Projects This Week

This week I am trying to finish up three outstanding projects:

  1. A review of the issues around one of our major products – I am involved with a cross-department work team that will use this data as guidance in June.
  2. A review of transfer rates from our call center back to the client.
  3. A review of level two assistance calls – These are calls where agents are routed to more senior agents to ask for assistance on the tasks performed on a call.

All of this is kind of a crunch since I will be in Orlando next week for the Verint users conference.

DNIS Problems

Does anyone know how to use multiple DNIS’ in a singe category? Right now I am creating a new category for each DNIS…but it would save a lot of time and effort if I could attach multiple DNIS’ to a singe category. HELP!

What are you working on now?

I am just finishing up a study that examines our level of system preparedness for a new line of business we will start to service later this year. I created a category that captured the call every time a verbal cue was used that showed the agent was waiting on their system to finish processing. Next, I compared this to a lay out of our traffic patterns for the same period. I used these two sources of data to show how slow our systems would get under different levels of traffic, i.e. 200 calls per hour, 1000 calls per hour, 2000 per hour, etc. This fulfilled the first requirement of this project, to perform a systems stress test. So now that we have shown a connection…the question becomes “How much does this cost us?”. To establish this I examined a random group of 100 calls that had video, and showed up in the category for system slowdowns, and timed the transactions. I compared this group to a second group that did not have any visible system issues, and contrasted the two sample groups. This will establish the negative impact that system issues cause. Finally, I will forward the results of this study to our IT and Resource Planning divisions so they can use it to prepare for this influx of new customers.

There are some concerns that I have about this project…mainly, those agents that use language that indicates the system is processing as a normal part of their lexicon on calls. I am banking on the fact that this group would be evenly represented in my sample, and would impact the time per transaction if there was no real system slowdown issues. What do you all think? Is this adequate? Or, what are you working on?

What do you use SA for?

So far, when I talk to other users of Speech Analytics software it seems that there are two camps:

  • Those that use it for enforcement of their various correctional policies, i.e. did agent X cold transfer a caller after being educated about not doing so?
  • Voice of the customer analysis – Why are they calling? How can I improve their experience?

Which camp do you belong to? Or, are there other camps? Do you feel you are getting a good ROI on your product?

Hello world!

Welcome to Speech Analysis! My name is Rob Bradshaw, and I am employed by a Pharmacy Benefit Manager (PBM) in the Omaha, NE area. In my job I am responsible for using Speech Analytics to improve the quality of our interactions with our members. I envision this blog becoming a place where best practices and new ideas in the use of Speech Analytics can be suggested and discussed.