The Most Dangerous Question in Tech Leadership
As an Agile Coach and delivery consultant, I spend a lot of time in sprint retrospectives and leadership planning sessions. Without fail, I eventually hear a well-meaning executive or stakeholder ask the exact same dangerous question: "Why isn't our team's velocity increasing every single Sprint?"
This question reveals a fundamental misunderstanding of agile frameworks. Leaders often treat velocity like a radar gun, assuming higher numbers mean developers are typing faster and working harder. This mindset turns a valuable planning tool into a weapon, ultimately destroying trust, predictability, and software quality.
Let's clear the air once and for all: Velocity is a measure of capacity, not productivity. When measured correctly, it answers one highly specific question: "How much work can this specific team predictably cross the finish line in a standard Sprint?"
If you want to build high-performing tech teams, you have to fundamentally change how you view this metric. Here is how elite engineering teams actually measure, interpret, and leverage velocity.
What Exactly is Agile Velocity?
Agile velocity is a historical measure of a specific team's capacity to predictably deliver work within a standard Sprint boundary. It is absolutely not a measure of developer productivity, individual performance, or raw coding speed. Instead, it serves as a forecasting flashlight for Product Owners trying to plan future releases.
Think about a delivery truck. Its capacity is defined by the volume it can hold and the weight it can safely transport. If you overload the truck to make it "faster" or more "productive," the axles break, and the cargo never arrives. Software development works exactly the same way.
When a Scrum Team calculates its velocity, they are figuring out the sustainable payload they can carry across the finish line every two weeks without burning out or skipping critical quality checks. Predictability is the ultimate goal. You want a team that consistently delivers exactly what they commit to, rather than wildly fluctuating between heroics and burnout.
The 100% "Done" Rule: Why Partial Credit is Toxic
You must only count work that strictly meets the team's Definition of Done because partial credit destroys predictability. If an 8-point story is 90% complete when the Sprint ends, the team scores exactly zero points for it. You do not get 7 points for effort.
This sounds harsh to new Scrum Masters. I get pushback all the time. Developers will argue that the code is written, and it just needs one minor QA sign-off. But the hard truth of software engineering is that value is only realized when software is in the hands of the user.
It hides process bottlenecks: If an item is stuck in QA every sprint, scoring it as "partially done" masks the testing bottleneck from leadership.It creates an illusion of progress: Imagine telling stakeholders a payment gateway is 90% done. It still cannot process payments. It holds zero business value until it goes live.It breeds partial quality: Rewarding almost-finished work incentivizes teams to rush the final 10%—usually documentation, testing, and edge-case handling.
Zero points forces a necessary, uncomfortable conversation during the Sprint Retrospective. It forces the team to ask why the work rolled over and how they can slice stories smaller next time.
How to Calculate True Velocity Using Rolling Averages
To calculate a baseline velocity, you must average the completed story points over the last three to five Sprints. You should never rely on a single Sprint peak or trough to forecast future releases, as short-term data is inherently noisy.
Sprints are messy. One sprint includes a major public holiday. The next sprint features an unexpected production outage that pulls senior engineers away for three days. A third sprint goes perfectly smoothly and results in a record-high point completion.
If your Product Owner uses the "perfect" sprint to plan the next release, they are setting the team up for failure. By using a rolling average, you smooth out the variance of human life and unpredictable systems.
For example, if your last four sprints yielded 32, 28, 35, and 30 points, your rolling average is 31 points. That 31 is your capacity limit. When Sprint Planning hits that threshold, the team stops pulling in work.
Velocity is a Vector, Not Just Magnitude
In physics, velocity requires both magnitude (speed) and direction, and agile delivery operates exactly the same way. High point output moving toward the wrong product goals or ignoring architecture just generates technical debt faster.
I have watched teams successfully push 80 points a sprint while management cheers from the sidelines. But when you look closely at what they actually built, the reality is grim. They delivered a pile of poorly tested features that didn't align with the product roadmap, skipped vital code reviews, and ignored essential refactoring.
High magnitude with the wrong direction is a disaster. A sustainable team balances feature delivery with technical maintenance. They use their capacity to build the right things, securely, and with a foundation that will not collapse under its own weight six months from now.
The Deadly Sin of Comparing Team Velocities
Comparing team velocities is a critical leadership failure because story points are completely relative to the specific team that estimated them. Team A’s 30 points will never equal Team B’s 30 points, and putting them on a shared leaderboard triggers immediate dysfunction.
There is a well-known concept called Goodhart's Law: "When a measure becomes a target, it ceases to be a good measure."
When executives pit teams against each other using velocity as a competitive scoreboard, developers adapt instantly. They simply inflate their estimates. A story that used to be a 3 is suddenly pointed as an 8. Overnight, Team A's velocity jumps from 30 to 80 points. Has their productivity increased? No. The only thing that increased was the inflation rate of their estimation scale.
Cross-team comparisons destroy psychological safety. They incentivize developers to protect themselves rather than collaborate. Keep velocity isolated to the team level. It belongs to the team, for the team.
The Leadership Reality Check: How to Actually Improve Delivery
If you want to increase team output, you must stop pressuring developers for higher velocity and start aggressively removing their daily impediments. Sustainable, predictable delivery happens naturally when leaders reduce context switching and pay down technical debt.
Managers who constantly demand higher velocity are treating the symptom while ignoring the disease. If a team is moving slowly, there is always a systemic reason. Your job as a leader is to find it and destroy it.
Actionable Steps for Agile Leaders
Audit the CI/CD Pipeline: Are developers waiting two hours for a build to complete? Automate testing and deployment to reduce feedback loops.Kill Zombie Meetings: Context switching is the enemy of deep work. Protect your team's calendar. Give them dedicated, uninterrupted blocks of time to actually write code.Address Technical Debt: If your codebase resembles a game of Jenga, every new feature carries massive risk. Allocate a specific percentage of capacity every Sprint strictly for refactoring and debt management.Measure What Actually Matters: Shift your focus from Velocity to Lean metrics like Cycle Time (how long it takes to finish work once started) and Deployment Frequency. These metrics reflect real flow and business value.
Velocity is a fantastic tool when respected as a localized capacity gauge. It gives teams the confidence to make commitments they can actually keep. When you respect the boundaries of capacity, you stop burning out your best engineers. Stop using it like a radar gun, put away the performance stick, and start clearing the road ahead.
Originally published at https://aiflowpm.com/agile-velocity-vs-speed-metrics/
Top comments (0)