People who share my background in traditional requirements analysis often have a hard time coming up with Product Backlog Items representing thin vertical slices of potentially-shippable product. Here is some material I've started using in teaching this.

GUIDELINES

User Stories may assume template form (see User Stories Applied by Mike Cohn):

or expressed as noun phrase:

or a 1-2 sentence story:

User stories, like all Product Backlog Items, should contain or clearly imply acceptance criteria (definition of “done”). Write these on the back of the index card, or in the "Description" field if you're using ScrumWorks. (By the way, if you're still using an information refrigerator like Microsoft Excel for your product backlog, maybe you haven't heard ScrumWorks Basic is still free!)

Bill Wake has given us the INVEST mnemonic to help remember the characteristics of a well-formed user story:

• I - Independent
• N - Negotiable
• V - Valuable
• E - Estimable
• S - Small
• T - Testable

EXAMPLES

What does this look like in practice?

1. A bank customer can change his PIN.
   • Acceptance Criteria: ....
2. As a student, I can find my grades online so that I don’t have to wait until the next day to know whether I passed.
   • Acceptance Criteria: ....
3. One level of undo
   • Acceptance Criteria: ....
4. As a book shopper, I can read reviews of a selected book to help me decide whether to buy it.
   • Acceptance Criteria: ....
5. As an author, I want the spell checker to ignore words with numbers so that only truly misspelled words are indicated.
   • Acceptance Criteria: ....

COUNTEREXAMPLES

How do we know when we're missing the mark? Ultimately "User stories are a promise for a conversation" (Ron Jeffries). If Product Owner and Team both know what they mean, you're off to a good start. But you can probably save some time by avoiding the common pitfalls below:

1. "Design brochure layout."
   • Drawbacks: not Independent, no business Value. This is a task representing a horizontal architectural layer or phase. The architecture will be done in a vacuum, possibly contributing to analysis paralysis.
   • Better: “As a dog owner, I can find a meal schedule on the brochure so I know whether this doggy day care center is appropriate for my hungry dog.”
     • This will lead to only the necessary amount of design to support this Sprint’s features. The layout might change the next Sprint, but rework is cheaper than no work.
2. "Write game rules."
   • Drawbacks: not Independent, no business Value, not Small.
   • Better: “As a newbie game player, I want to know who goes first so we can start the game.”
   • Better: “As a competitive gamer, I want a way to leapfrog my opposing players.”
3. "I want the brochure to be colorful."
   • Drawbacks: not Independent, not Estimable (without knowing other features of brochure), not Small.
     • This is an easy trap for those of us who grew up with the habit of writing “the JFIDM _shall_ comply with the IEEE-488 interface specification.”
   • Better: Use “colorful” and other cross-cutting requirements as acceptance criteria on each of the specific features in the backlog they apply to.
4. "As Product Owner, I want a list of highly-rated restaurants on the brochure."
   • Drawbacks: It’s not only about you!
   • Better: Focus on your end users and stakeholders. “As a gourmet tourist, I want a list of highly-rated restaurants on the brochure.”
   • Better: “As the Chicago Public Health Department, I want warnings about restaurants that serve raw ingredients so that tourists don’t get sick on our dime.”
5. "Play test the game."
   • Drawbacks: Not Independent. Encourages phasewise development.
   • Better: Make testing, refactoring, etc. a default acceptance criteria on every Product Backlog Item.
     • But: If you failed to fully test and refactor in previous Sprints, you are in technical debt! You are already working on a legacy product. In this case you may need to make testing and refactoring first-class Product Backlog Items to make up for your sins.

User stories are simple thing. "Simple" is not always the same as "easy." Happy unlearning!

--mj
Michael James
Software Process Mentor