What Is Effective Velocity? The Metric That Corrects for AI-Assisted Rework
By Anshul Gupta · Published 2026-09-12 · Technology & Innovation
Short answer: Effective Velocity is a corrected version of standard Scrum velocity that subtracts rework points from raw completed points, revealing how much of a sprint’s output was genuinely done, not just marked done. It’s calculated as Raw Velocity minus Rework Points. A team that completes 40 points but later reworks 5 of them has an Effective Velocity of 35, a very different number from the 40 that shows up on a standard velocity chart.
The Formula
Effective Velocity = Raw Velocity - Rework Points
This is one of the core metrics inside the Accountable AI Framework (AAF), built specifically to correct a blind spot that standard Scrum velocity has always had, one that became far more consequential once AI started generating a meaningful share of sprint output.
A Worked Example
Say a Scrum team completes 40 story points in a sprint. Everything looks like a strong sprint on the burndown chart. Two sprints later, 5 points’ worth of that work comes back for rework, a calculation error, a missed edge case, a detail an AI-generated summary got wrong.
Effective Velocity = 40 - 5 = 35 points
The sprint that looked like a 40-point sprint was genuinely a 35-point sprint. The other 5 points were never actually finished the first time, they just looked finished until someone caught the problem.
Why Standard Velocity Doesn’t Catch This
Velocity was built to answer one question: how much can this team reliably complete per sprint? It answers that question well when every task has a human directly producing it, because completion and correctness tend to track closely together.
That assumption breaks down with AI-assisted work. A task can look completely finished, code that compiles, a document that reads cleanly, a summary that sounds coherent, while quietly containing an error that won’t surface until later. Standard velocity has no mechanism to catch this, because it’s measured at the moment of completion, before problems have had a chance to show up.
The result is a genuinely misleading pattern: a team’s velocity chart can look strong and stable sprint over sprint, while its Effective Velocity, the number that reflects what was actually verified as correct, is quietly declining underneath it.
Effective Velocity vs Trust Score: What’s the Difference?
These two metrics are related but answer different questions, and teams often confuse them.
Trust Score is a percentage. It tells you what proportion of completed work was right the first time, useful for tracking quality as a rate over time.
Effective Velocity is a point value. It tells you the actual, corrected throughput of a sprint, useful for sprint planning and forecasting, since it’s expressed in the same story-point units a team already plans with.
A team can use both together: Trust Score to monitor the trend, Effective Velocity to correct the actual planning number a Scrum Master uses when committing to future sprints.
Why This Matters More as AI Involvement Grows
The gap between Raw Velocity and Effective Velocity tends to widen as a team’s AI-assisted work increases, not because AI is unreliable, but because verification hasn’t scaled alongside adoption. A team that was 90% human-authored a year ago and is now 40% AI-assisted is carrying a genuinely different risk profile, even if its raw velocity chart looks identical across both periods.
Tracking Effective Velocity specifically surfaces this shift. A stable raw velocity paired with a widening rework gap is often the earliest visible sign that AI-assisted work isn’t being reviewed at the depth its actual risk requires.
How to Track Effective Velocity Without New Software
This doesn’t require a dedicated analytics platform. In JIRA, Raw Velocity is already available through the standard Velocity Chart report. Rework Points can be captured with a saved filter for items marked as reopened or requiring correction within a sprint or shortly after. Subtracting the second number from the first, even manually at the end of each sprint, produces a usable Effective Velocity figure.
Frequently Asked Questions
Is a gap between Raw Velocity and Effective Velocity always a problem?
Not necessarily a small, occasional gap is normal in any delivery process. What’s worth investigating is a gap that’s genuinely widening sprint over sprint, particularly alongside increasing AI-assisted work, since that pattern often means review isn’t keeping pace with adoption.
Should Effective Velocity replace Raw Velocity in sprint planning?
Effective Velocity is generally the more honest number to plan future sprints against, since it reflects what a team can reliably deliver without rework, not just what initially got marked complete. Many teams track both, using Raw Velocity for historical comparison and Effective Velocity for forward planning.
Does Effective Velocity apply to Kanban and SAFe as well?
The underlying correction, Raw Throughput minus rework, applies across Kanban, where it’s calculated as Effective Throughput over a chosen time window rather than a sprint. In SAFe, it can be rolled up across an Agile Release Train for a Program Increment.
How is this different from AI Error Rate?
AI Error Rate isolates a percentage specific to AI-assisted tasks. Effective Velocity is a point-based, whole-sprint correction that applies to all rework, regardless of whether it originated from AI-assisted or human-authored work. Used together, they show both the rate of AI-specific issues and the actual point-cost of rework across the sprint.
Effective Velocity is one of five core metrics inside AAF, the Accountable AI Framework. Learn the complete formula set, including Trust Score, AI Error Rate, and Override Rate, at https://www.agileaipro.com/certifications