Utility customer experience programs usually start in the wrong place: a new front end, a journey map, a satisfaction dashboard. Meanwhile the thing every customer actually touches twelve times a year, the bill, is produced by a customer information system nobody in the CX meeting wants to talk about. This guide takes the opposite route. It starts from the structural facts that make utility CX unlike any other industry, then works through the CIS capabilities that decide most of the experience, and ends with a sober look at where AI helps and where it reliably hurts.
Key Takeaways
- Utility CX hit a measured low in 2025: the J.D. Power residential electric study recorded satisfaction of 499 on a 1,000-point scale, the lowest across its residential utility studies, with rising costs the main driver.
- Customers cannot churn, so dissatisfaction flows to regulators instead. The CX budget case is cost-to-serve and rate-case credibility, not retention.
- The CIS is the CX backbone: billing accuracy, start/stop/transfer, payment arrangements, and every fact a portal or bot can show a customer.
- Affordability is now a CX workload: household energy arrears reached $23.0 billion by mid-2025 per NEADA, which makes payment-arrangement design a frontline experience, not a back-office one.
- Proactive beats reactive: high-bill alerts and outage notifications measurably cut inbound calls.
- AI belongs behind the agent and in front of routine tasks. It does not belong in front of a billing dispute.
“A utility’s CX budget is usually spent on the front door while the bill is treated as plumbing. The bill is the product. Fix the data behind it before you buy another engagement layer.”
Nikhil Jathar, founder of AvanSaber
Why Utility CX Is Structurally Different
A retailer with unhappy customers loses revenue. A utility with unhappy customers keeps every one of them and inherits their frustration through other channels: the contact center, the mayor’s office, and the public utility commission. Because the customer cannot leave, the feedback loop routes through the regulator, who scrutinizes service quality and customer complaints when the utility asks for a rate increase. That is the structural difference, and it means utility CX is partly a regulatory asset.
The current numbers show how much pressure that loop is under. The J.D. Power 2025 U.S. Electric Utility Residential Customer Satisfaction Study, based on 127,103 interviews across the 152 largest electric utility brands, put overall satisfaction at 499 on a 1,000-point scale, the lowest score recorded across J.D. Power’s residential utility studies. The same study found average monthly residential electric costs up 34% since 2020, reaching $189 for 2025. Utilities cannot CX their way out of a price increase, but the study’s own commentary points at what they can control: reliable service, easy access to information, and frequent updates on outages and restoration.
Affordability stress compounds it. The National Energy Assistance Directors Association reported that household energy arrears rose about 31%, from roughly $17.5 billion to $23.0 billion, by June 30, 2025, with about 21.5 million households behind on their energy bills. One in six residential customers being in arrears is not a collections statistic. It is a customer experience population, and how a utility treats it is increasingly what commissions judge.
The CIS Is the CX Backbone
Strip utility CX to the interactions that actually happen and almost all of them resolve to the customer information system. What SAP, Oracle, and Cayenta platforms actually cover is mapped in our overview of what a utility CIS includes; here is why it decides the experience.
Billing accuracy is the foundation. An accurate, explainable bill is the single largest CX deliverable a utility produces, and every estimated read, misapplied rate, or unexplained true-up manufactures a contact. Exception handling inside the CIS (catching a billing anomaly before the bill goes out, rather than after the customer calls) is invisible CX work, and it is where modern AI earns its keep, as covered in AI in utility billing.
Start, stop, and transfer is the first impression. Move-in and move-out is often the first and last interaction a customer has, and it is pure CIS workflow: identity, service point, deposit rules, final bill. A move-in that completes online in minutes sets the relationship’s tone; one that requires a phone call and a follow-up sets a different one.
Payment arrangements are frontline CX now. With arrears at the levels NEADA documents, the flexibility of the CIS’s payment-arrangement engine (installment plans, deferred balances, assistance-program integration, safe dunning rules) determines whether a struggling household experiences the utility as workable or hostile. The collections side of this, and where agentic AI is starting to help, is the subject of our piece on AI in utility collections and dunning.
The wider modernization story, why utilities are re-platforming these functions at all, is covered in customer experience as a CIS-led digital transformation.
Self-Service Portals: The Second Front Door
A portal or app is where most routine interactions should land: bill view, payment, usage history, outage reporting, and service moves. The catch is that a portal is only as good as its CIS integration. A portal that shows a three-day-old balance, or cannot set up a payment arrangement because that logic lives only on the agent’s screen, teaches customers to call instead, and every taught call is a permanent cost. J.D. Power’s commentary on the 2025 study makes the same point from the data: utilities that deliver streamlined billing, digital self-service, and timely outage alerts see satisfaction rise even in a cost-pressured environment.
