Determining what constitutes a valid call monitoring sample is a matter that generates considerable debate. Different people will offer a variety of opposing answers to the question, but the only one answer actually is right, according to Craig Antonucci, VP Client Services at BPA Quality, and that is “as many as it takes to improve performance, while not exceeding the cost of quality.”
In a post on BPA Quality’s blog, Antonucci writes that many people believe in a statistically valid sample of calls for each agent. However, he says, “if you are looking for the typical 95/5 percent confidence level and interval, you simply need to know how many calls your agents take in a month’s time to derive a ‘statistically valid’ sample.” Antonucci notes that there is a website that can provide a clear number estimate of a valid statistical sample.
What this boils down to is that when determining the number of monitored calls to include in the sample, you need to calculate the average total number of calls an agent handles within a specified time frame. “For instance, if your agents take 1000 calls a month, you’d need to sample 278 calls in a month’s time for it to be significant for that specific month,” Antonucci writes. “Time is the key variable in determining this…over what time period is it relevant to you.”
Still, it is possible to come up with a number of calls that would require too much time and effort on your part. “Frankly, 278 calls being monitored is overkill and the benefit you’d get out of it would be outweighed by the cost of doing so. You can augment this with tools such as Speech Analytics, however, while SA is a great tool, it cannot be a replacement for human auditing, coaching and feedback,” Antonucci notes in the blog.
That brings up the question of how BPA Quality determines the best sample size for its clients.
“When we recommend a sample to a client, we recommend no less than 2 monitors per week per agent. The idea is to create an environment of coaching and feedback,” Antonucci explains. “You want the agents to know that they will be getting feedback and have that present in their minds as they take calls…you want them to perform. If you only do one call, and they are coached early in the week, they will figure out quickly that they aren’t being monitored for the rest of the week.”
Still, a larger sample size usually beats a smaller one in best assessing accuracy. “We also tell our clients that ideally, more is better, and a daily review and feedback is as close to ideal as you can get without breaking the bank,” Antonucci writes. “The statistics don’t necessarily support it, however, the human psychology does. Agents will perform better, if they are being monitored. The Hawthorn Effect demonstrates this in human behavior and psychology.”
In the end, Antonucci says that call centers must decide what level of auditing works best for their personnel and operations. Remember the ultimate goal, Antonucci says: “Quality Monitoring should be used as a tool to make agents better at servicing your customers.”