Why does "version report" projects a release date much farther in time than expected?

[Andrea Caponio]

I am currently tracking a version in JIRA. 

The current status is that 28 story points have yet to be completed. Team average velocity should be around 29 story points per sprint, and each sprint has a duration of 2 weeks. I would expect a projected end date of ~2 weeks, but I see that the version reports indicate the versions to be ready in ~2 months from now.

Am I missing something? Is the version report taking into account something I do not see (e.g. the rate with which new issues are created in the same version)?

[Andrea Caponio · Accepted solution]

To have a consistent prediction in the Version Report it is important to:

1. Make sure that the Version start date is correct and corresponds to when the work on the version actually started (and not to when the work was planned).
2. Make sure that working days of the team are correctly configured (e.g. are days off of the whole team properly taken into account?)

Other bits and pieces may influence. This article should help figuring out if still something is missing.

[Jack Brickey]

Without seeing the details e.g. screenshot, I can only suggest the reason is that the current velocity of getting work done. There is the theoretical value and the predicted value. Theoretical = sum of estimates and predicted considers velocity. 

[Andrea Caponio]

I would assume that the chart is based only on story points and only stories have story points estimation: ech sub-tasks should not be considered. 

Maybe I did not understand what you exactly mean, but I do not really see how the theoretical should be 4 times slower than the mean velocity.

[Andrea Caponio]

Here are the screenshots.

[Jack Brickey]

@Andrea Caponio, I agree on the surface that the prediction shouldn't be 4x. I wish I could offer more advice but w/o seeing the details I can't say for sure. I haven't used the Version report for a couple years now and recall when I first started it was confusing for me but ultimately was able to decipher the deltas between the various prediction models. I'm guessing you have already read this article but sharing just in case it helps.

Hopefully others that are actively using this report can provide some better assistance.

[Andrea Caponio]

@Jack Brickey I already found the first issue: the stats are calculated since the start date of the version, therefore is important to start it on a meaningful date. Adjusting this already anticipated the prediction to 1 month less.

[Jack Brickey]

Precisely! I was just looking at your charts a bit ago and was thinking that might be a contributing factor. Good find!

[Andrea Caponio]

This should help a lot: https://community.alfresco.com/community/ecm/blog/2015/07/29/inside-the-jira-version-report

[Jack Brickey]

indeed! great article. Once you have everything figured out you should either accept this answer OR maybe better construct a new succinct answer and accept it as this will help other in the future.

[Ross Yeager]

@Jack Brickey @Andrea Caponio , while it makes sense that you can adjust the start date, it does not make sense to me why JIRA is just doing a very simple linear fit to the data. Our team's velocity has increased as more people have started ramping up on the release, and this is not reflected at all in the predicted dates, making it pretty useless.

See screenshot. It's pretty easy to eyeball and know that we'll likely be completing right around the completion date, however, we wont get any indication in the prediction until about a week before completion. Seems like they should be doing a mixture of moving averages or non-linear fit to more accurately predict. The fit currently there literally just interpolates a line from the start date to the current date completion count...making it pretty useless.

[Mr_ Gray]

Where you able to solve this (I consider it a problem, because it is not accurate).