Quality Monitoring – Sampling Methods

When Call Center Quality Teams conduct Transaction monitoring, to understand ‘Health of the process’, it’s rarely possible to monitor 100% of population. Instead, they select a sample. To draw valid conclusions from such monitoring, one has to carefully decide how to select a sample that is representative of the population as a whole. TransMonQa has seen multiple sample selection methods that organizations use & broadly we can say there are two types of categorizations that we can make.

1. Probability sampling that involves random selection, allowing you to make strong statistical inferences about the population.
2. Non-probability sampling that involves non-random selection based on convenience or other criteria, allowing you to easily collect actionable insights. This type of sample has a higher risk of sampling bias. That means the inferences you can make about the population are weaker than with probability samples, and your conclusions may be more limited. In interest of reader’s time, we will not talk about these methods in this blog.

Let’s talk about Probability sampling methods in a bit detail now. Probability sampling means that every transaction managed by Customer Service Representative has a chance of being selected. There are four main types of probability sample.

• Simple Random Sample
• Systematic Sample
• Stratified Sample
• Cluster Sample

Simple random sampling In a simple random sample, every transaction managed by Customer Service Representative has an equal chance of being selected. To conduct this type of sampling, organizations use tools like random number generators or other techniques that are based entirely on chance. Systematic sampling Each Customer Service Representative is listed with a number, but instead of randomly generating transactions, you first pick CSRs at regular intervals and then randomly pick a transaction of those CSRs. NOTE: make sure that there is no hidden pattern in the Customer Service Representative list, else you might end up introducing a bias inadvertently. For example, let’s say you usually group your employee database basis tenure, and team leaders. In this case you might end up skipping all less tenured CSRs (assuming that you start picking from the top in systematic order) Stratified sampling Stratified sampling involves dividing the total transactions handled by Call Center into subpopulations that may differ in important ways. (Example: by call type customer category type etc.) It allows organizations to draw precise conclusions because it ensures proper representation of every sub-group in the sample. Note: Based on the overall proportions of the population, organizations calculate how many transactions needs to be sampled from each sub-group. Once decided teams apply random or systemic sampling process to select a sample from each sub-group. Cluster sampling Cluster sampling also involves dividing the population into subgroups, but each subgroup should have similar characteristics to the whole sample. From TransMonQa’s experience, there is more risk of error in using this sampling method, as there could be substantial differences between clusters. Note: let’s say you have 3 outsourced partners managing Customer service calls for you. If you decide to pick one partner (cluster), it is difficult to guarantee that the sampled partner is really representative Website: www.transmonqa.com Contact us: hello@wyzmindz.com