Probabilistic Feature Forecasting vs. PI Planning

Executives want a simple answer to a simple question:

“What can we commit to delivering in the next quarter?”

In many organizations, PI Planning exists to answer exactly this question. It produces a list of features, aligned across teams, with a shared commitment to deliver over a fixed time horizon.

But there is a problem.

Those commitments are often unreliable.

The Illusion of Certainty

PI Planning creates a sense of certainty at the start of a quarter. Teams align. Dependencies are mapped. plans are negotiated. Commitments are made.

But as execution begins, reality intervenes:

• new information emerges
• dependencies behave differently than expected
• priorities shift
• work takes longer than planned

By the end of the quarter, what is actually delivered has diverged significantly from the original plan.

PI Planning provides certainty at the start and uncertainty at the end.

A Different Approach: Forecasting Instead of Committing

A flow-based delivery system answers the same executive question in a different way.

Instead of committing to a fixed scope upfront, it uses observed system performance to forecast what is likely to be delivered.

This is probabilistic feature forecasting.

It replaces speculative commitment with evidence-based prediction.

How It Works

A flow-based ART continuously measures its delivery performance using simple flow metrics:

• throughput (features completed per iteration)
• cycle time
• work-in-progress

Over time, this creates a stable understanding of delivery capacity.

For example:

Now project forward across a quarter (e.g. 6 sprints):

Finally, map this onto a ranked backlog:

What Executives Actually Get

Instead of a brittle commitment, executives receive:

• a realistic delivery range
• clear confidence levels
• a continuously updated forecast

Uncertainty is made explicit at the start — and reduced over time.

This is the opposite of PI Planning.

Flow-based forecasting starts with uncertainty and converges toward certainty as work is completed.

Why This Works Better

• It reflects actual system performance, not optimistic planning
• It adapts continuously as new information emerges
• It exposes risk early through flow signals
• It reduces the need for large coordination events

Most importantly, it aligns with how complex delivery systems actually behave.

The Real Shift

This is not about removing planning.

It is about replacing speculative commitment with evidence-based forecasting.

Or more simply:

Stop committing to scope. Start forecasting capability.

Read my other article for a deeper dive into on Probabilistic Forecasting.

Final Thought

As delivery systems become faster, more automated, and increasingly AI-assisted, the ability to generate work is no longer the constraint.

The constraint is the system’s ability to absorb, coordinate, and complete that work.

And that is exactly what flow-based forecasting measures.