Scrum Forecasting and Release Planning with Empirical Data
Introduction
Scrum forecasting and release planning use past delivery and current constraints to give stakeholders a range they can act on. If a team usually finishes 16 to 24 points per Sprint, that history helps it estimate when a release scope could be ready. The forecast should change when the scope, capacity, or evidence changes.
This PSM I lesson covers Forecasting & Release Planning in the Managing Products with Agility focus area. You will calculate and explain a forecast, interpret progress charts, and distinguish the Definition of Done from the arrangements for delivering an Increment.
Learning objectives
- Explain what the Scrum Guide says about forecasting, plans and uncertainty in complex work.
- Compute a simple empirical forecast from past Sprints and explain its limits.
- Read a Product Burn-down, Release Burn-up and cumulative flow diagram.
- Describe release strategies: releasing within a Sprint, at the end of a Sprint, or after several Sprints.
- Explain how the Definition of Done and an organization’s release process affect when an Increment can be delivered.
- Describe the Scrum Master's role in helping with empirical product planning.
- Recognise typical anti-patterns: velocity as a target, a forecast turned into a commitment, and a "hardening Sprint".
1. Forecast versus commitment
Scrum is built for complex work, where the relationship between cause and effect is only clear in hindsight. In that environment a precise long-term prediction is not possible. The Scrum Guide therefore uses the word forecast:
- At Sprint Planning the Developers forecast which Product Backlog items can be Done in the Sprint.
- The Sprint Goal is the commitment for the Sprint Backlog. The selected items are a forecast and can be renegotiated with the Product Owner as the Developers learn more.
- The Product Backlog commitment is the Product Goal. Items in the backlog are not promises.
The Scrum Guide also states that the Scrum Master helps the Product Owner by helping establish empirical product planning for a complex environment. Empirical planning means plans are built from observed data (what the team actually finished) and are inspected and adapted regularly.
Forecasts and Scrum commitments
| Term | Meaning | Where it appears in Scrum |
|---|---|---|
| Forecast | An informed estimate, updated as data arrives | Items selected for the Sprint; how much of the Product Backlog may be finished by a date |
| Commitment | A pledge to achieve a goal | Sprint Goal, Product Goal, Definition of Done |
Many exam questions test exactly this: an option that calls the Sprint Backlog a "contract" or the release plan a "guarantee" is wrong.
2. Empirical forecasting from past Sprints
Because the Scrum Guide does not prescribe estimation units, teams use their own evidence. A common approach is to look at how much work was finished (Done) in recent Sprints and use that range to forecast. Many teams use velocity, the amount of Product Backlog (in the team's chosen unit) turned into Done Increments per Sprint. Velocity is a technique, not a Scrum requirement, and the Scrum Guide does not mention it.
Worked example: when will the release scope be done?
Scenario. A product has 120 points of Product Backlog items required for a first release. The last six Sprints finished 18, 22, 20, 16, 24 and 20 points. Sprints are two weeks long. A stakeholder asks for the date.
Reasoning.
- Average velocity: (18 + 22 + 20 + 16 + 24 + 20) / 6 = 120 / 6 = 20 points per Sprint.
- Range: lowest 16, highest 24.
- Pessimistic forecast: 120 / 16 = 7.5, so 8 Sprints = 16 weeks.
- Average forecast: 120 / 20 = 6 Sprints = 12 weeks.
- Optimistic forecast: 120 / 24 = 5 Sprints = 10 weeks.
Solution. The Product Owner explains that the illustrative scenarios span 10–16 weeks, with 12 weeks at the historical average, assuming comparable work, stable capacity, and unchanged scope. Six past Sprints and their extremes do not establish a probability or confidence level. Update the estimate whenever the assumptions or evidence change.
Why other answers are worse. "Exactly 12 weeks" treats the average as a promise. "Nobody can know" is unhelpful because empirical data does give useful information. Asking the Developers to commit to 20 points every Sprint turns a measurement into a target, and people will respond by inflating estimates.
Limits of velocity
- Use the history of the same team, and reassess comparability if capacity, work mix, or estimation changes.
- Teams estimate differently. Ranking them or setting a point target can reward changing estimates without improving delivery.
- Adding people or pushing for more output in the short term rarely increases it; unfinished or poorly tested work only appears in later Sprints as rework.
- It says nothing about value. A team can have high velocity and deliver items nobody needs.
3. Charts that make progress transparent
The Scrum Guide does not mandate any chart, but it says that increasing transparency of progress is needed because the Scrum artifacts must be understood by those who use them. It also warns that, although trends can be useful, they do not replace empiricism. In complex environments, what will happen is unknown. Only what has already happened may be used for forward-looking decision-making.