The practical test for any portal project is coverage of the top ten call reasons. If a customer can see why the bill went up, fix a payment problem, and report an outage without calling, the portal is doing its job. The build-versus-integrate details are in our guide to customer portals in utility services.
Proactive Communication: Outages and Usage
The cheapest contact is the one the utility sends first. Two categories dominate.
Usage and bill alerts. Oracle’s Opower Proactive Alerts watch interval data for customers trending toward an unusually high bill and message them before it lands; Oracle’s documentation credits high bill alerts with a 9% reduction in high-bill calls plus 0.3% energy savings on top of the 1.5% from Home Energy Reports. The scale is no longer experimental: Oracle reported approximately 45 million North American households in Opower AI-driven programs as of April 2026. A bill shock predicted is a dispute avoided.
Outage communication. Customers tolerate outages far better than silence about them. The J.D. Power finding above lists outage updates and planned-restoration times among the top drivers utilities control. The prerequisite is unglamorous: the outage system and the CIS have to agree on who is affected, so the alert reaches the right households with a restoration estimate the utility can stand behind.
Where AI Helps and Where It Hurts
AI has a real place in utility CX, and a well-documented failure mode. The map of what is production-grade across the whole operation is in our pillar on where AI in utilities actually lands; this is the CX-specific cut.
Where it helps. Behind the agent: Oracle shipped AI call summarization and tagging for utility contact centers in May 2025, cutting after-call work so agents spend time on the conversation instead of the notes. In proactive alerts, as above. And in routine self-service: balance, due date, outage status, straightforward payment arrangements, where the rules are deterministic and the bot is just a faster interface.
Where it hurts. In front of a billing dispute. A Verint survey of 1,500 consumers found more than two-thirds had experienced a bad chatbot interaction, with failure to understand the issue as the top complaint, alongside self-service flows that never reach a live agent. A Five9 study of 4,000 consumers in October 2024 found 75% prefer a human for customer service, 56% are often frustrated by AI chatbots, and 48% do not trust information from them. A disputed bill is high-stakes, emotionally loaded, and account-specific, which is the exact profile bots handle worst. A customer who is already angry about a $400 bill and then loops through a bot that misunderstands the problem does not file that experience under technology; they file it under the utility’s name, sometimes in a commission complaint.
The design rule that follows: automate the status update, the outage map, and the payment plan setup, and hand the dispute to a person, quickly and with the full account history on screen. Escalation speed is a feature, not an admission of failure. How to build the agent-assist side of that is covered in AI in utility customer service.
“Never let a chatbot argue about money. The bot’s job is to notice the dispute early and get it to a human with the context attached. Every vendor demo shows the bot answering; ask instead to see it escalating.”
Nikhil Jathar, founder of AvanSaber
The CX Moments, Mapped to the CIS
| CX moment | CIS capability behind it | Common failure mode |
|---|---|---|
| First bill after move-in | Start/stop/transfer workflow, rate assignment | Wrong rate or deposit surprise on bill one |
| Bill higher than expected | Interval data analytics, proactive high-bill alerts | Customer discovers it first, calls angry |
| Payment trouble | Payment-arrangement engine, assistance integration | Rigid plans, dunning that ignores an active arrangement |
| Outage | OMS-CIS linkage for affected-customer lists | Silence, or alerts sent to the wrong households |
| Billing dispute | Exception history, adjustment audit trail | A chatbot in front of it |
| Routine account tasks | Portal integrated to live CIS data | Stale balances that teach customers to call |
What to Measure and What to Ask
Measure utility CX where it is actually created: first-bill accuracy for new accounts, exception rate per billing run, percentage of top call reasons fully closable in self-service, alert reach before high bills and during outages, arrangement completion rate for customers in arrears, and time-to-human on escalations. Those six numbers predict the J.D. Power score and the commission’s mood better than any survey the utility runs on itself.
And when a vendor pitches a CX layer, ask the CIS questions first: what data does it read, how fresh is it, what actions can it complete rather than merely display, and what happens at the moment of escalation. The platform context for those questions is in our CIS-led digital transformation guide and the vendor AI landscape in the SAP vs Oracle vs Cayenta comparison.
For utilities working through these CIS assessment questions, the AvanSaber team provides vendor-neutral advisory on CIS capability gaps, portal integration, and customer experience program design across SAP, Oracle, and Cayenta platforms.