K cameras,or streaming real-time AR feeds to a mobile headset.More devices With the growth of the Internet of Things(IoT),many more devices,and increasingly different kinds of them,will connect to the n

9、etwork.Security concerns As more use cases are connected and increasingly critical,the demands on security rise dramatically.Continued customer experience demands Not only will networks have to cope with diverse use cases,but they also need to be able to deliver a great experience for each of them.N

10、ew revenue opportunities The enterprise market is hungry for new,relevant services to accelerate their digitalization journeys.CSPs need to bring new services to market at a faster pace to mee these demands.How AI can helpFor those that cut through the complexity,a range of benefits await.In a study

11、 conducted with telecom strategy consultancy Arthur D.Little,covering 80+CSP professionals across the world,71 percent mentioned improved customer experience as one of the primary benefits of implementing AI in their operations.Within customer experience,AI use cases relate to more efficient handlin

12、g of customer queries and improving the network.In the close second place,with 63 percent of respondents,came optimizing network operations.Within network operations,primary benefits lie in anomaly detection and management.Many CSPs have automated what they can to cope with the increasing complexity

13、.To take automation to the next level,companies need AI,as it generates insights that enable the automation of things the company may not have previously known.State of CSP AI usageFor most CSPs,utilizing the power of AI is a rather new phenomenon.Amongst the surveyed professionals,about 50 percent

14、mentioned having implemented AIGraph 1:Main AI benefits according to CSPs,as a percentage of respondents naming it within the top 5.10%Supporting new product and service developement14%Improving sustainability19%Increasing sales24%Enhancing existing products and services29%Decreasing operation expen

15、ses(OPEX)31%Improving decision-making39%Improving employee efficiency63%Optimizing current network operations71%Improving customer experience“We have received extremely good results.Quite remarkable.”Frank van der Rijst,Deutsche Telekom,Product Manager NOC,Europe4Ericsson|AI business potential under

16、standing the value of AI for telecom operationsin the past three years,partly driven by the increased focus on operational efficiency due to the COVID-19 pandemic.Most of the surveyed CSPs did have some form of AI deployment ongoing or in the pipeline.For those CSPs that have deployed AI,the primary

17、 areas of their business where they are utilizing its power are customer and service experience,security,and IT operations.Adoption enablersWhen implementing AI solutions,the main challenges that CSPs face are related to security and data privacy.The issues are connected,but different.Security conce

18、rns regard unauthorized access to data,stemming from a lack of control when scattered data across the organization is pooled.Data privacy concerns stem from authorized,but perhaps unwanted,access to data,where one part of the organization may be reluctant to share its data with another part.One addi

19、tional top challenge is that of AI integration.According to CSPs,these challenges point to a larger issue that of organizational acceptance of AI solutions.Worries around data privacy,security,and integration may be symptomatic of this issue.However,to build innovative programs with the use of AI,da

20、ta is crucial.The front-runners,those that have come the furthest with their AI implementation,address this issue head-on.Their implementation is user-centric;they start by addressing common challenges faced by organizational users,which highlights the benefits,thereby outweighing potential concerns

21、.In addition,some run organization-wide events and training to build awareness.One front-runner,who had worked extensively with overall organizational buy-in and succeeded,mentioned being faced with another challenge adopting data analytics in the steering of the company.Business decisions,such as s

22、ales targets,were made based on suboptimal data which was dependent to a degree on guesswork.Currently,the company is working to optimize their decision-making by building granular AI predictions into these types of steering decisions,increasing their accuracy.Read more on how the front-runners deal

23、 with this challenge in the next part of this report-AI business potential:unleashing the value of AI for telecom operations.“It became a very hot topic.Especially during and post COVID-19,we tried more to deflect transactions smartly due to the availability of workforce.”Salam Ibrahim Al Khafaji,Di

