I was recently asked to lay down the things I think about when estimating stories for development. Here's what I came up with. Given that a lot of this is second nature, I'm sure I've missed something in my attempt to get it all out of my head. ConstructiveCriticism is strongly encouraged.

Design

• Will this story fit into the current design of the code?
  • best case scenario, generally lowers estimate
• Will I need to refactor how we currently do things in order to accommodate this story?
  • probably the most common case
  • refactoring can sometimes add more time to the estimate, but not always
  • refactoring should always be done in context
• Will I need to remove and/or rewrite significant portions of the code in order to accommodate this story?
  • should be rare, but these do happen on occasion
  • generally adds significant time to story, but not always (e.g. removing code can be quite quick and not add a lot of time)

Testing

• What will the tests/specs look like for this story?
  • each category (developer, functional, integration, feature, other?) of tests/specs adds time and complexity -- but well worth it
  • not all stories require a feature test/spec per se (e.g. developer and functional tests/specs may be sufficient)
• Will this story affect current tests/specs?
  • usually adds time

Research

• What new research do I need to do for this story?
  • variable affect on story, but for the purposes of estimating it is assumed that this will add more time
• Are there libraries/frameworks that will help with this story?
  • usually decreases estimate
• Would this story benefit from a quick proof of concept (PoC) first?
  • adds time
  • usually separate from the story and written to include time (e.g. "take a day to investigate Foo" or "take a few hours to build a PoC to validate our approach")
  • sometimes these stories are written after work on the initial story has begun, in which case the initial story is paused in favor of the PoC and usually re-estimated based on findings
• Do I have everything in place already and thus the need for new research is mitigated?
  • adds no time

Documentation

• What sort of documentation will I need to reference in order to finish this story?
  • Is it readily available? adds little to no time
  • Do I need to find it? adds time
• What sort of documentation will I need to write for this story?
  • usually adds some time
  • this can be a significant investment of time (e.g. documenting an API for external customers)
  • or not much time at all (e.g. just documenting an algorithm by copying and pasting some other resource)

Deployment

• Will this story affect the current deployment scheme?
  • adds time
• Will this story have no effect on the current deployment scheme?
  • freebie, obviously