From Velocity to Throughput: The Metric Shift AI Forces on Leaders
- Darren Emery
- Dec 22, 2025
- 7 min read
For years, enterprise leaders have been told the same story:
“If teams are delivering faster, the organisation must be performing better.”
So dashboards fill up with reassuring signals:
Velocity trending upward
Capacity fully utilised
Roadmaps on track
Delivery dates confidently forecasted
And yet, right now, many executives are staring directly at a frustrating contradiction:
The dashboards are green.The outcomes are red.
And no one can quite explain why - without blaming the teams.
AI didn’t create this gap. It just made it impossible to ignore.
Velocity Was Always a Proxy - We Just Forgot What It Was Standing In For
Velocity was never meant to be a goal.
In its original context, it was a local learning signal - a way for a stable team to understand its own delivery rhythm and improve predictability over time.
Somewhere along the way, it lost its original purpose.
It became:
A performance metric
A comparison tool
A planning input
An executive assurance mechanism
In short, velocity stopped being a learning aid and became a control signal.
And like most control signals applied to complex systems, it delivered exactly what it incentivised.
More activity.
More output.
More work in progress.
Not more value.
AI Makes the Proxy Problem Impossible to Ignore
According to a recent BCG report, 66% of executives are dissatisfied with their progress on AI and GenAI - despite record levels of delivery activity.Why the disconnect?
AI dramatically lowers the cost of execution.
Ideas become backlog items faster.
Code gets written quicker.
Analysis takes minutes instead of weeks.
On the surface, this looks like a win.
But there is a catch:
When execution accelerates, measurement errors scale as well.
If you measure activity, you will get:
More activity
Faster activity
Larger backlogs
AI doesn’t fix that. It amplifies it.
Velocity climbs.
Utilisation improves.
Delivery “speed” increases.
And yet customer impact barely moves.
This is why so many AI initiatives feel productive but underwhelming - fast locally, but flat systemically.
The Metric That Actually Matters: Throughput
If velocity measures how busy teams are, throughput measures how effectively the organisation converts intent into outcomes.
Throughput asks different questions:
How long does it take for an idea to become customer value?
Where does work queue, stall, or fade away?
How much value actually reaches customers - and stays there?
This is why throughput is a leadership metric, not a team metric.
It spans:
Strategy choices
Funding models
Architecture constraints
Governance latency
Decision-making speed
No single team controls throughput.
Executives do.
A Statistic That Should Make Leaders Uncomfortable
If there is a disconnect between how busy your teams look and how slow delivery feels, this statistic explains why:
Studies consistently show that in large enterprises, less than 20-25% of total lead time is spent doing value-adding work.
The rest is waiting - for decisions, approvals, funding, prioritisation, or coordination.
In other words:
Teams are rarely the bottleneck
The system almost always is
Velocity can increase inside teams while flow efficiency across the organisation remains catastrophically low.
AI doesn’t change this dynamic.
It just helps work reach the waiting points faster.
Why AI Forces This Shift Now (Not “Eventually”)
Before AI, inefficiencies were expensive but slow.
Now, they are fast and visible.
When teams can:
Generate features faster than governance can approve them
Run experiments faster than funding can adjust
Respond to customer signals faster than strategy can absorb them
The bottleneck moves upward.
Execution stops being the constraint.
Leadership systems become the constraint.
This is the moment many executives are experiencing right now - even if they haven’t named it yet.
Green Dashboards, Red Outcomes: A Familiar Pattern
If this feels abstract, it usually isn’t.
It shows up in familiar executive frustrations:
“We’re delivering more than ever, but customers aren’t noticing.”
“Everything is ‘on track’, yet benefits never materialise.”
“We keep shipping, but the needle barely moves.”
These aren’t execution failures.
They’re measurement failures.
“If you tell me how you measure me, I will tell you how I behave.”- Eli Goldratt
Most organisations are getting exactly what their metrics are asking for:
the efficiency of the build - not the effectiveness of the outcome.
How Velocity Distorts Behaviour at Scale
When velocity becomes the signal of success, predictable behaviours emerge:
1. Work Gets Broken Down to Look Fast
Features are sliced smaller to keep numbers high - even if the slice delivers no standalone value.
2. Teams Optimise Locally
Teams deliver “their part” efficiently, even when the end-to-end flow remains blocked.
3. Dependencies Multiply
Work completes faster into queues, approvals, and integration bottlenecks.
4. Decision Quality Degrades
Speed of output replaces quality of choice. Doing something becomes safer than deciding not to.
AI accelerates all of this.
The organisation doesn’t become faster.
It becomes busier.
A Real Example: When Flow Improves but the System Pushes Back
At a mid-sized financial services firm I worked with, we attempted to redesign product delivery around throughput rather than activity.
