/
Estimation Accuracy

Estimation Accuracy

Owned by Ivan_Kaschenko
Last updated: Jan 19, 2022 by Anastasia Baranouskaya
5 min read

Purpose

Estimation widget shows how far team estimations from a reality are, how many tasks are not estimated - for the whole project team, per each sub-team, for particular individuals on a project. Additionally it helps to analyze a personal productivity of team members.

In general, a correlation of estimations and reported efforts is one of key indicators about a stability (and predictability) of a process. To some extent this tells how we can trust to other metrics and forecasting.

Effort variance (overall)

Shows a ratio of Estimated efforts (a sum of Original Estimate) to Actual efforts (Logged time) per last 12 months for the whole project. Computed over completed work having non-zero Original Estimates and non-zero time reported. Calculated as (Total Logged Effort - Total Estimated effort) / Total Estimated Effort * 100%.

NOTE!

This chart shows a global view on estimation variance for the whole team. It is not impacted by "custom time period" or "selected issue types". 

Controls panel

This is a group of controls to set right context for charts on this tab. 

  • Show for <time range> - predefined intervals to check team estimation accuracy. Available presets: Last 90d - Last 30d - Last 7d - Sprint - Release - Custom interval. 

  • Issue type - ability to narrow the analysis to a particular set of issue types. By default, all types are enabled.

Summary

This section shows a total estimate accuracy for selected context (time interval, selected scope of issue types) - originally estimated hours vs. actually logged resulting in % of a ratio above or below initial estimate. 

Actual effort vs. Original estimate

Shows the distribution of completed items based on the comparison of their Original Estimate and Logged Time. An item is considered as estimated if estimate is given in Hours or any of its sub-items is estimated in Hours. Context: the chosen time interval and selected issue types. The more green and less red - the better. The goal of this view is to identify team estimation mistakes timely via amber and red segments - and retrospect that with the team in order to improve their estimation accuracy going forward. If a team doesn't estimate their work in hours - this chart will be blank. 

The left figure shows how much of accomplished work had been estimated. An item is considered as estimated if estimate is given in Hours or any of its sub-items is estimated in Hours. This gives a kind of "precision" for the right chart - because only mentioned % of items on the left participates in the right chart.

 

Estimation thresholds

Estimation thresholds are set in Project settings. 

Default values:

  • Effort exceeds estimates significantly: default value is 2, i.e. red zone in your chart includes items where logged effort exceeds estimates by >100%.

  • Effort exceeds estimates: default value is 1.2, i.e. amber zone in your chart includes items where logged effort exceeds estimates by >20% and <100%.

  • Effort less than estimates: default value is 0.8, i.e. green zone in your chart includes items where estimate accuracy is between -20% and 20%, and blue zone in your chart includes items where estimate accuracy is < -20%.

Personal estimate accuracy

Personal Estimate Accuracy shows a kind of productivity measurement for every team member as well as accuracy of their individual estimates. Sorting via various attributes helps to see leaders in appropriate categories. 

  • User - a team member name. 

  • Logged work, h - a sum of all hours logged by a team member to any items within a selected context (time interval, selected work item types). 

  • Unestimated work done, h - a sum of all hours logged by a team member to items originally NOT estimated in hours and completed within a selected period.

  • Unestimated work % - a ratio of the effort reported on non-estimated work vs. an estimated one. For example, if tasks on your project are estimated, and bugs are not, this value shows a % of effort spent on bugs.

  • Estimate accuracy, items

  • Estimate accuracy, actual hours

How does it work? Examples:

  1. A task, estimated as 6h, 1 person worked on it and completed in 7h. Thus, actual exceeds estimated with 17% => the task added into a green area. 

  2. Another task, estimated as 10h, 2 people A and B worked on it, logged respectively 6h + 8h = 14h. Thus, estimation variance is 40% => the task will be added into amber area of both A and B (in general, of all people who was involved into accomplishing of a task).  

This report works the best if tasks are accomplished by dedicated people in JIRA i.e. "single-person" assignments are used instead of sharing tasks between multiple people.

Otherwise, if multiple people work on same tasks (and report time to them) or tasks re-assigned through multiple people (e.g. business analyst to developer, developer to tester, etc.) - stats will be less accurate.

Multi-person (shared) tasks

As personal details works best in case of single-person assignments, not shared or chained ones, the above grid about Personal Estimation Accuracy didn't include shared tasks. Although, there are projects where the most of the work is accomplished via re-assignments from one person to other. In such case it is not possible to detect if a particular individual exceeds an estimate - because estimates are usually totals. So, this section at least shows how much work on the project is being done in a "shared" way and what is the overall accuracy in those tasks - via comparison of original estimation in those tasks to the total work logged (by all people involved).

Related content