Product Burn-down chart
A Product Burn-down chart shows the remaining work in the Product Backlog over time (Sprint by Sprint). The line falls as items become Done. If new items are added, the line may go up. It answers "how much is left, and at what rate are we going down?"
Example: start of Sprint 1 = 200 points remaining. After Sprints 1 to 4: 180, 164, 150, 138. Average burn is 15.5 points per Sprint, so remaining 138 / 15.5 ≈ 9 more Sprints.
Release Burn-up chart
A burn-up chart plots work completed on one line and total scope on another. It makes scope change visible: when the scope line rises, stakeholders see why the end date moved, instead of blaming the team. This is usually clearer than a burn-down, which can hide additions by simply dropping the slope.
Example: the total scope line moves from 120 to 140 points in Sprint 3 when the Product Owner adds a new compliance item. The done line is unchanged. If the team continues at about 20 points per Sprint, the additional 20 points extend the forecast by about one Sprint. The scope line makes the assumption visible.
Cumulative flow diagram
A cumulative flow diagram (CFD) stacks the number of items in each state over time, such as To Do, In Progress, Testing and Done. The vertical thickness of a band shows how many items are in that state. The distance between comparable arrival and completion counts can approximate elapsed time under stable conditions, but does not show an individual item’s cycle time. A widening band shows that items are accumulating. Investigate the cause before deciding how to respond. It is typical for teams that use flow metrics (cycle time, throughput, work in progress) alongside or inside Scrum.
Example: the "Testing" band widens from 4 to 14 items over six weeks. The team inspects why items are waiting and can try limiting work in progress or testing earlier, then measure whether the queue shrinks.
Table: choosing a chart
| Chart | Shows | Useful when | Weakness |
|---|---|---|---|
| Product Burn-down | Remaining work | Scope is stable | Hides scope changes |
| Release Burn-up | Done work and total scope | Scope changes often | Needs two lines to read |
| Cumulative flow | Work per state over time | Looking for bottlenecks | Needs consistent states |
| Sprint Burn-down | Remaining Sprint Backlog work | Developers inspect Sprint progress | Only meaningful inside one Sprint |
Exam tip. The Scrum Guide does not require any of these charts. If an answer says "Scrum requires a burn-down chart", it is wrong. If a question asks what a Scrum Master should do to help stakeholders understand progress, a transparent, empirical option (such as a burn-up showing scope and progress) is usually right.
4. Release strategies
The Scrum Team creates a valuable, usable Increment each Sprint. Delivery can happen when the Increment is ready and operational conditions allow it. A release schedule is an organizational decision.
The Scrum Guide allows delivery before the Sprint ends. Use the Scrum Definition of Done to determine whether the work is part of a usable Increment. Follow the organization’s process to authorize delivery.
| Strategy | Description | Typical context | Benefit | Risk |
|---|---|---|---|---|
| Release during the Sprint | Deliver an eligible Increment when it is ready | Web services with automated deployment | Fast feedback, small risk per change | Needs strong automation and a strong Definition of Done |
| Release at the end of each Sprint | Use the Sprint cadence as a release schedule | Mobile apps, B2B products | Regular rhythm | Waiting can delay value |
| Release after several Sprints | Group Increments into a larger release | Regulated products, hardware, marketing launches | Fits external constraints | Larger batch, later feedback, more risk |
| Fixed date | Target date is set; optional scope can vary | Event deadlines, laws | Gives a planning constraint | Mandatory scope can still be at risk |
| Fixed scope | Scope is fixed, date varies | Contractual feature sets | Scope is complete | Date may slip |
Worked example: fixed date or fixed scope?
Scenario. A ticketing product must support a new national tax rule that comes into force on 1 July. The Product Owner has 10 items related to it. With the current velocity the team forecasts that 8 of the 10 items will be Done by 1 July in the average case, and all 10 by 1 August.
Reasoning.
- The date is fixed by law, so a fixed-date strategy applies.
- The Product Owner identifies the mandatory outcomes with the appropriate stakeholders and orders that work first. See product value and Product Backlog management for how evidence and constraints shape the order.
- The Product Owner and Developers update the forecast as evidence changes. If mandatory work may miss the date, they raise the risk promptly, discuss feasible changes with the responsible stakeholders, and seek help removing obstacles.
- Adding people late is not assumed to help. New members need time to learn and may slow the Scrum Team in the short term.
Solution. Aim to deliver the mandatory behavior by 1 July and schedule optional work later. Release only if the Increment meets the Definition of Done and the applicable requirements. The Scrum Master helps the Product Owner explain the forecast and the trade-offs transparently.
Why other choices are worse. Cutting the Definition of Done or skipping testing to hit 10 items creates hidden technical debt and an unusable Increment. Moving the date is not possible because of the law. Promising all 10 items by 1 July contradicts the data.
