send survey after every 10 closed incident?

[ash36]

I need to send incident survey on repeated interval of 10 incident closure. I tried setting the trigger condition based on Number ends with 0, but that's not a concrete solution. Has anyone done this before?

Help much appreciated, thanks !

[The SN Nerd]

Let me frame it for you this way:

You could spend hours developing a custom script that then needs to be tested, documented and maintained over time.

The end result will be that every 100 incidents closed, 10 surveys will be sent.

OR

You could spend 15 minutes setting up a Survey Trigger Condition for 10% probability, using the condition 'State' is Resolved.

That is, 1 in 10 closed incidents will have a survey sent (in theory).

The functionality is already documented and well tested by the vendor.

The end result will be that every 100 incidents closed, ~10 surveys will be sent.

[Rakesh Mamidi]

Setting 10% probability is OOB and its good.

I have seen a corner case in my past like where in say 30 tickets are closed and so 3 survey's must be sent.

The Survey might be sent at 10th,20th and 30th or it can be sent for 12th,19th,27th incidents.Gap interval is not consistent because it triggers random.

In above case, if we take first 10 incident closure there is no survey sent.

In my case QA and business would like to have for every 10 or 15 incidents.

[The SN Nerd]

I guess if they are willing to accept the extra cost(s) associated with that solution then it's their call.

All you can do is present the options, costs and risks.

[drjohnchun]

In data sampling, what Ashish is using is called systematic sampling while what Paul is suggesting is called random sampling. Whether you're selecting every 10th record or 10% at random, you'll end up with about the same sample size given it's sufficiently large (determining the proper sample size is yet another topic). But do they provide the same or equivalent results? If the sampling is designed properly, the results should be equivalent within the statistical error.

Unless there's an overriding reason, my preference is to go with what's already built into SN, which, in this case, is random sampling. However, Ashish may have the requirement he has to fulfill by using systematic sampling.

BTW, for systematic sampling, you may also be able to use a custom System Property as an incrementer to keep track of the number of closed tickets and send out a survey when it hits 10, 20, etc. Seems straight forward.

I've written a blog series on Data Sampling Techniques and get into some concepts I mentioned here in Part 3:

Data Sampling Techniques Part 3: CMDB Data Quality

Hope this helps.

Please feel free to connect, follow, mark helpful / answer, like, endorse.

John Chun, PhD PMP

Winner of November 2016 Members' Choice Award

[ash36]

Thanks John for the deep insight. However, when you say "for systematic sampling, you may also be able to use a custom System Property as an incrementer to keep track of the number of closed tickets and send out a survey when it hits 10, 20, etc.", can you give any example how to configure this for OOB incident surveys?

Thanks,

Ashish