Release Forecast

Release Forecast is the projected outcome of a release plan, expressed as an estimate of when a given scope will be delivered, or alternatively how much scope will be achievable by a given date, derived from capacity and delivery data inputs rather than from unvalidated assumption. It represents the point at which release planning translates raw historical and capacity data into an actionable prediction that stakeholders can use to plan around.

Core Concept

A Forecast, Not a Promise

A release forecast is explicitly probabilistic in nature, reflecting the inherent uncertainty in any prediction about future work, rather than functioning as a guaranteed commitment. This distinguishes it conceptually from a fixed release date announced with full confidence, even though in practice a forecast is often the basis from which such commitments are eventually made.

Two Common Forecasting Directions

A release forecast can answer either of two related but distinct questions: given a fixed scope, when will it likely be completed, or given a fixed date, how much scope is likely to be achievable by then. The appropriate direction depends on which variable, timeline or scope, is treated as the fixed constraint in a given planning context.

Methods for Producing a Release Forecast

Deterministic Velocity-Based Forecasting

The simplest approach divides the total estimated scope by the team's average historical velocity, producing a single projected completion point without explicitly representing the underlying uncertainty.

Monte Carlo Simulation

A more sophisticated approach runs many simulated iterations of the release, each drawing randomly from the team's historical velocity distribution, producing a range of possible completion dates along with the probability associated with each, rather than a single deterministic figure.

Throughput-Based Forecasting

Rather than relying on story point estimation at all, some teams forecast based purely on the historical rate at which discrete work items, regardless of size, are completed, using that count-based throughput to project how many items are likely to be finished within the release horizon.

Communicating Forecast Uncertainty

Confidence Intervals

Presenting a forecast as a range, such as a date by which there is a specified probability of completion, communicates the genuine uncertainty inherent in the prediction more honestly than a single point estimate.

Where $p$ represents the probability, derived from simulation or historical variance, that the release will be complete by date $d$.

Multiple Confidence Levels

Presenting forecasts at several confidence levels, such as an optimistic, likely, and conservative date, gives stakeholders a fuller picture of the range of realistic outcomes than any single number could convey.

Visualizing Trend Over Time

Tracking how the forecast itself shifts across successive updates, rather than presenting only the current forecast in isolation, helps stakeholders understand whether the release is trending toward or away from its original projection.

Updating the Forecast

Recalculating After Each Sprint

As each sprint completes, actual velocity and remaining scope are used to refresh the forecast, incorporating the latest real performance data rather than relying on the assumptions made at the outset of the release.

Incorporating Scope Changes

Additions or removals from release scope during the course of the release should immediately feed into an updated forecast, since a static forecast that ignores scope changes quickly becomes misleading.

Responding to Significant Forecast Shifts

A forecast that shifts substantially between updates warrants investigation into the underlying cause, whether a genuine change in team capacity, a scope change, or a one-time anomaly, rather than being passed along to stakeholders without context.

Risks of Poor Forecasting Practice

False Precision

Presenting a release forecast as a single, confidently stated date, without acknowledging the underlying uncertainty, sets stakeholders up for disappointment when actual delivery inevitably varies from that exact prediction.

Anchoring on an Early, Overly Optimistic Forecast

An initial forecast produced with limited data or before accounting for known risks can become an anchor that is difficult to revise upward even as later evidence suggests it was too optimistic, particularly once it has been communicated externally.

Infrequent Updates

Failing to refresh the forecast regularly as new sprint data becomes available allows it to drift further from reality the longer it goes unrevised, undermining its usefulness as a planning tool.

Best Practices

Base Forecasts on Historical Data, Not Aspiration

Grounding forecasts in actual measured velocity and delivery variance, rather than in what the team hopes to achieve, produces predictions that hold up better under real execution.

Communicate Ranges, Not Single Dates

Presenting forecasts with explicit confidence ranges, even if stakeholders initially prefer a single number, builds more durable trust than repeatedly missing a falsely precise point estimate.

Refresh the Forecast on a Regular Cadence

Updating the release forecast after every sprint, incorporating the latest velocity and scope data, keeps it a genuinely useful and current planning tool throughout the life of the release.