Forecasting

Forecasting gives a realistic delivery window for work that is already in progress.

We use it to support better decisions about sequencing, dependencies, planning, and when to expect outcomes, while reflecting the fact that work often evolves as it moves through delivery.

How forecasting works

For each in-progress project on the current roadmap we show:

• A target date, based on the current expected completion date
• A pessimistic date, which extends that target date based on confidence

Confidence is translated into a forecast extension:

• High confidence: +20%
• Medium confidence: +40%
• Low confidence: +60%

This gives a visible delivery range instead of a single point estimate. If the target and pessimistic dates fall in the same month, we show one month. If they span multiple months, we show a month range.

Why we do not forecast more precisely

Project delivery dates are influenced by factors that become clearer as work progresses. In practice, that often includes:

• Work enters delivery with uneven levels of definition
• Clarifications often continue after work has already started
• Dependencies and edge cases emerge during build and QA
• Issue scope can expand as the real shape of the work becomes clearer
• Teams absorb interruptions, BAU, and urgent changes alongside project work
• Some delivery effort is consumed by coordination and rework rather than net-new progress

Taken together, these factors make a more precise date less useful than it first appears. A forecast range gives a better view of the likely delivery window.

What the forecast is useful for

Forecasting is most useful for:

• Understanding whether a project is likely to complete within a given reporting window
• Seeing which projects have a wider delivery risk envelope
• Making prioritisation and sequencing decisions earlier
• Setting stakeholder expectations with appropriate caution

It is less useful as a promise of a specific day or week.

What would improve forecast accuracy

Forecast accuracy improves when flow quality improves. The biggest levers are:

• Better issue definition before work starts
• Fewer mid-flight scope changes
• Clearer ownership of clarifications and dependencies
• Reduced interruption load during active project delivery
• Smaller, better-shaped pieces of work entering implementation

As those conditions improve, forecast ranges should narrow and confidence should rise.

Summary

The forecast is designed to be useful, not artificially precise.

It shows the most likely delivery month, the plausible overrun window, and where there is still delivery uncertainty. That makes it more useful than a precise target date that appears exact but may change as the work develops.