CASE STUDY: National Grid

National Grid is one of the largest investor-owned energy companies in the world, delivering electricity and gas to millions of customers across New York and Massachusetts. Its 2,500-agent contact center handles some of the most consequential customer interactions in the utility sector, including billing disputes, payment hardship, and service reconnections during heating season.After four years on a previous platform procured through a two-tier reseller chain, the organization moved its quality, analytics, workforce, and engagement operations onto NiCE Workforce Management (IEX) with a direct vendor relationship. The team eliminated $600,000 in annual third-party QA vendor cost, shifted supervisors from evaluation quotas to behavior-based coaching, and gave 2,500 agents mobile self-service for the first time. Coaching consistency improved by 20 percent, supervisor efficiency by 30 percent, and repeat call volume dropped 15 percent.

01 Before

A vendor chain, a quota model, and a system nobody liked.

National Grid is one of the largest investor-owned energy companies in the world, delivering electricity and gas to millions of customers across New York and Massachusetts. Its contact center operation, handling approximately 2,500 agents, sits at the front line of some of the most consequential customer interactions in the utility sector: billing disputes, payment hardship, service disconnections, and emergency reconnections. When a customer calls in November because their heat has been cut off, getting it right is not a service level metric. It is a matter of basic welfare.

The contact center technology supporting those agents was not getting it right. National Grid had been running its recording, workforce management, and quality operations through Calabrio, procured via a two-tier reseller chain: the organization contracted with a vendor, who had contracted with another vendor, who had contracted with Calabrio. When something broke, the path to resolution ran through every layer. Resolution took weeks. As Stefan Singh, Product Manager for Customer Experience and Digital at National Grid, described the experience: “By the time I get a resolution, I forgot about the problem. I moved on to something else.”

Layered on top of the vendor friction was a quality model measuring the wrong thing. Supervisors completed a set number of evaluations per agent per month. They hit the number. They checked the box. Whether those evaluations were driving behavior change was impossible to know. Customer satisfaction was not improving. A third-party QA vendor was being paid $600,000 per year to produce numbers, not outcomes.

Agents, meanwhile, had no self-service. No mobile access to their schedules. No visibility into their own performance data. No ability to request time off without going through a process. The contact center’s workforce skewed young, recent graduates and early-career workers who expected to manage their working lives from their phones. That expectation was not being met.

02 Desire to change

One platform. One relationship. One vendor to call.

Three failures had become impossible to ignore. A support structure that took weeks to resolve basic issues. A quality program generating activity rather than improvement. A workforce technology stack invisible to the agents who depended on it.

The consolidation vision was straightforward. NiCE Workforce Management (IEX) offered everything that was desired on a single platform, with a direct vendor relationship replacing the reseller chain. The technology was available. The question was whether the organization had the appetite to invest in it. Leadership did have some trepidation about cost. The counter-argument was direct: if you want best-of-breed, you pay for it. End-user dissatisfaction with the incumbent system provided the supporting evidence. Buy-in was not the obstacle.

The more significant shift was conceptual. Fred Daum, Vice President of Customer Contact at National Grid, had been sitting on a correlation analysis that connected AI-detected agent behaviors to after-call customer satisfaction scores. The data showed that when agents demonstrated the right behaviors, customer satisfaction increased. The implication was clear: the evaluation quota model was not just inefficient. It was actively measuring the wrong thing. What mattered was not how many evaluations a supervisor completed. What mattered was whether agents were greeting customers properly, demonstrating empathy, and taking ownership of the call.

Moving to behavior-based coaching required a platform that could surface those behaviors automatically, at scale, across every interaction. Interaction Analytics could do exactly that. No custom configuration was needed. The behaviors were available out of the box.

The tools were there. The data was there. The only thing missing was the decision to use them. For National Grid, the vendor model itself was part of the case: going direct to NiCE meant one relationship, one point of contact, and one place to go when something needed fixing. After four years of working through a reseller chain, that simplicity was its own argument.

03 NiCE solution

From contract to first recorded call.
One platform across 2,500 agents.

National Grid signed its contract with NiCE in January 2024 and recorded its first call in November of that year. Quality Management, Interaction Analytics, and NiCE Workforce Management (IEX) came online together. Desktop Analytics followed in the months after. Full Desktop Analytics deployment is expected by the end of June 2026.

The quality transformation: from quota to behavior

The foundation of the new quality model was Interaction Analytics. Every call is now transcribed in full. The AI engine scans each transcript against a defined set of agent behaviors: appropriate greeting, customer empathy, and ownership of the call. Each behavior is scored automatically. The system surfaces the result: this agent is consistently demonstrating empathy; this agent is struggling with taking ownership. Supervisors receive a data-anchored starting point for every coaching conversation they have. When supervisors do listen to calls, it is for targeted coaching from those AI-driven behavioral insights, not to score. Calibrations have been eliminated entirely.

The shift from quota to behavior changed what supervisors do with their time. Previously, a supervisor completing their monthly evaluation requirement was a supervisor who had listened to a handful of calls, submitted scores, and moved on. Now, with the evaluation burden removed, that same supervisor has a dashboard showing behavior trends across their team, can identify which agents need which conversations, and can track whether coaching is working over subsequent weeks. Coaching frequency increased from approximately 30 minutes per agent per month to 40 to 50 minutes. The content of those conversations changed completely.

