CMT CX leaders, move past your cost and fix your debt to unlock customer value

This HFS Take 5 report is for CMT customer experience, IT, and operations leaders scaling agentic AI from pilots to journey-wide customer value.

Executive summary

Communications, media, and telecom (CMT) enterprises have been heavily investing in AI for customer experience (CX), yet the outcomes have plateaued. Generic models cannot reason over the customer, contract, or workflow specifics, so value stalls on the growth side even as cost-to-serve improves. The bottleneck is operational, with integration complexity and data fragmentation as key constraints. The demand for implementing agentic AI in CX is real, but the application and data layer needs work before agents can act seamlessly across silos.

To realize the full value of AI for CX, enterprises must address the integration and data debt blocking journey-wide AI and shift the business case from cost to customer value. That means connecting fragmented systems and customer data so agents can act across the journey, grounding the work in governance and CX domain depth.

HFS Research, in partnership with Cognizant and Google Cloud, surveyed 51 CMT enterprise leaders across North America ($1 billion+ in revenue) to understand where agentic CX stands today and what it takes to move from pilots to scaled value.

The survey uncovered five key takeaways:

The Bottom Line: Agentic CX represents CMT executives with a credible path to closing CMT’s growth-side CX gap. Capturing value requires partners with proven capabilities in integration, governance, and CX domain expertise. This also means that enterprises should shift their lens from cost to customer value.

This is more than double the share of those who already scaled. Most are looking to make the leap from GenAI-augmented AI to agentic AI or fully autonomous systems.

• Most CMT executives expect to jump two maturity tiers in 24 months. This requires a step-change in the operating model, not business as usual.
• The Level 3-to-Level 4 leap is where value compounds. Moving from copilots that suggest the next action to agents is where CX economics change. It is also where integration and data debts should be addressed.
• CMT leaders are planning a bigger maturity leap than most have made to date. The winners will be those who fix integration and data now.
• Agentic CX delivers a Level 4 agent that handles a billing dispute end to end, including pulling the contract, issuing the credit, and updating the customer. Getting there is an integration project, not a single-system deployment.

Most CX use cases sit at 20% scaled or below. Cross-system workflows such as billing (14%) and authentication (15%) lag the most.

• Across every use case, most deployments sit in piloting or siloed production. Under 20% have scaled enterprise-wide across most use cases, well short of the 76% that plan to reach agentic CX in two years.
• Single-domain use cases will scale first; cross-function ones are harder to crack. For example, a retention agent that knows the customer is in a billing dispute during a service outage is the kind of cross-silo move that protects retention. Today, that is the exception, not the default.

Less than 40% of CMT leaders are satisfied with AI outcomes in any CX domain, as horizontal copilots and bolt-on tools fail to encode the workflows, rules, and customer data that decision-grade CX requires.

• CMT leaders are pouring investments into AI for CX. However, satisfaction with outcomes is capped well below ambition because generic models cannot reason over the specifics of a customer, a contract, or a workflow.
• The gap is widest where context matters most. Personalization and cross-channel integration depend on customer signals, history, and downstream system state, precisely what horizontal tools cannot see.
• Large neutral blocks indicate that pilots are running while outcomes are not yet good enough to defend a scale-up budget. It is a quiet form of failure that delays the harder conversation about the underlying systems.
• A copilot that summarizes a support ticket without knowing the customer’s contract tier, prior escalations, or open invoices cannot drive resolution. Closing the gap requires CX-specific workflows, expert rules, and integrated customer data.

Nearly half of the CMT leaders surveyed cite integration complexity (45%), data fragmentation (41%), and governance (41%) as the top blockers.

