PLANNING POKER
Entinex Style!

We've made up a set of our very own Planning Poker cards

Here's how it works:

(Adapted from Grenning 2002)
1. Discuss the story/feature/theme/requirement/etc.
2. Place estimate face-down.
3. Turn over all cards at once.
- No Convergence? Discuss High & Low then return to 1 (with timer as needed).
- Convergence? Note most common value and move to next story.

Helpful practices:
- No one's opinion trumps.
- Use the timer to keep the discussion focused.
- Anyone can use the timer.
- Avoid hair-splitting. (Remember the timer.)
- Decide on your own rules such as whether values are days or hours, how long a "day" is, what constitutes "convergence", and when to call for a new round of estimates.
- Note attributes used to rationalize/justify the estimates.
- Note risks associated with each estimate.
- Agree on how long breaks should be.
- Experiment with allowing each estimator to bid a "range".

These instructions are included in each deck of cards.
Our friend, Tom Loveland, suggested that perhaps it might be good under certain circumstances to collect all the cards and display them so that the values are anonymous thereby disarming any possible fears/intimidation that might exist in a project team that hasn't yet gelled.

In any case... A few other ideas:
» The "?" mark card and the "I need a break" card (the one with the beer, nuts, and pretzels) card may trump other cards and table a round of discussion for now.  The "?" mark card can eliminate all other estimates regardless of convergence, whereas the "I need a break" card can simply table discussion until after the break.

» Use an egg timer or stop-watch to keep on time.  Taking longer is not equal to adding value.  If convergence is not happening, consider that a clue that the requirement is too vague still, high risk, or requires other action and estimation at this time is *not* a good idea.

» Usually, the product owner, Scrum master, or whoever is facilitating the discussion on features is *not* one of the estimators.

For more information, please visit Mountain Goat Software, the folks who popularized the idea.

Each estimator gets a set of cards and each set includes the following cards:

• 0
• ½
• 1
• 2
• 3
• 5
• 8
• 13
• 20
• 40
• 100
• ?
• Break

For those who can't help but ask.... The cards follow a simplified Fibonacci series, and uses a simplified version of the Wideband Delphi estimation method.

To use Planning Poker as an estimation approach that helps with CMMI practice, Project Planning, we strongly encourage the capture and recording of attributes that contribute to the estimate, both leading to and upon convergence of the numbers.  We would also encourage capture of the attributes of the actual game as far as number of rounds before convergence for each product feature/task, number of features/tasks that were broken into smaller parts, and so forth.

When recorded, these attributes can be revisited and updated throught the projct and reflected upon during project/iteration retrospectives.

Planning Poker is a value/metric-rich, inclusive (and fun and easy!) approach to performing estimates.  Best of all, it likely most closely resembles how estimates are typically derived in absence of any other formulas, only now there's a good bit more structure without imposing useless and unrealistic activities.

CMMI doesn't say estimation can't be fun, which is why we created these cards!

Don't forget, if you *do* use planning poker for project estimation, your defined process can't just be "we use planning poker like Entinex says to" You'll want to record how each project actually plays the game.  Otherwise there are just too many unknowns for that approach to be realistic, manageable, or to yield repeatable results.

Please contact us if you have any questions about it!

You can get your own decks!

You can make your own decks of Planning Poker cards, but if the novelty of marking-up regular playing cards or using sticky-notes wears off, we'll be happy to ship you ours!

Each deck comes with 6 complete sets of cards, plus 1 instruction card. 

This means up to six people can be in on the estimation process at once. 
We picked 6 because it was within the recommended 7 +/- 2 value for number of estimators who should play at one time.  Also influencing our decision was our observation that it's rare that our clients (and other users of this approach) ever have even that many people working on estimation, and, even getting them as far as 6 participants would be beneficial in some cases.

Because we are not set up to sell products, you can order cards by contacting us.

Any donations to offset costs would be appreciated and negotiable.

CMMI, and SCAMPI are registered in the U.S. Patent and Trademark Office by Carnegie Mellon University.