Mervi Sepprei

‘What Percentage of Conversations Should You Evaluate?’ is one of the most asked questions in the QA space.

On average, only 1-2% of support conversations are reviewed, placing a premium on selecting the right ones.

The goal for QA reviews is often to obtain a representative sample. With the overall average review coverage of just 1-2% and diversity in languages, channels (voice, email, chat, social media), markets, product areas, and handling tiers, achieving a truly representative sample manually is unrealistic. Just creating the filters to address all areas is a serious data mining task.

The Role of AutoQA

AutoQA (automated quality assurance) steps in here, handling 100% of support tickets and eliminating biases associated with manual reviews. It evaluates key metrics like tone, grammar, empathy, greetings, closings, readability, and comprehension. Although it lacks the process-specific insights of human reviewers, it excels in general human communication aspects and offers a bias-free, quantified quality metric.

Balancing Automation and Human Touch

AutoQA identifies areas requiring focus, where it outperforms humans, such as grammar checks. However, human QA specialists are still vital for providing nuanced feedback and detailed analysis. Combining both approaches ensures comprehensive coverage, consistency, scalability, and reduced human error.

AutoQA provides an efficient way to obtain a representative overview of support conversations. Strategic use of both automated and manual reviews can significantly enhance your customer support team's quality and performance.

What percentage of conversations do you think your team should review?
Let me know in the comments.