Archive for January, 2011

January 12, 2011

Estimation using Planning Poker

by Sampath Prahalad

Now imagine that each team member is holding a deck of cards, containing the following cards:

Image001

Let’s do the estimation. The product owner says:

Image002

Before we continue, a small note on the numbers.

1: Small UI enhancement or bug fix. Easy to develop, easy to test.

2: Adding a small change/enhancement to existing functionality

3: A slightly bigger change to existing functionality or a fairly simple module to develop and test

5: A feature or enhancement that touches few other modules, We understand it well. It is complex. Testing impact is significant.

8: A feature or enhancement. We have some understanding, Some investigation is needed, it is quite complex and will touch few modules. Very significant testing impact.

13: Pretty complex. Might need more than a Sprint to accomplish by 1 person. Better to split into smaller User stories.

20: I don’t even know what you are talking about. Very complex, Needs good amount of investigation. Let us revisit it once the PM provides more clarity. Cannot estimate at this point of time.

The team starts thinking about how long the story will take.

Image003

All participants have to present a card, face down, containing their estimate. Everybody has to present a card, so Mr D and E wake up. Mr D admits that he was sleeping and asks what the story is about. It’s harder to slack off when doing estimates this way :o)

When they are done, all cards are turned over simultaneously, revealing everyone’s estimates.

Image004

Whoops! Big divergence here. The team, in particular Mr A and Mrs C, need to discuss this story and why their estimates are so wildly different. After some discussion, Mr A realizes that he has forgotten some important tasks that need to be included in the story. Mrs C realizes that, with the design that Mr A presented, the story might be smaller than 20.

After the discussion (3 minutes in total) they do another estimation round for that same story.

Image005

Convergence! OK, not complete convergence. But they agree that an estimate of 5 should be close enough. Next story.

%d bloggers like this: