Hi Everyone

Been reading on some of the topics and it has been renewing.

We have been doing performance based account for a year now. Now for the problem at hand.

Our coaches (people previously performing very well in sales before sales) monitor live calls of the agents who have not been performing well for 2-3 days, they are tasked to identify the problem of the agent and provide an action plan for that problem. I'm sure everyone would agree that the easiest to identify would be quality related problems, like pronunciation, accent, etc and rebuttals as these already have a standard/script to follow. But those things do not entirely cause failure, I talked to my team and we came up of things I find hard to teach to our coaches (to identify): intro spiel and tone of voice. Why do I find it hard? We compared agents using the same intro spiel verbatimly and different tones of their voices. Some were successful and some weren't.

So now the even more frustrating problem that has risen was that the coaches now just focused QA related problems or on just very obvious mistakes like wrong rebuttal, too many fillers, agent not following instructions...leaving me with unsatifactory evaluations and coaching sessions and slow improvement of performance or basically reliance on agent to acquire the skill by themselves.

if only I could edit: sales = promotion :(

Alexander,

Do you measure variation for agents and across agents? I am worried that what might seem like poorer performance from some agents will be within common cause variation. And if you are taking action on common cause variation - perhaps by doing some of the things you describe - you will be making things worse.

I only suggest this because you seem frustrated that what is happening is not improving performance, which if the variation was from common cause would be what I would expect to see. You can't blame the agent for what are common causes, you have to change the system.

If you are having trouble with what I am saying, send me the data in a time series for two agents, one with no problems and one with problems, preferably going back as far as you can for each. I will take a look and we can put up the results for the rest of CCV so they can see what I am talking about. Of course, make the data anonymous before sending.

Best,

Rob