24、rector Customer Operations Development,DU,Middle-East“User acceptance is a challenge.When it comes to managing things,you need to see the value in this domain,then they change their mind.”Chief Network Planning officer,Middle-EastAbout 66%of CSPs surveyed had implemented AI to some degree.66%Graph 2

25、:AI implementation challenges,as a percentage of respondents who have implemented AI65%Data privacy59%Securityconcerns52%Problem to integrate AI48%Fear of losing control due to automation48%Cost of AI technologies/solution development5Ericsson|AI business potential understanding the value of AI for

26、telecom operationsThe future of AI in telecomAs the adoption of AI in telecom continues to increase,most CSPs will focus their efforts on improving customer and service experience,security,and network operations.Where CSPs see potential going forward is similar to the areas they are currently invest

27、ed in.One potential explanation for this continued focus is the fact that AI was introduced fairly recently in telecom,and will need time to mature.Technology enablersAs AI technology becomes more and more advanced,the value it brings to CSPs increases.The current use cases that support CSP operatio

28、ns and allow for partial automation of processes will become increasingly complex,move towards working fully autonomously and allow for full process automation.This evolution towards intent-driven management and zero-touch operations will be dependent on several key technologies.These can be divided

29、 into three areas:AI,tools,and connectivity.(See Figure 1.)Within the AI area,next-generation use cases will be dependent on the transition from the machine and deep learning to machine reasoning,which enables machines to act with agility on learned data.Within tools,the transition towards end-to-en

30、d MLOps platforms for developing,executing and life-cycle managing AI solutions and new connectivity technologies will also be important for driving next-generation use cases.These technology enablers will allow CSPs to move from reactive operations to proactive,more data-driven service delivery.“A

31、couple of countries are collaborating with hyperscalers;they have the OTT data.If the industry brings their heads together,we can correlate this data and get a better customer experience.”Frank van der Rijst,Product Manager NOC,Deutsche Telekom,Europe“A multitude of use cases within closed-loop auto

32、mation.And then the system does it all itself.From the subscriber side of things,we are also working on churn.Whether a subscriber has a high likelihood to churn.Zero-touch operation is the goal.Its the journey.”Ajay Jain,Head of Automation and Digitalization,Airtel,South East Graph 3:Potential goin

33、g forward,as a percentage of respondents“For operations,you need more real time data,24/7,and be more proactive than reactive.And as data becomes more and more prevalent,the challenge will be on how can we drive the whole organization to be data-driven.”Nabeel Alheider,Advanced Analytics GM,STC,Midd

34、le-EastSecurity64%57%40%Customer&service experienceNetworkoperationsKey technology enablersUse casesNon-Exhaustive?AIToolsConnectivityMachine&deep learningAI management applications4G&first wave 5GMachine reasoningEnd-to-end platformSecond wave 5G&6GAI allows CSPs to move from reactive towards a pro

35、active enabler of service deliveryAI used for partialautomationAI supportingoperators Intelligentalarm andincidentcorrelationAPI testautomationZero touchoperationsIntent drivenoperationsAI used for fullautomationAI workingautonomously Batterybackup predictionTower climbpredictionCurrent use casesNex

36、t generation use casesValue of AI for CSPsFigure 1:AI going forward and technology enablers 6Ericsson|AI business potential understanding the value of AI for telecom operationsAI carries benefits across CSP organizations02AI use case opportunities AI carries benefits across CSP organizations,involvi

37、ng hundreds of use cases.A majority of them are specific to telecom and require both deep telecom domain expertise and AI expertise.Navigating them can be challenging for CSPs.To guide CSPs to AI use cases related to their business goals,we have clustered hundreds of use cases into a conclusive fram

38、ework.This framework is based on the familiar TAM TMF framework,which outlines the various parts of CSP organizations.For example,if you are looking to improve your network operations,look to use cases within this cluster.(Example use cases are on the next page.)In total,we have clustered use cases

39、into 10 categories,ranging from customer and service experience to telco enterprise management.AI can optimize processes in various parts of CSP organizations,from customer and service experience to network optimization.7Ericsson|AI business potential understanding the value of AI for telecom operat

