AI Powered Quality Assurance (QA) and Agent Reviews

Post author: 
Est. Reading: 3 minutes
Edgetier Blog Images

The Quality Assurance Process

To keep track of, and improve, customer experience at the contact centre, the management of contact quality, consistency, and outcomes is critical. The QA (Quality Assurance) process is designed to ensure that agents are dealing with customers in line with the operating guidelines and brand tone of the company, and is used as part of training and continuous improvement efforts in contact centres.

QA is most often carried out through a regular assessment of randomly selected customer interactions for each agent. This review process is carried out monthly by a team supervisor or dedicated QA team, and interactions are scored across multiple metrics that capture adherence to protocol, friendliness, efficiency, etc. There has been a move away from “speed” metrics such as handling times to manage agent performance, with contact centres now focussing more on customer satisfaction and contact quality.

According to this 2017 study from NICE, QA reviews operate, most often, with monthly reviews of approximately 7 interactions per agent (on the order of 1% of interactions).

The greatest challenge for comprehensive QA review is a shortage of time and resources. While performance and quality is high on contact centre’s priority list, the process is manual, time consuming, and requires highly trained staff. Listening to 100% of calls, or reading 100% of chat transcripts is clearly an impossible task, and random selection of interactions for review can easily miss important information or points for improvement. Very often, results are recorded in Excel or Google Sheets in smaller centres for analysis, but digital scorecards with centralised reporting are becoming increasingly common.

Using Artificial intelligence for QA

Artificial intelligence (AI) and machine learning can vastly improve the coverage of analysis efforts for QA processes at the contact centre. Often falling under the umbrella term of “interaction analytics” or “conversational analytics”, AI systems can analyse every contact, providing contact centre administrators a much higher level of visibility across agent activities.

Many of the technologies mentioned in this series, including text classification, sentiment analysis, and other NLP techniques can be used to help with QA processes and applied across both text-based communications and phone conversations (using speech-to-text transcriptions).

AI systems can help the QA process in multiple ways, for example:

  • Identifying interactions where agents did or did not follow protocol
  • Automatically tagging interactions for review that contain difficult or at-risk customer behaviour
  • Identifying average customer sentiment per agent, allowing more focussed review of individual interactions
  • Analysis of grammatical errors and spelling mistakes by agent
  • Analysis of brand tone and banned statements by agents when speaking with customers
  • Allowing comparison between agents in a unified view to isolate the “key areas for improvement” for individual agents to improve the efficacy of the monthly review process
  • For calls, detection of emotion or silence during interactions to find inefficiencies in process flows
  • Detecting repetition in transcripts, which is often not encouraged in centres, or may indicate a bad connection on phone lines.

As real-time and post-contact interaction analysis tools become widely employed, the QA process is poised to become more efficient and more personalised for each agent. Ultimately, these improvements will allow 100% coverage of interactions in a cost-effective manner, driving increases in customer satisfaction as well as overall efficiency.

Expectations from AI-enhanced QA processes

Implementing an AI system in your agent review process will not remove the requirement for detailed analysis from team leaders, or for 1:1 meetings between agents. However, a well-implemented system can help review meetings become much more effective and customised to individual agent's and team performance.

Ideal Outcomes

  • Improved agent quality through more targeted and specific QA reviews
  • Increases in customer satisfaction scores due to improvements in agent quality
  • Lower QA process costs due to improvements in efficiency
  • Improvements in consistency across customer communications
  • Faster response to difficult customer issues or agents operating outside of training

Potential Pitfalls

  • Staff performing reviews will need education on how best to use analytics outputs for agent review
  • Agents will need reassurance that these types of analytics are not surveillance
  • Inconsistent application of automated QA systems across different teams and channels
Discover More
Let us know your details and we will contact you with more information regarding EdgeTier and our Platform.
Main-Contact-From (#5)
linkedin facebook pinterest youtube rss twitter instagram facebook-blank rss-blank linkedin-blank pinterest youtube twitter instagram