The operating model was deliberately simple:
A cross-functional business outcomes team (product, architecture, delivery, change, and business stakeholders), accountable for a single value stream
A standardised flow process, designed to minimise handoffs and keep work moving
A single weekly alignment forum, where this group could review upcoming initiatives, assess signals, and decide whether to pivot or persevere
The results were hard to ignore.
Flow efficiency increased by 240%.
Work in progress was reduced by 87%.
Decisions happened faster.
Dependencies surfaced earlier.
Value moved more predictably.
On paper, this was exactly what every modern operating model claims to want.
And then… it stalled.
Not because the model didn’t work.
But because the organisation wasn’t ready to trust it.
Funding remained annual and project-based.
Governance stayed layered and risk-averse.
Additional approval steps were added “for safety”.
The very system that needed to change wrapped itself around the new model - adding friction, overhead, and complexity - until the flow advantage collapsed.
The lesson was clear:
You cannot improve throughput inside a system that refuses to change how it decides, funds, and governs work.
This wasn’t a delivery failure.
It was a leadership one.
Throughput Exposes the Real Constraints
When leaders shift their lens to throughput, uncomfortable truths surface quickly.
Throughput reveals:
That funding decisions lag strategic intent by months
That governance forums are the slowest part of delivery
That architecture choices throttle speed more than teams ever did
That prioritisation is reactive, not intentional
Throughput turns vague frustration into visible constraint.
And constraints are leadership work.
The Hidden Mismatch: Roadmaps Change, Funding Doesn’t
Here’s a pattern most executives recognise immediately:
Enterprise roadmaps change materially every 90 days.
Funding models, governance structures, and success measures rarely do.
This creates a structural contradiction:
Strategy adapts
Priorities shift
Teams reorient
But money, approvals, and incentives remain fixed.
Throughput suffers - not because teams are slow, but because the system resists change.
AI simply exposes this mismatch sooner.
Throughput Connects Strategy to Reality
Velocity answers: “How fast are teams moving?”
Throughput answers: “Are we moving in the right direction - and does it matter?”
Throughput forces alignment across:
Strategic intent
Portfolio decisions
Value streams
Delivery systems
It removes the comfort of local optimisation and makes systemic issues impossible to ignore.
That’s why throughput feels uncomfortable.
It removes plausible deniability.
What Leaders Must Redesign (And What They Must Stop Delegating)
Shifting from velocity to throughput is not:
A tooling change
A framework rollout
A coaching initiative
It is an operating model decision.
Specifically, leaders must take ownership of:
1. Decision Latency
How long does it take to:
Kill work?
Reprioritise?
Fund new learning?
Throughput collapses when decisions lag reality.
2. Funding Flow
Is funding aligned to value streams or locked into annual bets?
If money can’t move, neither can value.
3. Architectural Constraints
Are teams constrained by dependencies they don’t control?
Throughput flows where architecture allows it.
4. Incentives and Metrics
Are leaders rewarded for:
Activity and utilisation?
Or outcomes and learning?
Metrics shape behaviour far more than intent statements ever will.
The Executive Shift AI Demands
AI removes execution scarcity.
That changes the leadership job.
The hardest problems are no longer:
How fast teams can deliver
How much work they can handle
How efficient individuals are
The hardest problems become:
What not to pursue
When to change direction
How quickly the organisation can learn
How to stop work without drama
These are throughput problems.They cannot be delegated.
AI makes every car faster.
Leaders are no longer engine tuners - they are traffic controllers.
Speed without orchestration creates collisions, congestion, and unrealised value.
A Practical Question for Senior Leaders
If you’re leading a large organisation today, ask yourself this:
Where does work wait the longest - and why?
That answer will tell you far more about performance than any velocity chart ever will.
Because where work waits, value stalls.
Why This Matters Now
AI is not waiting for governance to catch up.
Markets aren’t pausing for funding cycles.
Customers aren’t impressed by internal efficiency.
The organisations that win won’t be the fastest at doing more.
They’ll be the best at:
Choosing wisely
Learning quickly
Redirecting decisively
That requires a shift from velocity to throughput - and from delegation to leadership.
A Practical Invitation
If you recognise this gap in your organisation, the work isn’t to add another metric.
Our Strategy to Execution workshop helps leadership teams:
Align metrics to strategic intent
Expose throughput constraints across the system
Identify where speed is being wasted
Create decision and funding signals that match reality
No dashboards theatre.
No framework debates.
No cosmetic change.
Just clarity - where it actually matters.
If you’re interested, let’s have a 30-minute conversation before the next planning cycle locks in another year of green dashboards and red outcomes.
AI won’t fix organisations that mistake movement for progress - but it will expose them faster than ever.
Comments