40、ionsFigure 2:AI use case clusters AI use case clustersDescriptionCustomer&Service ExperienceAI use cases related to improving the overall customer experience and/or specific service experience,also covering use cases for customer relationships,channels,sales,and marketingApplication Development&Mana

41、gementAI use cases related to the development and management of different applications and capabilities,such as smart code completion and optimized testingNetwork Planning&DesignAI-powered network planning,design,evolution,and deploymentNetwork OptimizationAI use cases related to the optimization an

42、d improvement of network performance and efficiencyNetwork OperationsAI use cases related to improving the end-to-end operations and automation of the network,including NOC and fieldIT OperationsAI use cases related to improving IT operations,performance,and service availability for systems such as

43、order fulfillment,charging,and billingCloud&Infrastructure Planning&DesignAI use cases related to improving the cloud and infrastructure planning and design,such as capacity predictionCloud&Infrastructure OperationsAI use cases related to cloud and infrastructure operations to improve effectiveness,

44、availability,and stabilitySecurityAI use cases related to IT and network securityTelco Enterprise ManagementUse cases related to improving and automating general enterprise applications8Ericsson|AI business potential understanding the value of AI for telecom operationsLTE RET OptimizationRemotely ad

45、just and optimize the electronic tilt of antennas.Helps to address coverage area and field density,and thereby minimize mobile coverage holes and interference issues.Cell issue root cause classificationIdentify the root cause of problems quicker.Allows CSPs to remove manual analysis that takes longe

46、r and decreases the effect of the issue on customer satisfaction.Capacity planningPredict capacity needs in advance;plan investments with maximal RoI using cell-level traffic forecasting and automatic dimensioning based on different services and performance requirements.Cell site energy managementPe

47、rform auto soft lock and unlock of radio based on traffic patterns.Enables CSPs to make large energy savings that impact both OPEX and carbon emissions without affecting important user experience and service KPIs.Traffic balancingOptimize the utilization of network cells by predicting high cell traf

48、fic and dynamically route traffic.With the help of AI,network performance can be increased and user experience enhanced.NPS insightsPredict future net promoter scores and recommend how to improve.With the help of AI,better insights into NPS can be achieved,which can have a positive effect on revenue

49、 and churn rates,as CSPs can manage network experience more proactively.Capacity forecast for telco cloud stackDevelop correlations between NFVi usage and performance metrics.With the help of AI,CSPs can gain better visibility into NFVi requirements and forecast the expected capacity usage,allowing

50、for more effective resource planning.VNF service degradation correlation with telco cloudFaster troubleshooting for service-impacting incidents.Identify issues in the application layer and correlate them with other layers of CSPs clouds.This can help to increase service stability and both MTTI and M

51、TTR for service degradation issues.Infrastructure fault prediction for multi-cloudAccurate and automatic fault prediction and data drifts.Helps to reduce QoS degradation and improve efficiency,as the system is compatible with several types of faults and is reusable.Example use cases from Network Ope

52、rations cluster.Example use cases from Cloud&Infrastructure Operations cluster.AI use case examplesTo exemplify how AI brings value to CSP organizations,we include three use case examples from three clusters.These clusters are of importance to CSPs and have a high adoption rate.They have been chosen

53、 to provide detailed insight into how AI can be leveraged to improve existing operations across CSP organizations.Example use cases from Network Optimization cluster.9Ericsson|AI business potential understanding the value of AI for telecom operationsHow to engageOver the next few years,all CSPs we s

54、poke to expect to increase their AI budgets.As they look to leverage these budgets,certain capabilities are needed.Two of the top three capabilities to have in place in order to successfully adopt AI relate to a better understanding of how to use it and why.CSPs are looking for clear answers.In addi

55、tion,with poor input,you get poor output,which gives reason for the third top capability improved data management.To help,vendors can provide secure,easy-to-integrate,scalable,and flexible solutions.This explains the preference among many CSPs for purchasing their AI solutions as Software-as-a-Servi

