Velocity: What if a story is partially completed in one sprint?

Question: If you had a user story (Story X) with four Story Points at the start of the Sprint, which was not completed by the end of the Sprint but the majority of the work was done (e.g., just two small bugs remained), would you recalculate the Story Points when adding this into the next Sprint – for example effort was now only one Story Point to complete the story? How would this have an effect on Velocity planning?

Answer: This first “answer” is not quite to your question, but first, we don’t play horseshoes. (Do you know that game?) There is no partial credit in Sprint N for doing part of Story X, and, if everything is completed in Sprint N+1 (according to the Definition of Done), then the team earns full credit in Sprint N+1 (four story points).

But, you say, that is unfair!

Well, in a way. So, yes, to get a fair Velocity estimate for Sprint N+2, you must use the average over the last X Sprints (people argue how large X should be, let’s say three or four Sprints).

If the team says: “Well, if we had divided that into two 2 SP stories, and we would have earned 2 more SPs of Velocity in Sprint N…”, then you say, “Yes, but you didn’t.”

This gives the team a big incentive to have smaller stories (and virtually 110 percent of the time, smaller stories would make everything better).

Why? Well, it does seem like making the stories smaller takes more time to slice them, to put Story Points on, etc., etc. There is some truth to that, but by making the stories smaller, it turns out that the feedback (and other factors) are much, much better. So, net net, there is a savings of effort if we are smart, and use the feedback (and other benefits) professionally. It is better to have smaller stories. (Well, one can imagine 40 stories in a two-week Sprint. So, not too small.)

Remember what Yogi Berra said: “It ain’t over till it’s over.”

Meaning, for us: We never know if our work is really complete until it fully matches the Definition of Done. Sometimes they say there are only two small bugs remaining, but only after they are fixed and re-tested do we really know the bugs are small. This is just so true for our kind of work. You probably see it every day on your team. (Give them the specific examples that have happened to them.)

_________________________

Next issue: How much work should the team take on in Sprint N+1?

Answer: Let us assume that they can predict with reasonable accuracy that Story X has only one SP worth of work “remaining.” Let’s assume that the average Velocity is 20 and that Story X will be done in Sprint N+1.  Then, for Sprint N+1, the team will “use” one SP for Story X and then have 19 additional SPs of “capacity” for other (new) stories. Still, they will end up earning 23 SPs (assuming they get everything done this time). That is 19 SPs for new work, and four SPs for Story X.

Again, because we average over several sprints, these “normal” ups and downs come out in the wash, as we say.

Help enough?

 

« « Is It Really important to estimate initial velocity? || Impediment List – Inhouse Course May 15 » »