Fiboshirti

Relative estimating for agile frameworks

 by Professor Olaf Bjorklund, Department of Informatics & Data, University of Jomala, Aaland

Relative Estimating - A New Approach

Consistent estimating at scale for agile projects in informatics

Research Summary: As agile projects have become commonplace, the role of story pointing to identify estimated effort has grown. Several techniques now exist for this (outlined in Mike Cohn's article of effort vs complexity) but each has limitations. Scrum teams have also found simple hours/days estimates can be inaccurate (as Dan Radigan on story points vs hours) This research aims to identify an approach that combines the best of the two main techniques, Fibonacci (David Pereira, April 2023) and T-Shirt sizing (Sean Blake, 2021).

In the Fibonacci approach, estimates are identified using the Fibonacci numbers, though often limited to the smaller numbers up to 21 as in 0,1,1,2,3,5,8,13,21. However some team members have been known to mistakenly select numbers outside the sequence, such as 20, leading to some confusion about the estimated effort. Additionally, the start of the sequence 0,1,1,2,3 is unhelpful in sizing smaller tasks and some guides on the subject leave out numbers (as the article in Scrum.org), perhaps to avoid this issue.

T-Shirt sizing has the advantage of being more familiar to software developers. However, the sizing sequence is not an internationally agreed standard with some teams adding sizes such as XXS, XXL or even XXXL depending on local cultural norms. Some projects have encountered difficulty with the gender-relative sizing, where it is not possible to agree whether a men's small size is larger or smaller than a women's medium.

The Fiboshirti Sequence is described as a technique to address these issues and proposed as an international standard for relative estimating. In Fiboshirti the smaller tasks can be described by XS instead of 0, 1, 1, 2 or 3. The more commonly used 5 and 8 of Fibonacci are then added, but with XL being used to describe the largest of tasks. So the full sequence is as follows:

XS, 5, M, 8, 13, L, XL

This allows teams to more accurately define their estimates with a range of values that remain relative but also familiar, while avoiding the arithmetic complexities of calculating the less familiar Fibonacci numbers and the cultural differences encountered with T-Shirt sizes. The full research paper is available for subscribers to the on-line university library.