56、ce.Graph 4:Preferred engagement model with suppliers,as a percentage of respondentsGraph 5:Capabilities required to adopt AI,as a percentage of respondents“You need a good data layer if you want to scale use cases.”Frank van der Rijst,Deutsche Telekom,Product Manager NOC,EuropeThe benefit of Softwar

57、e-as-a-Service is scalability.However,to consider this option,CSPs require data to be stored locally.When purchasing AI solutions,vendors must satisfy a list of security requirements;doing so is a hygiene factor,and the main concern is local data storage.SW-as-as-ServiceManaged serviceSW licensingBe

58、tter understanding of deploying newbusiness models 56%A thorough and alignedAI strategy and roadmap54%Improved management of data50%Developed internalprocesses47%Improved skillsets ofthe internal resources43%20%14%66%“From an operational perspective,SaaS is scalable;its easy.But its case to case.Dep

59、ends on the process;if its not important information,then fine with SaaS.”Director Customer Operations Development,Middle-East10Ericsson|AI business potential understanding the value of AI for telecom operationsWhen it is implemented,a majority of CSPs believe that AI can result in IT and network OP

60、EX reductions,as well as returns on investment of between 5 and 10 percent.5-10%The high impact of AI implementation could be a potential explanation for why CSPs,on average,scored the overall importance of AI as 4 on a scale of 15.Across all use case clusters,CSPs aim to address a wide range of ind

61、icators with the help of AI.Operational efficiency was deemed the most important to address,followed by complaints and security indicators.Graph 6:Expected annual return from AI investments,as a percentage of respondents Graph 7:Indicators to address with AI,as a percentage of respondents51%39%0%10%

62、Less than 5%Between 5%and 10%Between 10%and 20%More than 20%56%Operational efficiencyComplaintsSecurityNetwork availability Size&scopeUser throughputAutomationMean time to repairIssue prediction accuracy%prevented investmentsRevenue leakageTicket&dispatch reductionEnergy consumptionFaster time to ma

63、rketMean time to identify,ticket basedCO2 emissionsOrder falloutTest planning time reductionNet Promoter ScoreSpeed&scale Other54%53%49%43%33%31%24%24%21%20%20%20%20%19%17%16%11%9%0%0%11Ericsson|AI business potential understanding the value of AI for telecom operationsMeasure AI impact in business v

64、alue03Although AI can be seen to impact important KPIs,such as percentage of reduced complaints,time without incidents,and percentage of reduced network unavailability,these KPIs might not always win in a budget discussion.To gain internal traction for AI implementation,measure impact in monetary te

65、rms.The question,then,is how to measure AI impact in a consistent,user-friendly way.To help CSPs further their AI projects,we undertook an effort to define such a framework.This industry-unique framework has four distinct steps,described further below.Step 1:Determine baselineInitially,a baseline is

66、 created for a base scenario in which no AI use case is deployed.Various input parameters are assumed to determine a baseline CSP for the calculation.Exact input parameters are determined by the specific AI use case.In general,a selection of 510 parameters is sufficient to have a solid base to calcu

67、late the impact on business value.Although these inputs may vary depending on the AI use case,examples include network OPEX,CAPEX,and revenue.Step 2:Specify the sought business outcomeWe must then specify the sought business outcome from implementing AI.AI is not an end goal;the end goal must be roo

68、ted in a business outcome.We have found four common types of business outcomes when deploying AI use cases:1.Reduced total cost of ownership Such as reduced CAPEX,network and IT OPEX,as well as SG&A.2.Increased revenue3.Enhanced customer experience Relates to service and support and network quality

69、of service.4.Improved sustainable business Environmental impact,risk and compliance,and strategy and brand impact.Step 3:Determine KPI drivers Having defined the intended business outcome,we must choose how to measure the impact.To measure the impact of AI,we must utilize KPIs relevant to the busine

