So you've accepted that forecasting and agile can be friends after all (if you missed that revelation, pop over to Part 1). Now comes the fun part: actually, making it work in the wild, where stakeholders roam free and deadlines have sharp teeth.

Welcome to the "how" guide. Not the theoretical how that works perfectly in conference presentations, but the practical how that works on Monday morning when your CEO casually asks for "just a rough timeline" while pouring coffee.

This isn't for the agile purists who get hives when they hear the word "forecast." It's for those of us living in the messy middle, trying to balance the agile dream with the reality of stakeholders who still – bless their hearts – just want to know when things will be done.

The Five Things You Actually Need (Not What the Books Tell You)

1. Batch Size Management (Or: Stop Creating Work Monsters)

Let's be brutally honest: your team is terrible at estimating large chunks of work. Everyone is. Those enormous work items that take weeks? They're unpredictability bombs waiting to explode your forecasts.

What actually works:

• Break down anything that feels "big" until it feels almost too small

• Ruthlessly enforce work-in-progress limits (your team will hate this at first)

• Track how long different sizes of work take to complete, then use your actual data instead of wishful thinking

• Stop pretending that complex work is predictable – embrace the variance and plan for it

Without this foundation, everything else is just fancy decoration on a house of cards.

2. Two-Way Estimation (Or: How to Speak Two Languages at Once)

Your executives want roadmaps with dates. Your developers want to work in sprints with story points. Both are right, and both are wrong.

The balancing act that works:

• Keep high-level portfolio planning deliberately vague (quarters, not days)

• Let teams estimate in whatever unit makes them comfortable

• Create a translation layer that connects the two without promising false precision

• Update the translation regularly as you learn more (monthly minimum)

• Never, ever treat early estimates as commitments

If you only show executives the top-down view or only work from the bottom-up view, you're setting yourself up for painful conversations.

3. Value-Based Prioritisation (Or: How to Actually Say No)

When everything is top priority, nothing gets done predictably. Most "prioritisation" frameworks in organisations are performative theatre.

What you need instead:

• Brutally clear criteria for what constitutes business value

• A documented process for making trade-offs that everyone is held to

• Regular (at least bi-weekly) prioritisation sessions with empowered stakeholders

• A "one in, one out" policy for urgent requests

• The courage to publicly track priority changes and their impact on delivery dates

Without this, your forecast is just a fantasy document no one believes.

4. Value Slicing Capability (Or: Delivering Something Useful Instead of Technical Pieces)

Most teams are terrible at this. They slice work by technical layers (database, API, UI) instead of by functionality that delivers actual value.

How to get better:

• Train your team in vertical slicing techniques (with real examples, not abstract concepts)

• Practise identifying the smallest releasable increment that provides value

• Map dependencies between slices so you know what must come first

• Create clear, testable acceptance criteria for each slice

• Celebrate when early slices deliver actual value, even if the "full feature" isn't done

This is where the real agility happens: the ability to deliver value incrementally instead of holding everything until the mythical "done."

5. Collaborative Estimation (Or: Stop Letting One Person Set You Up for Failure)

If your estimates are coming from one person (especially a manager), you're doing it wrong. If they're coming from people who won't do the work, you're doing it wrong.

The approach that doesn't lie:

• Get every discipline in the room (dev, test, UX, product, etc.)

• Use techniques that surface disagreement instead of hiding it (hint: if you're not arguing, you're not a high-performing team)

• Give extra weight to the voices of those who've done similar work before

• Create psychological safety so people can say "this seems harder than you think"

• Regularly calibrate by comparing estimates to actuals

One-person estimates are just guesses. Group estimates with the right process are forecasts.

The Support System You Can't Skip

Regular Review Cadence (Or: How to Spot the Train Wreck Coming)

You need:

• Weekly progress tracking that's honest, not performative

• Monthly forecast reviews that adjust based on what you've learned

• Quarterly planning sessions that embrace the uncertainty of the future

• Clear outcomes for each meeting so they don't become status theatre

• Defined roles so everyone knows who makes what decisions

Without this discipline, forecasts go stale and lose credibility fast.

Actually Skilled Facilitation (Not Just a Person with a Whiteboard)

This is the unsung hero of good forecasting. You need someone who can:

• Guide planning sessions without imposing their own agenda

• Surface and address disagreements productively

• Help groups reach consensus without forcing false agreement

• Keep discussions focused on outcomes, not process debates

• Create an environment where bad news surfaces early

Bad facilitation leads to artificial consensus and forecasts no one believes in.

Unified Data (Or: Stop Playing Chinese Whispers with Your Metrics)

The fastest way to kill forecast credibility is having different numbers in different places.

You must have:

• A single source of truth everyone can access

• Consistent tool usage (pick one and stick with it)

• Regular data quality checks (data needs love and attention)

• Clear, agreed-upon metrics everyone understands

• Thoughtful balance between manual and automated data collection (yes, Excel has its place)

When data conflicts, trust evaporates.

Making it Work When Nothing Else Has

The real magic isn't in any one of these elements. It's in how they work together:

• Consistent, small batch sizes make estimation more reliable

• Collaborative estimation surfaces risks before they become problems

• Clear prioritisation prevents constant scope churn

• Good facilitation keeps the process honest and focused

• Reliable data builds trust that the forecast means something

This isn't about perfect prediction. It's about providing just enough information for good business decisions while maintaining the flexibility to respond to change. That's the pragmatic middle ground where real delivery happens.

The Uncomfortable Truth

Most organisations want the benefits of forecasting without the discipline required to make it work. They want certainty without investing in the capabilities that create predictability.

But if you're willing to put in the work, you can have both agility and reasonable forecasting. Not perfect forecasting, but good enough to make sound business decisions.

Start small. Build capability. Earn trust through transparency. And remember that the goal isn't perfect prediction—it's helping your organisation navigate uncertainty while delivering value.

Need a practical tool to get started? Download our free Excel-based burn-up forecasting template - purpose-built for the messy reality of hybrid delivery environments.