The vendor consolidation: $600K out, accountability in

The $600,000 per year third-party QA vendor was eliminated entirely. The work that vendor was doing, call evaluation and quality scoring, is now done by the AI and by National Grid’s own supervisors. Bringing quality in-house also changed accountability: it is harder to point a finger at a third party when the work is yours. Supervisors own their agents’ performance in a way they did not before.

The agent ownership shift: mobile, self-service, in their hands

Agent self-service was a parallel transformation. The NiCE Workforce Management (IEX) mobile app gave agents access to their schedules, time-off requests, overtime management, and their own quality scores from their phones. The previous platform had no mobile capability at all. For a contact center workforce that skews young, this was not a convenience. It was a generational expectation finally being met. Agents could take ownership of their performance without being tethered to a desk or dependent on a supervisor to tell them how they were doing.

One of the most significant operational changes addressed the PCI-compliant payment call. A customer calls in November, heat cut off, needing to pay a balance to get service restored. Previously, the agent had to transfer them to a third-party payment processor and drop the call entirely: recording had to stop to protect payment card data, which meant the customer had to hang up, pay alone, call back with a confirmation number, and wait again. NiCE solved it with an automated stop-recording trigger. When the agent dials the payment processor, recording pauses. When the customer returns, it resumes. The agent stays on the line. Service is restored in one call.

Ten minutes of training. One instruction: stay on the phone.
For a customer sitting in the cold, that matters.

Without any configuration or work needed from National Grid, the behaviors were available right out of the box. Supervisors don’t need to listen to calls to score them now. When they do listen, it’s for targeted coaching.

Stefan Singh

Product Manager,
Customer Experience and Digital,
National Grid

04 Results

Fewer calls back. More coaching. One less vendor.

The $600,000 annual QA vendor cost has been eliminated in full. The third-party vendor is gone. The AI and National Grid’s own supervisors now handle all quality management.

Supervisor efficiency improved by approximately 30 percent. Time recovered from manual call-listening and quota tracking has been redirected to actual coaching. Those sessions are behavior-anchored, data-supported, and trackable over time in ways the old quota model never allowed. Repeat call volume dropped by close to 15 percent. Interaction Analytics surfaced two primary drivers. First: agent compliance issues, where agents were incorrectly directing customers to call back when they did not need to. Second: agent education gaps, where incomplete product knowledge was generating follow-up contacts. The elimination of the PCI payment callback was a significant contributor, removing an entire category of mandatory callbacks from the operation.

Agent self-service adoption has exceeded 60 percent of the workforce, reached progressively over the 18 months since the November 2024 go-live. Agents are managing their schedules, reviewing their own quality scores, and requesting time off via the NiCE Workforce Management (IEX) mobile app. They do not need to be in the office to do any of it.

We were just measuring the wrong thing. Doing four, doing six, doing ten evaluations has no measurable impact on customer experience if the model isn’t driving behavior change.

Stefan Singh

Product Manager,
Customer Experience and Digital,
National Grid

The analytical foundation for the coaching framework is already proven. Daum conducted a correlation analysis between post-call SMS survey scores and Interaction Analytics behavior scores, confirming that agents hitting defined behavior benchmarks drives higher customer satisfaction.

That analysis is being presented at NiCE World, with broader CSAT impact data to follow as the behavior program matures.

The shift has been felt by the people doing the work.
A quality manager with more than 30 years of industry experience described how the day-to-day reality of the role has changed:

In my day-to-day quality work, I used NiCE Workforce Management (IEX) to gather all the details and asked AI to analyze and create the customer journey info, which let me focus on understanding patterns and outcomes rather than compiling data.”

Quality Manager,

National Grid Contact Centers

The measure National Grid trusts most requires no spreadsheet. For the entire four years on Calabrio, end users came to the product team with the same complaint: in NiCE, we could do this. Those complaints became known internally as the death words. Since the return to NiCE, they have not been heard once.

Silence, in this case, is the best result of all.

“If my users aren’t coming to me and telling me the old system did it this way, I’m happy,” Singh said. “I have not heard those death words. I never hear that anymore.”

05 Future

AI on the roadmap.
The foundation already in place.

Desktop Analytics deployment completes by the end of June 2026, closing out the original consolidation scope. What comes next is an expansion of AI capability across the platform. National Grid has been evaluating NiCE Copilot for Agents, NiCE Copilot for Supervisors, Actions, Interaction Insights, and Automated Quality. These are on the roadmap for the next couple of fiscal years.

The foundation for that expansion is already in place. A daily relationship with the NiCE Technical Account Manager means new capabilities can be assessed and implemented without the weeks-long lag that characterized the previous support structure. The platform is stable. The team knows how to use it. The business case for going further is built on results that are already proven.

The behavior-CSAT correlation analysis Daum is presenting at NiCE World is not just a conference session. It is the internal proof point that behavior-based coaching works. That data will drive the next phase of coaching program development, the expansion of AI-assisted quality, and the case for additional product investment.

Agents who once had no visibility into their own performance now own it entirely. The roadmap ahead builds from there.

Go direct to NiCE. It might be a little bit more money, but the stability and the support you get is second to none. One throat to choke. Something happens, I’m only choking one person.”

Stefan Singh

Product Manager,
Customer Experience and Digital,
National Grid