March 8, 200917 yr Hi All..I'm working on a project for Quality Improvement in a BPO...and I found that the Quality evaluation form used doesn't give me consistent results..in other words the form is difficult to comprehend. one error gets marked under several parameters, which doesn't give me a clear picture and thus the scores are misleading....For eg. There is a parameter "Ownership"..which can encompass anything and everything a CSR should do...The solution I thought to break down Ownership like parameters into specific actionables. Which throws me a new challenge of having a quality form which is very extensive....Please let me now if there is any a way to engineer a form...(currently I'm going by the historic data ie. VOC).....
March 12, 200917 yr Hi Sreejit,Can you try some sort of an Attribute Measurement Systems Analysis (MSA) to determine if people are correctly interpretting the questions?If your scores on MSA (Kappa values or percentage agreement) are low, you could redesign the questions and try it again until people are correctly interpretting the survey form.SJ>PS: This is an advanced topic that should be posted in the private benchmark six sigma forum.
March 12, 200917 yr Hello Sree, The most important thing you must start with, while re-designing is to understand the basic parameters affecting your CSAT... If you have an extrenal CSAT measure - An IVR or a third party assessment, the parameters should be callibrated. 1. The cascading effect is an old thing which such forms - for example, how do you define 'ownership' - following the troubleshooter on the knowledge base, or being able to go a step further in deciding that the troubleshooter followed is insufficient and the query needs to be escalated or the resolution itself? 2. Again, the knowledge level of your advisor is an extremely subjective thing here - if it is a tenured advisor you're dealing with , are you expecting enough of him? The answer is not to breakdown such parameters. The answer is to make this a 'rating' rather than a binary marking - Yes, No. , i.e. , if you find it difficult to narrow down on the root cause of the 'ownership not taken' - skill/will? The key is to define ratings - specifically aligned to behavior... (am sure a couple of callib sessions among QAs should get you the idea!)0 - Solution available but not rendered (basics)1 - Incomplete information (could be same as accuracy of info)2- Troubleshooter followed and necessary steps taken.3- Issue resolved though beyond troubleshooter/explained well 4 - Ensured avoidance of recurrence.Dunno if you guys have a 1-5 scale, but a zero marking will ensure the agent doesn't even get 'grace marks'...If ownership in your contact center is defined as 'resolution' (Wow!), and you also have different types of resolution (Case raised to correct department, Case closed, Follow up call made/ closed, Service Order raised etc) - you could break it down accordingly. It'll make sense not to go by VOC, because they are very generic... however, if you have been following an IVR script which captures Resolution rates, and you have a further drill down of how many times the customer has called back, you could take that as a further support to the rating... But, of course, the process is more complex - to pull out such info for every call you monitor... :|Hope this helps... Am not sure if I have been able to write legible... Best of luck!
Create an account or sign in to comment