John Linton ....or an amazing aspect of the future of end user support.....take your pick.....and the majority, by a very long way vote for the former.
We began the main part of the SL reviews today with Annette beginning the one to one discussions with each one of the SL personnel and me starting the 'analysis' of what was happening in the overall SL operations. Anybody vaguely associated with being a part of the running of any commercial operation would understand how bad or good such processes can be. If you are aware of these processes then I need say no more and if you aren't then there is no amount of detail that can make it any clearer.
The most enjoyable part of the day was the hour or so where the AI program that Exetel is funding at SLIT was updated via the people undertaking the research. The progress is now tangible and we will begin to see how it can actually become integrated in to our, or anyone else's support and sales processes over the coming months. We had a brief demonstration of how the base engine is currently operating which is little more than a refined version of the pattern recognition that we 'bought' commercially some 5 -6 years ago but at least it is now 'our' code and the two levels beyond word/syllable pattern recognition will become available to this process.
Our five year objective, like our objective for the past 7 years, is to develop an intelligent front end to our telephone 'human' services that will filter out the need for human beings to be involved in something like 75% of the questions asked of any support/sales/provisioning facility. If this proves possible then the cost of providing support to any company's end users reduces by 75% from whatever it is without an AI function. We have been working on this assumption for almost ten years now and have not got very far but we now seem to be on the road to achieving this self evident goal.
It has always been evident that it required the intellectual rigour and sharpness of intellect that we had been unable to bring to it on our little better than part time basis we were able to devote to it. By funding the 'professorship and the associate researchers/programmers we have been able to see some real progress over the past six months. We will now engage Exetel customers in 'populating' the "unanswerable" question logs of the infant data base so that we can move beyond the current concepts of pattern recognition to the more sensible levels of 'intelligent' discussion.
If we can progress beyond this point we will begin the process of moving a text to text process to a voice to voice process with the objective of having a computer to human initial base answer service to all of our human based telephone support and sales functions by the end of 2011. Eliminating 50% of the repetitive, and dumb, questions before a human has to be bored by the drudgery of answering them will be a key factor in improving the experience of being a CSR in the call centres of the future.
Scary stuff. But more sensible than:
http://www.smh.com.au/executive-style/management/blogs/management-line/how-do-we-fix-bad-customer-service/20101003-162b3.html?rand=1286162057959
Copyright © Exetel Pty Ltd 2010
ABN 350 979 865 46