70、ss outcome we are looking to improve.AI impacts a number of common CSP KPIs:For a reduced total cost of ownership Predicted capacity issues solved with optimization,spectral efficiency improvements,reduced MTTI and MTTR,and ticket and dispatch reduction.For increased revenue Reduced revenue leakage,

71、reduced order fallout,and increased ARPU.For enhanced customer experience Reduced complaints,improved NPS,reduced unavailability,and reduced number of outages.For improved sustainable business Reduced energy consumption and CO2 emissions,time without incidents,and more.Step 4:Select the AI use case

72、and assess the business impactTo select an AI use case,we must consider its attractiveness and implementability.(See Figure 3.)Finally,we measure the business impact using a structured approach.All data points are included in a single calculation tree.(See Figure 4)CSPs can measure AI impact through

73、 common KPIs,but to secure buy-in,it needs to be a business value.12Ericsson|AI business potential understanding the value of AI for telecom operationsStep 1:Determine baselineFar EasTone(FET),Taiwan,is building a 5G network targeting premium performance and coverage to maximize customer experience.

74、By August 2021,FET had achieved about 1 million subscriptions on this network.During this period,FET also had a monthly ARPU of about 20 USD 1.A conservative estimate of monthly revenues from these subscribers would then be about 20 mUSD and 240 mUSD annually.This is conservative since 5G subscripti

75、on plans are typically more expensive than others.Step 2:Specify business outcomeReturning to our example of FET,the sought business outcome is enhanced customer experience.Step 3:Determine KPI driversFor FET,which is looking to improve customer experience,the relevant KPI is improved NPS.Step 4:sel

76、ect the AI use case and assess the business impactIn order to increase customer experience through improved NPS,Far EasTone leverages AI-powered data analytics.For 5G network deployment,site selection is based on AI analytics of real traffic usage and user behavior.This enables a more precise cost-p

77、erformance analysis to prioritize the sites that most urgently need capacity enhancements.With an internal data analytics team in place and clear indications of positive KPI impact,this AI use case is a natural choice.Finally,to assess the business impact,we first look at the KPI impact,and then the

78、 baseline revenue impact associated with it.In the real-world case,FETs 5G subscriber NPS was about 18 p.p higher than for its 4G subscribers in September 2021,and AI was one of the key contributing factors.In one of the initial academic studies of NPS impact on revenue,it was found that a 7 percent

79、age point increase corresponded to a 1%increase in revenue growth 2.This would imply 18 p.p./7 p.p.,equaling about a 2.6 p.p.increase in revenue growth due to the increase in NPS.If we assume the previously calculated 240 mUSD in annual revenue stemming from new 5G customers to be the before-mention

80、ed growth,about 2.6%of 240 mUSD would be the first-year revenue impact of using AI for 5G network deployment for FET,i.e.,about 6.2 mUSD.Over five years,this accumulates to 32 mUSD.FETs 5G deployment is still in its infancy.For a tier-1 CSP with 270 m subscriptions on more mature technologies,this f

81、ive-year value would instead represent 32 mUSD x 270 m,i.e.,about 8.6 bUSD.This 270 m value is based on the average mobile subscriptions of 10 of the worlds largest communication service providers 3.The framework in practiceFigure 3:Business outcomes and KPI driversReducedMTTIReduced number of outag

82、esImproved NPSReduced order falloutReduced SG&A related costsReduced CO2 emissionsSize&scopeReducedsecurityandcompliance issuesTime withoutincidentsReducedenergyconsumptionReducedcomplaintsReducedrevenueleakageIncreased automationPredicted capacity issues solved with optimizationStrategy&brandRisk&c

83、omplianceEnviron-mentalistNW QoSService&supportRevenueSG&AIT OpexNW OpexReduced Total Cost of OwnershipImproved sustainable businessEnhanced customer experienceIncreasedRevenueCAPEXReducedunavailabil-ityTicket&dispatchreductionFastersightacceptanceKPIsBusiness sub outcomeBusiness outcome13Ericsson|A