• Integration and data are key blockers. The leading barriers are all plumbing problems: connecting systems, unifying data, and encoding rules.
• Governance is now table stakes, not a brake. Blanket policies create more friction than they remove. Scaling requires governance controls built into the workflow itself.
• Demand is not the problem. Business case and change management sit at the bottom of the list. Leaders know what they want agentic AI to do, but they cannot get the underlying systems to deliver it.
• An AI agent that recommends a retention offer without seeing the customer’s billing disputes, tenure, or product entitlements is solving the wrong problem. Unblocking value requires CX-grade integration, trusted data, and built-in controls, not another model.

Buyers rank integration, governance, and domain expertise above agentic AI-platform flexibility and pre-built agents.

• Enterprises prefer partners whose agentic CX work is defined by disciplined delivery and assurance, not by who has the best model.
• Governance is now a procurement criterion. Buyers want guardrails built into agentic platforms, not bolted on after deployment.
• Pre-built agents come last. Sector-specific workflows differ enough that customization beats configuration.
• Deep partnerships matter. Vendors will need to align their frameworks and partnerships with what their customers have installed to be successful.
• A partner that integrates the agent into the customer data and existing systems, demonstrates the governance trail and proves that the workflow in the buyer’s sub-segment will outsell one with a richer model menu.

The Bottom Line: Agentic CX represents CMT executives with a credible path to closing CMT’s growth-side CX gap. Capturing value requires partners with proven capabilities in integration, governance, and CX domain expertise. This also means that enterprises should shift their lens from cost to customer value.

CMT CX, IT, and operations leaders must take ownership of operationalizing agentic CX as a journey-wide capability, not a contact-center upgrade. The 18-month window for differentiated capability is open now.

Five moves CMT leaders must take to convert ambition into scaled value:

1. Fix integration and data foundations: Fragmented data and disconnected systems are the binding constraint; no model compensates.
2. Stop piloting and start scaling: Use the next 12 months to scale what already works.
3. Treat journey-wide use cases as integration projects, not AI projects: Connect customer data and workflows before deploying the agent.
4. Reframe the business case around CLV, not contact-center economics: Cost-out and CSAT are necessary but no longer differentiating.
5. Select partners that work with the systems you already have: Integration, governance, and CX and industry domain expertise matter more than model choice.

The largest unrealized value in CMT agentic CX is the re-orientation from cost-first to CLV-led investment thesis.

Appendix

Survey demographics

AI has delivered efficiency, the next frontier is lifetime value

AI in CX programs at CMT companies are incomplete: efficiency is done, but growth and experience haven’t caught up. The next dollar pays off in retention, cross-sell, and customer perception, not further productivity gains.

• AI in CX has matured along the path of least resistance, with a focus on productivity and stalled elsewhere.
• Cost-to-serve and first-contact resolution are shaped by volume, routing, and containment, which are exactly what large language models, classification, and retrieval do well.
• AI falls short on the KPIs that depend on context it does not yet hold:
  • Experience requires reading tone, history, and intent across a relationship, not a single ticket.
  • Retention plays out in the moments before churn, which sit outside the contact center.
  • Cross-sell needs an agent built to expand the relationship. Most agents are built to close the ticket.
  • The result is an incomplete program with AI industrializing the cost side of CX, while growth and experience wait for a capability that understands the customer, not just the CX silo.

CLV is the undervalued prize; enterprises are still funding agentic CX on cost-out

Cost reduction sits at the top of the priority list. However, executives need to be wary of a singular focus on cost as it may inadvertently impact growth side metrics.

• The priority list is a wake-up call. CMT leaders want better retention, higher CSAT, and growth-side outcomes. However, they are funding agentic CX primarily through a cost-reduction lens. You cannot buy your way to CLV with a cost-out budget.
• CLV uplift is the underplayed frontier. The one outcome that signals genuine lifetime-value thinking sits at the bottom of the list, while every other outcome (cost, satisfaction, productivity, retention) clusters in the same band.
• An agent that resolves a billing issue prevents a churn moment and surfaces a relevant upsell in one interaction as the CLV unit of value. Today, that outcome lives in three priority buckets and three different teams.