Skip to main content
Question
Ola Weintraub
Head of CS Operations
Asked a question 8 months ago

Hi guys, I am thinking of switching my 5 point rating scale to a 3 point rating scale for quality scoring. We score about 1-2% of all of our conversations and use it to teach and motivate our agents. I am getting the sense that the 5 point scale, while accounting for all the finer points within a response, send a wishi-washi message to the agent. Do you guys have any best practices in this regard?

Where am I?

In Quality Tribe you can ask and answer questions and share your experience with others!

Lisa Hunt
Support Engineer at Klaus

Hi Ola,

We do have a brief overview of our scales to help you make a decision, which you can find here: https://www.klausapp.com/blog/customer-service-rating-scales/89

We also have some work planned to allow you to offer more nuance for your rating categories so that, for example, if you have one question that doesn't make sense using a rating scale rather than a binary choice you can use a different rating type for that one in particular (i.e. "did we answer the customer's question?" is really a yes or no answer so an answer on a scale of one to five doesn't really tell us much).

In terms of what works best for you and your team, here are some thoughts specifically from me as someone who has worked in CS for the last 5 years (and not from Klaus as a whole):

When you choose the rating scale, you should be aware of the error of central tendency. If you give a neutral option, as with the 3-point and 5-point scales, folks are more likely to choose it as their default answer. You might find that people giving the reviews don't want to give anyone a low rating but might also believe that no one ever deserves to get the highest possible rating unless they have done what could be considered truly outstanding work.

The error of central tendency might also affect you with the 4-point rating scale, with reviewers tending towards either the 2 or 3, rather than giving ratings of 1 or 4. 

One way to avoid this error is to use a binary scale so that a review is either good or bad but you might be part of a team that is reluctant to ever leave negative reviews for one another. This is certainly the case in support teams that I have been part of - everyone has been so empathetic towards one another that no-one ever received a bad review unless something had gone dramatically wrong. This rendered the results essentially meaningless.

The value of a rating scale that offers more variance, such as the 5 point scale is that it offers more nuance to help you understand the performance of your team but this is only true if everyone has the same understanding of what these values mean or, at least, you have an understanding of what they mean from your team members.

In that case, you might find that your team members, for example, only use the top three options on your five-point scale. But this still tells you something - you can adjust your thinking to consider that a neutral rating is actually negative. In a sense, you might still have a three-point scale but with results that you wouldn't see if you switched to an actual three-point scale.