84、I business potential understanding the value of AI for telecom operationsFigure 4:Choosing AI use casesLowHighUse case attractivenessLowHighUse case implement abilityCommentsThis calculation methodology is simple to outline and explain and works well with all types of use cases.The split between the

85、 specific baseline and use case variables helps break down the value drivers and makes it easy to conduct sensitivity analyses.Moreover,it makes it easy to modify the baseline and thereby assess the business value if the use case were to be deployed in a different context,market,or CSP altogether.Fi

86、gure 5:Choosing AI use casesMonetary change in business value%change in baseline value enabled by the AI use caseBaseline revenue/costCompany specific dataUse case KPI specific dataInput parameter 1Input parameter 2-KPI 1 impactKPI 2 impact-CommentsThis calculation methodology is simple to outline a

87、nd explain and works well with all types of use cases.The split between the baseline and the use case specific variables helps break down the value drivers and makes it easy to conduct sensitivity analyses.Moreover,it makes it easy to modify the baseline and thereby assess the business value if the

88、use case were to be deployed in a different context,different market or at a different CSP altogetherGeneralized viewSource:Arthur D.LittleUse case attractivenessBased on the impact of the use case on the selected KPIsUse case implementabilityBased on Capability build-up needs Capital investment req

89、uirements Customer,partner,and service diversification Execution risk profile14Ericsson|AI business potential understanding the value of AI for telecom operations15Ericsson|AI business potential understanding the value of AI for telecom operationsConclusion Reference04AI will deliver value across in

90、dustries.To capture this value,focus on the added business value.AI will deliver value across industries,including the telecommunications sector.However,AI implementation is a practice more than a technology;it is a more efficient way of working that is present in data-driven organizations.As such,i

91、t is challenging to discern and convey its value since it is often indirect.To aid CSPs in communicating this value,we surveyed and interviewed 80+CSP professionals on AI value creation in their organizations,and developed an industry-unique standardized measurement framework for AI impact in CSP or

92、ganizations.The main benefits that professionals see with AI is improving customer experience and optimizing current network operations.We also found that while many CSPs had reached mature and orchestrated states,a majority only had a pilot in place or had not yet started.The primary challenges the

93、y see include security and data privacy concerns.In addition,cultural barriers to technology adoption go a long way in explaining the current adoption rate.Read more in the next part of our report,Unleashing the value of AI for telecom operations,where we detail some of the actions the front-runners

94、 are taking to overcome these barriers.To highlight the range of AI opportunities for CSPs,we have developed a conclusive framework,categorizing AI use cases into 10 different clusters,such as network optimization and cloud and infrastructure operations.Finally,to communicate the business value of A

95、I,our industry-unique standardized measurement framework for AI value creation recommends four distinct steps:1.Determine baseline2.Specify sought business outcome3.Determine KPI drivers4.Select AI use case and assess business impact1.Ericsson,Mobility report(November 2021)2.Brand Strategy,Advocacy

96、drives growth(December 2005)3.Company websites;Arthur D.Little(July 2022)Interested in more details and recommendations for specific AI clusters?Find out more about selected AI clusters deep dives and customer case in our second brochure.Download from:https:/ content of this document is subject to r

97、evision without notice due to continued progress in methodology,design and manufacturing.Ericsson shall have no liability for any error or damage of any kind resulting from the use of this documentEEAB-22:006062 Uen Ericsson AB 2022Ericsson enables communications service providers to capture the ful

98、l value of connectivity.The companys portfolio spans Networks,Digital Services,Managed Services,and Emerging Business and is designed to help our customers go digital,increase efficiency and find new revenue streams.Ericssons investments in innovation have delivered the benefits of telephony and mobile broadband to billions of people around the world.The Ericsson stock is listed on Nasdaq Stockholm and on Nasdaq New York.