Quality assurance has long been a necessary part of contact center management, helping leaders to monitor service, meet compliance obligations, identify coaching needs and understand where the customer experience is breaking down.
Yet the traditional model has a clear limitation, as human assessors can only review so much.
“Given it is a very manual process, the sampling in a busy contact center is really minimal. You can maybe do three or four calls for one agent per month,” Stephen Butler-Emmett, CX and AI Specialist at QContact, told CX Today.
This leaves quality teams making decisions from a small sample of the overall customer picture. It also means agents’ positive work can go unseen.
“It’s quite irritating when you do 400 really good calls and you get picked out for one that’s sampled by chance, and doesn’t take into account all the other good work as well.”
For organizations managing increasing volumes across voice and digital channels, this is becoming harder to sustain.
The Growing QA Burden in an Omnichannel Environment
The emphasis on omnichannel engagement has changed the nature of quality assurance. Contact centers now need to assess interactions across calls, email, chat, messaging and self-service journeys.
Digital volume has increased as businesses encourage customers towards self-service, bots and automated flows.
“QA teams are really struggling with digital because it’s much more difficult to go through a digital interaction than listen to a voice call,” Butler-Emmett said. “It’s quite mundane to go through a transcription.”
Digital interactions also present a practical challenge. A single customer journey may involve numerous messages across different channels and times, making it difficult for assessors to review consistently and understand the full context. This creates a widening gap between the complete view quality teams need and what manual processes can realistically provide.
AI-assisted QA can analyze voice calls, transcribe conversations, and examine digital interactions, giving quality teams a larger body of evidence from which to identify patterns.
“Where AI has helped and supported in the quality assurance space is it gives a much wider penetration,” Butler-Emmett said. “It can sample voice calls, it can transcribe them, it can go through digital interactions and give you a much more nuanced answer around what your customers are saying and also how your contact center staff are working.”
AI can identify patterns such as a rise in handling time for a particular type of interaction but the insight still requires context to understand why certain outcomes are happening.
“If it does pick up a trend, for example a long handle time for a certain interaction, the manual assessor can then go in and see the reason for that,” Butler-Emmett said.
In this model, automated analysis provides scale and direction, while human assessors then investigate the underlying factors and decide the appropriate action.
Avoiding the Compliance Trap
The technology alone does not determine whether AI-supported QA succeeds. How organizations introduce and position it with agents matters just as much.
Quality assurance can be perceived as a punitive exercise, especially when agents feel it exists only to catch mistakes, a perception that can become stronger when the process is automated.
“A lot of people will see it as the traffic police trying to catch you out,” Butler-Emmett said.