We already know that we can correct and sustain a particular behavior by providing positive feedback. So we observe the behaviors, and when we identify the "correct" one, we make sure to positive feedback.
I was curious to see if a machine shares this feedback, will it have the same effect as it would have shared by a person. What do you think, which was the conclusion?
Disclaimer: I haven't run the experiment for an extremely long period and haven't incorporated any statistical best practices.
Behavior to sustain:
We have a particular type of JIRA tickets in our organization for which we have an agreed SLA. In general, it's acceptable to have 75% of those respecting the SLA. However, on average,~80% of those tickets are within the SLA, and I've wanted to see if we can increase the percentage.
Automation Setup:
The trigger setup would be a CRON job that runs each working day and checks that the status of the ticket was changed earlier than 2days (In reality, the JQL was more complicated, but I've simplified it)
project=X and AND Status BEFORE -2d and assignee in membersOf("devs")If the condition is true, then it will post a comment on the ticket tagging the assigned person:
Hey [~accountid:{{assignee.accountId}}]! Most probably, you haven't noticed, but you were able to reply to this ticket within SLA (max 2 days). Great job! !https://media.giphy.com/media/kyiHXDANuOg8Jy6ByA/giphy.gif!
Conclusion:
The team had some fun with these comments. However, the percentage of tickets respecting the SLA was still around the average, so there was no real improvement. Still, as I've said in the disclaimer, the experiment didn't consider statistics best practices.