Key Influencers not Weighing volume?

[blm001]

Hello,  

I am using the new "Key Influencer" visualization to analyze First Contact resolution.

The KI visual tells me that a certain ticket type category (call it Category A) is 14x more likely to give me a failure. The failure Rate is 96%. This I can confirm by looking at simple charts. The volume of tickets is 490 out of 16k.

However, it also tells me that other ticket categories (like Category X) are also key influencers even though I may only have 1 or 2 tickets fail out of the 16k tickets in the report. The failure rate for these tickets is 100%.

Basically it shows me that "When" Category is X "... then the likelihood of Tickets FCRed being No increases by " 10.94x.  Albeit true, it seems pointless since we're talking about 1 out of 16k tickets.

It does not seem it is accounting for the volume (weight) of tickets properly.

Is it me or a bug?

[justynalucznik]

Hi I would love to follow up with you and dig into this scenario a little bit more. We are running some significant tests (calculating p-values) to take volumes into consideraiton so I'd like to take a look and see what's happening. Would you mind reaching out to me at juluczni@microsoft.com? Thanks!