iSixSigma

Is Auditing BOTS a Good Idea?

Six Sigma – iSixSigma Forums General Forums General Is Auditing BOTS a Good Idea?

This topic contains 3 replies, has 3 voices, and was last updated by  Joe W 2 months ago.

Viewing 4 posts - 1 through 4 (of 4 total)
  • Author
    Posts
  • #238087

    Karthik Dharmalingam
    Participant

    During one of business meeting today, we were discussing about solution through Automation. A process/product developed were tested during the UAT and reported 99% accuracy and it goes-live. However in the real time, when there is 1% error which crops up and if there is a Client escalation, Team goes back and do 100% inspection (to ensure we don’t get errors), which completely defies the purpose of automation . while it is a confidence issue (people related), I would like to know if auditing BOTS is a good idea? Any thoughts are welcome.

    0
    #238093

    Mike Carnell
    Participant

    @karthikdharmalingam Auditing is a waste of time and money if you have automation. If it is actually at 99% anything less than 100% audit (which isn’t an audit it is sorting). Anytime you are under about 10% defective you are pretty much wasting time with a sample.

    You should have never bought off on 99%. You are operating at a Cpk less than 1.0. That has been passed for about the past 20 years. I would get the robot people in and tell them you need at least a Cpk of 1.5.

    0
    #238107

    Karthik Dharmalingam
    Participant

    Thank you Mike.

    0
    #238656

    Joe W
    Participant

    Hi Karthik, deciding whether 99% is good enough or not can be difficult. I’m not familiar with the acronyms you’ve used, I’m guessing UAT is unit acceptance test and BOTS are SW bots? Anyway, some kind of automation (whether a robot or a SW bot the issue remains the same.) In hindsight, you seem to be saying that the 1% customer escalation rate initiates a technical support “damage control” activity- 100% inspection. I’ve heard it said that 100% inspection is 80% effective… whether true or not, it illustrates the point that service or product quality cannot be inspected in. It takes some effort, but you can assess the total cost of quality of the 1% customer escalation rate and it’s effect both on customer satisfication and the lost productivity resulting from efforts to contain a problem that may or may not be there. A key consideration is that SW defects don’t occur randomly, but are the result of specific combinations of inputs. The difficulty is in presenting enough variation in the inputs to determine whether the BOTS can be made to be robust to this variation and deliver the desired input. It’d be great to hear more about your application and I hope that some of this is close to the mark.

    0
Viewing 4 posts - 1 through 4 (of 4 total)

You must be logged in to reply to this topic.