How many calls per agent per month do you analyse for quality purposes?

We are running a poll this week

"How many calls per agent per month do you analyse for quality purposes?"

I was expecting the average answer to come out in the region of 4-6 calls per month.

However the survey is showing a different answer.

The most common answer (36%) showed was that they are monitoring more than 10 calls per agent per month.

What do you find to be the best number of calls to monitor?

We are client lead in terms of the quantity monitored but would advise at least 10 calls per agent per month with more (possibly 3-4 times the set amount) for those agents with identified quality issues.

The more investment in back-end quality checking, the better the service provided however, as Sanman has already mentioned, this can depend on logistics and cost.

I always find it interesting that people talk about numbers of calls they monitor as if this is an indicator of how effective their QM programme is. To me it's another example of let's measure the thing that's easy to measure!

I've not yet come across anyone who can really give me any logic behind why they monitor the number of calls they do. The argument normally goes along the lines of "industry best practice". I reckon though that the industry doesn't really know what the correct amount of evaluations should be.

I'm not too sure about the question raised in the poll "how many calls per month do you analyse for quality purposes". This really depends on what you want out of your quality programme. If the purpose of a QM programme is to assess how well an organisation / contact centre is performing as a whole then a random sample of calls across all agents will give this information (assuming that what is actually being measured is correct; but that's a whole separate discussion). Also you don't actually need to assess that many calls to get this view as few as 400 calls can give you a +/- 5% accuracy based on accepted sampling theory.

However, once you get into the realms of individual agent quality and making assumptions of how good / bad they are and what they are doing that needs attention, typical sampling theory doesn't apply. You need to get much smarter and more targeted on what calls you are assessing and why. For example, as Ant Marketing have mentioned, the number of calls you need to monitor per agent should vary based on an agent's current performance. Mind you this is only one factor that will influence the volumes.

Sorry for the rant, but this has always been a question that gets me agitated! In my opinion a catch-all metric of "x number of calls per agent per month" is an irrelevant measure which at its worst can tie up expensive resource for little operational gain.