Baseline Your Estimates Using Reference User Stories
By using reference user stories, teams can calibrate their story point estimates to real work they’ve completed in the past and confidently arrive at accurate-enough estimates more quickly. To do so, try the following:
- Pick 10 good user stories your team has completed in the past. Be sure the selected stories (a) span the typical range of story sizes the team most commonly works with (helps calibrate estimates to story size) and (b) represents stories on parts of the system or technology the team is most likely to work on in the future (helps calibrate estimates to story type).
- One-by-one, have the team estimate each user story using the standard fibonacci sequence scale of 1, 2, 3, 5, and 8 (discard any user story larger than an 8). Be sure the smallest user story is anchored as a 1 and the largest user story is anchored as an 8. All other user stories should fall in between.
- Discuss differences of opinion and re-estimate until the team arrives at group consensus for each of the user stories. Try to have at least 1 story for each of the fibonacci sequence scale values.
- Use these reference user stories in future estimation sessions by asking if the user story in question is less than, equal to, or larger than each of your reference user stories. This will make it much easier to hone in on an accurate estimate more quickly.
- Re-baseline your team’s estimates periodically — either once per quarter or whenever the nature of the team’s work changes in any significant way (ex. — the team begins working with new technology or on completely different parts of the system with legacy code, etc). In your Sprint Retrospective ceremony, ask yourself “do my reference user stories make sense to compare to upcoming work?”. If the answer is no, re-calibrate!
Finally, try developing a set of rules for when to change an estimate based on factors such as familiarity with the technology or system, uncertainty of design approach, amount of collaboration with other teams, etc. Using a rule set in parallel with reference user stories is a great way to provide a sanity check on estimates to ensure their accuracy based on other, more soft factors that influence how long it takes to complete a story.