Fibonacci vs T-Shirt Sizing - Which is Better?
Our team is debating between using Fibonacci numbers (1, 2, 3, 5, 8, 13) vs T-shirt sizes (XS, S, M, L, XL) for story point estimation. We've been using Fibonacci for 6 months but some developers find it confusing. Which approach has worked better for your teams and why?
I've used both extensively. Fibonacci is superior for mature teams because the gaps force better conversations. When someone says '5' vs '8', they have to articulate why. With T-shirt sizes, people drift toward 'Medium' for everything.
That said, T-shirt sizing is excellent for initial roadmap planning with stakeholders who aren't technical. They immediately grasp "this epic is XL, this one is S" without needing training on points.
Switched from Fibonacci to T-shirts 3 sprints ago and haven't looked back. Our junior devs were constantly confused about the difference between 5 and 8 points. T-shirt sizes are intuitive - everyone knows what Small vs Large means.
We map them internally: XS=1, S=2, M=3, L=5, XL=8. Best of both worlds. Report T-shirts to stakeholders, track Fibonacci in Jira.
The real question is: do you need precision? For sprint planning with a stable team, Fibonacci's non-linear scale is perfect. For high-level quarterly planning or portfolio estimation, T-shirts work better because stakeholders can grasp them instantly.
We actually use both - T-shirts for epics during PI planning, Fibonacci for stories during sprint planning. Different tools for different contexts.
Unpopular opinion: Modified Fibonacci (1, 2, 3, 5, 8, 13, 20, 40, 100) beats both. Gives you the conversation-forcing gaps of Fibonacci but adds 'too big' markers at 20, 40, 100 that scream "split this story!"
T-shirts lack that forcing function. I've seen teams happily estimate "XXL" stories that should obviously be split. With 40 or 100 points, it feels absurd and forces decomposition.