Upping the accuracy in PI planning


This is beyond my ken, so I seek the wisdom of the group.

I was part of a conversation about what to do if your PI planning session suffers from real lack of confidence in their ability to estimate upcoming work. What do you do to make a PI Big Room session work and give some reasonable indication of upcoming capacity?

Some of the ideas I was given by a colleague are as follows:

There are preconditions to make PI planning hum and the estimates useful.

Planning (about 2 weeks prior to the event) including;

  • Stable estimation teams. People need to know who they are working with so there isn’t any loss of time and get to work together
  • Firm features - If the features aren’t ready to be estimated (DoR) side of things then estimating turns from a reasonable guess to a shot in the dark
  • Strong facilitation lead by a facilitator that really knows how to run the room.
  • Understanding that once a capacity is reached, trying to stuff more on to the train is madness
  • Getting the right people in the room. If there is a domain expert that will be needed for an estimate or review, have them there
  • Visible dates for the event and a list of people that must be there no “ands”, “ifs” or “buts”

Also noted, was that PI Planning was like sprints, they in that they improve over time; that it isn’t about getting it all right, it is about getting better each time. That the facilitator should probably be tracking that.

Better to roughly right than exactly wrong


Let’s ask the resident SAFe gurus…

@ryan :wink:


@bradstokes SAFe recommends using relative estimation so that could be an approach. I am not a huge fan of that method but helps when just starting. Was this there first PI? I coach my ARTs to this mind set. to have features refined and ready for story breakdown 6 weeks before the next PI. then 2 weeks prior to the PI event have close to 80% of the stories created so the teams go into PI planning to do that. PLAN…

When dealing with Features or Capabilities yes its important to utilize something like a DoR. But its also important that the those features run through refinement sessions where its handled like a team sport to quickly get a common understanding of the feature and if there is any runway that needs to be built or risks and or constraints as well as any NFR’s. I find that a lot of the ART’s I have coached do Just in time refinement so always dealing with a fire drill. Very hard to work ahead.

When it comes to facilitation is there no RTE(release train engineer)? I agree with someone that is a strong facilitator but also needs to understand why we are doing PI planning to help keep focused on the value of PI planning and keeping in line with the framework to help the teams learn.

Capacity should be just that capacity. Not sure why SAFe would complicate that. Capacity planning should be handled as capacity planning. how do you plan for capacity for 2 weeks just now do it for 10 weeks (or however long the PI is) . Make time off, vacation, etc visible and then work back from there.

I Keep going back reading through your questions and the more i review it the more it seems like you need someone that understand SAFe such as an SPC that can run through PI planning in the way that SAFe has laid it out to build muscle memory. Does anyone have any SAFe experience and or training? Seems like you should kick off doing SAFe for Teams with the Entire ART prior to the planning event so that you get everyone to understand the goal and expected outcomes.