From Manual Mayhem to Scorecard Sorcery: How Automated Quality Scoring Can Save Contact Center Sanity!
Early in my career, I standardized and deployed a QA program across multiple contact center locations for a large global company. Our initial process was painfully manual. We listened to live calls, completed paper monitor forms, transcribed data into spreadsheets, and distributed monthly summaries by email to various teams. When we purchased a solution to record interactions, post online scorecards, and deliver real-time reports, our QA process dramatically changed for the better.
The rightful challenge from leadership was that of sample-size. From a statistical point of view, we were light years away from the number of evaluations needed to achieve a confidence level within an acceptable margin of error. At the time, very few automated scoring solutions were available, and the costs were prohibitive.
Fortunately, contact center agents are creatures of habit, who tend to repeat both good and bad behaviors, and the potential of being audited on any given interaction is a motivator. Also inspirational: a culture where correct behaviors are plainly explained and are consistently and positively reinforced. Even without an automatic scoring solution, we created a new culture, amplified the voice of the customer, improved compliance and set CSAT records.
Today, with advanced speech analytics tools within budgetary reach for many contact centers, monitor rates can be vastly improved for many key behaviors, even to 100% of interactions with surprising accuracy. Especially in high-risk or regulatory scenarios, the ROI for automated QA measurement is compelling.
A good automatic scoring solution can spot-on tell you whether an agent said or didn’t say certain words or phrases on every voice interaction. With channel-separation, it can tell who said what, and can often determine if certain statements were required or prohibited. Most recording transcription solutions today include the ability to perform ad-hoc searches across a set of recording transcripts for specific words or phrases, and many include some form of sentiment and customer level of effort analysis.
Without an ability to compare what is spoken against what the agent is viewing on their screen or steps actually taken, there are limitations to what speech analytics can measure. Transcript-analysis solutions cannot measure non-voice channels or processes, and automated QA scoring via voice transcription alone cannot and should not replace an entire human QA team.
If I were shopping for an automatic contact center quality scoring product today, here are four key features I would be looking for in a solution.
For accuracy and sample size reasons, speech-to-text vs. phonetic transcription, especially in a regulated industry.
Ability to pre-populate answers on manual scorecards, with auditors completing any questions not able to be scored by automation.
Quality analysts able to change automated answers on scorecards, and the ability to track scoring changes made by auditors to identify needed automatic scoring adjustments.
Unified reporting, combining all manual and automated quality data.
The adoption of automated scoring solutions in contact center quality assurance processes has transformed the way contact centers evaluate agent performance. While it’s important to acknowledge the limitations of speech analytics alone, a holistic approach that combines automated scoring with human expertise ensures a comprehensive evaluation of agent performance and adherence to quality standards. Ultimately, a unified reporting system that consolidates both manual and automated quality data provides a comprehensive overview for better decision-making and improvements.
Lorne Wood is an experienced contact center leader with a career in customer service dating back to the early 1990’s. As an industry colleague and CCNG member, Lorne shares his perspectives and experience developing and implementing strategies that ensure efficiency and drive extraordinary customer experiences.