In Adobe and Incisiv’s latest research, telecommunications respondents report that an average of 16 out of every 100 proposed AI initiatives reach production, compared with the global average of 15. The largest numerical drop comes at IT evaluation. Integration has the steepest proportional drop: 49 proposals enter that stage and 31 pass, at 37% reduction.
These numbers don’t explain why proposals drop out. But they raise a question worth asking before the next pilot: Which requirements could teams address while the idea is still taking shape?
Take a provider planning an AI assistant for billing queries. The proposal needs to account for how the assistant will access billing records, verify the customer’s identity, and hand unresolved queries to an agent. Without those requirements, a strong demo still leaves key deployment questions open. Bring the people responsible for those decisions into the work before the pilot’s scope is fixed.
Clarify who can authorize wider deployment.
An AI initiative that spans enterprise functions like customer care, technology, and network operations needs someone who can set priorities across all of them. Only 15% of respondents report having a formal AI owner with a defined role and budget authority. The rest rely on part-time owners (32%), committees (23%), or ad hoc arrangements (21%), and 8% say no one is accountable at all. Whatever ownership model a provider uses, it needs to be clear who can commit resources and resolve decisions across functions.
The adoption figures raise another question. In the report, 60% of telecommunications respondents are Piloting or Deploying, while 20% are Expanding or Transforming. This concentration is what the report calls the Frozen Middle. It tells us where respondents are today, not whether they have stopped progressing or why. For a provider considering wider deployment, ask who can authorize expansion and what evidence is needed before doing so.
Bring business and IT into the work earlier.
Sequential handoffs can make integration questions harder to resolve early. Among telecommunications respondents, 76% describe their business-IT relationship as siloed, transactional, or consultative. These models support requests, delivery, and advice, but they stop short of shared accountability. Consultation can surface constraints, but teams still need agreement on who resolves them and owns the outcome.
Agree on the evidence before deployment.
Only 11% of telecommunications respondents measure AI returns comprehensively. Most rely on partial metrics that confirm whether a model ran or an interaction was completed, but not whether an initiative improved retention, service resolution, customer experience, network performance, or cost. Baselines are hard to reconstruct once an offer, interaction, workflow, or network decision has changed, and pricing and service conditions have moved with it. Without a pre-deployment baseline, separating AI’s contribution from pricing, service, and operational changes becomes harder.
What providers can learn from cross-industry Leaders.
The research identifies 342 Leaders across industries, representing approximately 6% of the 5,323 active respondents within the wider survey of 5,633 executives. They aren’t a telecommunications-specific group. Compared with telecommunications respondents, they score 60% higher on organizational structure and 46% higher on workflow transformation.
The report highlights five practices more common among Leaders:
- Real ownership. Leaders are six times more likely to have a formal AI owner with a title and budget authority. That mandate should also carry decision rights across the customer, commercial, technology, network, and oversight functions.
- Day-one measurement. Nearly half of Leaders (44%) measure AI returns comprehensively, compared with 11% of telecommunications respondents. They define outcomes, baselines, and attribution methods before an initiative moves forward.
- Genuine business-IT partnership. Collaborative or integrated relationships are the norm for 80% of Leaders, compared with 24% of telecommunications respondents. Both sides shape objectives, data needs, and controls from the start.
- Learning culture. Destructive failure patterns of blame and repetition affect 37% of telecommunications respondents but fewer than 1% of Leaders. Constructive review preserves the lesson without weakening accountability.
- Empowered champions. When asked to name their single most impactful culture move, 42% of Leaders point to finding and empowering internal champions, ranking practitioner-led adoption ahead of executive sponsorship and formal training.
None of these figures prove that ownership or measurement drives deployment maturity. These comparisons give providers operating practices to examine. They don’t establish what caused greater maturity.
Start with one priority initiative.
Start with the billing assistant described above. Name who can approve its expansion to other query types. Before implementation, record the current resolution and repeat-contact rates, define the improvement the pilot should demonstrate, and agree on accuracy requirements and when the assistant must hand a query to an agent.
Bring the relevant delivery and oversight teams into the work while the scope is still open. They can identify integration and governance requirements before implementation. Formal reviews should still determine whether the proposal is ready to advance.
Read the full report to examine telecom’s approval funnel and compare ownership, measurement and collaboration with cross-industry Leaders.
Recommended for you
https://business.adobe.com/fragments/resources/cards/thank-you-collections/telecommunications