Why You Don’t Need to QA Score 100% of Calls with Auto QA Software
Example we are using for this:
50-seat contact center answering 60,000 a month.
Managing 60,000 calls with 50 agents can seem overwhelming, especially if you think you need to quality score every single call. But guess what? You don't have to! Thanks to the Law of Large Numbers, you can confidently score a smaller sample and still get accurate results. This is where Auto QA software and contact center QA software come into play.
What is the Law of Large Numbers?
The Law of Large Numbers tells us that as we sample more calls, our results get closer to the true average of all calls. This means you don't need to score every call to understand your overall quality. With QA software powered by AI, you can leverage this principle to save time and resources.
Understanding Confidence Level and Margin of Error
When we talk about sampling with Auto QA software, two important terms come up: confidence level and margin of error.
- Confidence Level: This is how sure we are that our sample represents the whole group. For example, a 95% confidence level means we're 95% sure our results are accurate.
- Margin of Error: This shows how much our sample results might differ from the true average. A 5% margin of error means our results could be 5% higher or lower than the true value.
Achieving Equal Representation with Auto QA Software
It's important to make sure the sample represents all agents and call types fairly. This helps you get a true picture of overall performance, which is crucial for effective contact center QA software.
How to Get Equal Representation:
- Divide Calls into Groups: Split the calls into different groups (like by agent or call type). Then, randomly pick calls from each group. This ensures each group is fairly represented.
- Proportional Sampling: Make sure the number of calls you pick from each group matches the group's size. For example, if Agent A handles 10% of the total calls, then 10% of the sampled calls should be from Agent A.
- Random Sampling: Within each group, pick the calls randomly. This reduces bias and ensures fairness.
Sample Sizes for Different Confidence Levels and Margins of Error
Here’s a quick guide on how many calls you need to score out of 60,000 to be confident in your results, depending on your desired confidence level and margin of error. Using QA software with AI can help automate this process efficiently.
90% Confidence Level:
- 1% Margin of Error: 6,531 calls
- 3% Margin of Error: 723 calls
- 5% Margin of Error: 270 calls
95% Confidence Level:
- 1% Margin of Error: 10,642 calls
- 3% Margin of Error: 1,187 calls
- 5% Margin of Error: 450 calls
99% Confidence Level:
- 1% Margin of Error: 16,613 calls
- 3% Margin of Error: 1,853 calls
- 5% Margin of Error: 664 calls
Using this example, I would use the 99% confidence level with a 3% margin of error, meaning you only need about 1800 calls a month or about 60 calls a day (30-day month).
To get this done with our OttoQa tool, you would need to budget only $1500 a month for your entire QA scoring.
I really think it's hard to justify full-seat licenses for QA software once you understand this.
Takeaway: Benefits of Using QA Software with AI
Using the Law of Large Numbers and understanding confidence levels and margins of error means you can save time and resources while still getting reliable insights into your call quality. So, don't stress about scoring every call—focus on the right sample size and ensure equal representation across agents and call types.
By using Auto QA software, you can streamline your quality assurance processes, reduce workload, and improve accuracy. Contact center QA software powered by AI can handle complex calculations and ensure that your sampling is both efficient and effective.