How much do you actually review monthly/weekly? What is the ideal sample size to have a good overview of the quality level in the team? I'm struggling to decide on the good number of percentage. Especially that in my team I'll have agents who are very productive and resolve a lot of interactions (chat) and others who have additional projects and automatically resolve less cases. Then reviewing the same number for both of them seems unfair.

Hey @Sylwia Miela104!

Based on my own experience in doing it + the experience of hundreds of companies that I've talked to about this, I would say that "the more, the better" but less than 5% of volume (total and per agent) probably won't give you a representative sample. 

It's of course a question of what resources you have at your disposal, how many agents you have and how many tickets they solve. If you do a million conversations, then a smaller number of well selected examples could also be more than enough because the absolute numbers would be already very high. 

But let's say you have 10 000 tickets a week and those are done by 20 agents. That's not far from the actual numbers that I was dealing with as a head of customer support. In that case, I would set a target of 500 reviews/week and would then distribute more or less proportionally between the 20 agents while making sure that everybody gets some reviews. So if an agent does 1000 tickets, the number would be a bit higher but still aim for 5% of their volume. 

I would also set some kind of minimum target per agent because reviewing 2 tickets per agent will neither help reviewers to gain an adequate understanding of the performance nor help the agent improve. So maybe a minimum of 20 reviews/per agent/week? This is all of course assuming that you have the power to do that many - if not, maybe do it more as project and less often. Once every two months or something like that. 

All in all, my recommendation would be to think deeply about a) what resources you have, b) what you would like to achieve with the review process and c) what would make the smallest representative sample per agent (10 emails, 10 chats and 2 phone call reviews for example - if that happens to be how the proportions look like). 

Good luck!