Roughly 80% of operators already use AI for analytics; nevertheless, these deployments are largely restricted to decision-support use circumstances
In partnership with Radcom, GSMA Intelligence has launched a brand new report, Service Assurance Traits within the AI Period, which revealed key focus areas for telcos in 2026 and paperwork a shift in direction of a extra customer-centric method by means of using unified assurance options, larger-scale agentic AI deployments, and real-time information infrastructure.
Prime takeaways from the report:
Finish-to-end visibility stays a significant hole regardless of “unified” platforms
Whereas 82% of operators declare RAN-to-core correlation, solely 41% have a real end-to-end information structure throughout domains and departments. This disconnect limits real-time perception and automation potential, whereas 21% say they’ve a mixture of each built-in structure and siloed information administration. The remaining 38% nonetheless have totally different groups or departments conduct information assortment and administration throughout their operations.
For GSMA Intelligence, this misalignment suggests that almost all of telcos might imagine that they’ve visibility, however most nonetheless lack the info basis required for AI-driven operations.
Service assurance is turning into customer-experience infrastructure, not only a community device
Operators’ major use of service assurance at the moment is buyer perception, churn discount, and Web Promoter Rating (NPS) enchancment, not pure community troubleshooting. “Operators rely on assurance options to provide them larger perception into buyer utilization and spending patterns, with a view to shoring up the general buyer expertise and driving NPS,” acknowledged the report.
Insights into buyer and utilization tendencies for upsell and advertising and marketing functions (53%) and analytics on buyer experiences to enhance expertise and cut back churn (52%) make up the highest two use circumstances for community information, adopted by community use circumstances: analyzing community issues on the core (48%) and for superior community troubleshooting (42%).
The takeaway is that greater than half of operators use community information primarily for buyer analytics and monetization, highlighting how assurance has shifted from OSS plumbing to a frontline buyer expertise system.
Right here, GSMA Intelligence additionally identified one other “disconnect,” this time between assurance and OSS/BSS. “The survey factors to a disconnect in that the a number of streams of information out there to operators from evaluation of their clients and components of the community aren’t essentially built-in with and streamed to present OSS/BSS,” it stated. “Whereas this can be work in progress, it might counsel a spot or alternative for present assurance platforms to completely combine with these methods.”
AI is already mainstream — however principally assistive, not autonomous
About 80% of operators already use AI for analytics reminiscent of anomaly detection, fault prediction, and root-cause evaluation. Nonetheless, these deployments are largely restricted to decision-support use circumstances, together with buyer grievance evaluation (63%), fault prediction (59%), and root-cause evaluation (55%), moderately than closed-loop, autonomous automation, which simply 39% have applied.
These findings present that whereas telecom is certainly injecting AI into its operations, it has but to take action in an AI-native method.
Agentic AI is the subsequent inflection level
Roughly 65% of operators see agentic AI as transformational or high-value, and over 70% of these planning deployments goal 2026. Precedence use circumstances embrace automated grievance decision (57%), autonomous fault decision (54%), and buyer expertise predictions (52%).
Moreover, almost half of the respondents (48%) stated that agentic AI can be useful for predictive community well being scoring.
Cell information high quality is now the dominant ache level
45% of service complaints relate to cellular information, particularly streaming and gaming, whereas voice generates minimal complaints. The low stage of complaints referring to voice isn’t a surprise,” famous GSMA Intelligence, explaining that voice is a “steady product section” and “extremely optimized” on the community. “Knowledge isn’t, nevertheless,” the report continued. “Relying on the precise operator, nation, protection areas, and so forth, information protection and capability can differ considerably, producing the best stage of complaints.”
Nonetheless, most operators (71%) declare to be dealing with these complaints nicely utilizing present methods. Even so, future assurance methods might want to align extra intently with data-centric providers — together with 5G standalone, community slicing, and rising AI-driven workloads — to satisfy rising expectations for low latency, constant throughput, and real-time efficiency ensures.
