Every call, heard. Our partner, GoConnection, brought AI to every qualified Endesa call with Speech Analytics.
27,309 interactions analyzed by AI in 30 days. Every analyzed call is checked for customer sentiment, silence, talkover and compliance questions, so the quality team reviews all qualified calls, not a sample.
27,309 interactions
analyzed by AI in a 30-day period (OneContact Portal, October 2025).
4 checks per call
on every analyzed call: customer sentiment, silence, talkover and compliance.
100% of qualified calls
analyzed by AI, reviewed by the quality team. No more small manual samples.
Same day feedback
to agents, based on real conversation data instead of memory and samples.
GoConnection runs the outbound sales operation of Endesa on Nubitalk, the Collab cloud Contact Center. More than 200 concurrent agents work predictive campaigns, and every call is a potential energy contract with legal text the agent has to say correctly. Quality control used to cover a small manual sample. Today AI analyzes every qualified call. The quality team reviews and decides.
Speech Analytics (AI Studio) added to the Nubitalk platform GoConnection already used, with results feeding Data Studio dashboards. No migration, no new infrastructure.
Speech Analytics tenant started on a few campaigns with a fixed Azure budget. Filters by call outcome and duration, a first compliance question (does the agent present the Endesa service correctly?), answered by AI and confirmed by the evaluator, and live Data Studio dashboards replacing static reports.
Every call is a contract
The operation
GoConnection has managed Endesa’s outbound sales operation on Nubitalk from day one. More than 200 concurrent agents work predictive campaigns. Every call is a potential energy contract. Every contract has legal text the agent has to say correctly.
The challenge
Quality control was performed manually, allowing for the evaluation of only 4% of calls, leaving 96% of interactions without any analysis. This limitation had a direct impact on operations: sales errors and compliance issues could go unnoticed, opportunities for improvement were identified too late, and agents received vague feedback, making it difficult to improve practices and results. In practice, for every 1,000 calls made, only 40 were evaluated, compromising the ability to identify risks, improve sales conversion, and ensure consistent service quality. The result was an operation with high supervision costs, limited visibility into actual performance, and a limited capacity for preventive intervention.
Start small, prove value, expand
GoConnection added a Speech Analytics tenant to the Nubitalk platform it already used. No migration. No new infrastructure. Four steps.
Start with a limit
The analysis began on a few campaigns, with a fixed Azure budget. Costs stayed under control from day one.
Extract what matters
Filters by call outcome and duration focused the AI on the interactions that matter, instead of every second of every call.
Define compliance
The team wrote the first compliance question: does the agent present the Endesa service correctly? Every analyzed call now gets an AI answer to that question, with the evidence one click away. The evaluator confirms.
Categorize
Categories follow the words agents have to say and the words customers use when they accept or decline an offer.
Costs stayed under control from day one. Coverage grew once the value was proven.
How the quality team works today
Every analyzed call gets four checks: customer sentiment from the transcript, silence, talkover and compliance. The team filters those scores to find the calls that deserve a human ear.
-
Low customer sentiment Flags calls to review. Often a failed sale or weak argumentation.
-
High silence with failed compliance Often points to a training gap. The evaluator confirms.
-
Talkover became a KPI Timing metric only. Acceptable when closing a sale. A signal to review at any other moment.
-
Word trends Show which terms repeat across calls, so coaches can act on them.
When a call needs review, the AI summary, transcript and compliance answers are on one screen. The evaluator decides without replaying the audio. Transcript, sentiment chart and compliance result sit on the same screen.
Endesa works with the same interface. Quality patterns are now visible to the end customer across every call, not a sample.
Data Studio
From static reports to live dashboards
Speech Analytics results feed Data Studio, alongside telephony data and GoConnection’s internal database. Supervisors track campaigns, contact lists and agents in real time. Alerts fire when a list is running out or a close rate crosses the threshold. Time spent on manual reporting went down.
Results
27,309 interactions
analyzed in a 30-day period (OneContact Portal, October 2025)
A compliance check
on every analyzed call, with the evidence one click away for the evaluator
Faster reviews
Average review time per call reduced
Real-data coaching
Coaching is now based on real conversation data, not memory and samples
“Speech Analytics has transformed the way we manage quality at Go Connection. Today, we’re able to analyze a much higher volume of calls, identify patterns in conversations, review alerts, and quickly pinpoint interactions that require attention. Analyzing metrics such as customer sentiment, periods of silence, voice overlap, and compliance with requirements gives us a much more comprehensive view of each interaction. But what matters most is our ability to act faster: identifying opportunities for improvement, fostering empathy among agents, and providing feedback on the same day. We’ve gained efficiency, speed, and the ability to intervene, making our quality management much more comprehensive and focused on continuous improvement.”
AI for customer experience, built on 20+ years of Contact Center expertise
Collab delivers AI for customer experience, with 20+ years of Contact Center specialization as its foundation. Nubitalk runs the Contact Center in the cloud. OneContact CC runs on-premises or hybrid. AI Studio adds Virtual Agents, Agent Assist, Speech Analytics and WorkAgents. Part of Lumine Group.
AI with a Human Touch.