The "hardening Sprint" smell
Some organizations schedule a final Sprint for stabilization and testing before a release. Scrum does not have such a Sprint. If work is not Done, the Increment is not an Increment. The team should improve its Definition of Done and automation so that every Sprint produces a releasable Increment.
5. Uncertainty, complexity and risk
Scrum assumes that the team does not know everything in advance. Teams use several tools to deal with that.
- Short feedback loops. A Sprint of one month or less limits the cost of being wrong.
- Thin vertical slices. Delivering a small part end to end (design, build, test, deploy) tests assumptions early and reduces risk more than building one layer at a time.
- Ordering by risk. Items with high uncertainty can be placed early, so that the product learns about them sooner.
- Experiments. A Product Backlog item may be an experiment, with a hypothesis and a measurement, such as testing demand with a prototype before building the full feature.
- Transparency of assumptions. The Product Owner shares what is known and unknown with stakeholders.
Example: reducing risk with a spike and a slice
A team plans to replace its payment provider. The Product Owner orders two items first: a small experiment to confirm the provider's API can handle 500 transactions per minute (risk), and a thin slice that processes one payment type end to end in a test environment (learning). After two Sprints the experiment shows that the limit is only 200 transactions per minute. The Product Owner can use the finding to change the plan after two Sprints, before committing to a full migration.
6. The Scrum Master and empirical product planning
The Scrum Master serves the Product Owner and the organization in several ways that relate to this lesson:
- helping establish empirical product planning, by showing how to build forecasts from data and revise them;
- helping stakeholders understand and enact an empirical approach, so that they do not mistake forecasts for promises;
- coaching the organization toward budgeting and contracts that allow change, such as funding a product incrementally;
- removing impediments that block frequent releases, such as manual deployment steps;
- helping the Scrum Team keep a Definition of Done that makes every Increment usable.
The Scrum Master does not set the release date, promise scope to stakeholders, or act as the project manager who reports on the team's productivity.
Table: stakeholder pressure and the right response
| Situation | Weak response | Better response |
|---|---|---|
| Manager asks for a fixed date and fixed scope | Say yes to avoid conflict | Explain the forecast range; offer fixed date with variable scope or the reverse |
| Manager wants velocity to increase by 20% | Developers inflate estimates | Discuss what limits flow; improve the Definition of Done and impediments |
| Release is delayed because testing is left to the end | Add a hardening Sprint | Move testing into each Sprint and automate it |
| Stakeholder asks for weekly progress report | Write a long status document | Make the Product Backlog, Increment and charts transparent; invite them to the Sprint Review |
7. Pitfalls and exam tips
Pitfall: turning a forecast into a commitment. The Sprint Goal is the commitment. A forecast of items or a release date is updated as data arrives.
Pitfall: velocity as a performance target. It encourages gaming and lowers quality.
Pitfall: waiting for a "release Sprint". Increments should be releasable every Sprint.
Release authority depends on the organization. The Product Owner brings the value perspective. Scrum assigns no universal release approver.
Pitfall: assuming the Sprint Review is a release gate. An Increment may be delivered before the review.
Exam tip. When two answers look right, prefer the one that is transparent, empirical and keeps decisions with the correct role: the Product Owner for Product Backlog ordering, the Developers for the Sprint forecast, the Scrum Master for helping everyone understand Scrum.
Exam tip. The words "guarantee", "promise", "fixed" and "must" in an answer about forecasting are warning signs.
8. Mini case study
Case. A Scrum Team has a velocity range of 25 to 35 points. The Product Backlog for the first release contains 300 points. The sponsor demands: "Deliver all 300 points in 8 Sprints, with a signed commitment, or the budget is cut." The Product Owner worries. The Developers want a hardening Sprint at the end. The sponsor asks the Scrum Master to confirm the plan.
Questions. What should the Scrum Master do and say?
Commented solution.
- Compute the forecast. 300 / 35 = 8.6, so 9 Sprints even in the best case; 300 / 25 = 12 Sprints in the worst case; the average of 30 points gives 10 Sprints. Eight Sprints at the historical rates suggests 200–280 points. At the midpoint rate of 30 points per Sprint, 240 points would be 20% less than the original scope. This is a planning scenario, not a guaranteed capacity.
- Explain options transparently. The Scrum Master helps the Product Owner present a fixed-date option (8 Sprints, with 200 to 280 points, highest-value items first) and a fixed-scope option (300 points, 9 to 12 Sprints). Both come with a plan to inspect progress after every Sprint Review using a burn-up chart.
- Reject the signed commitment on scope. The Scrum Team commits to the Sprint Goal and to the Definition of Done, not to a guaranteed output of 300 points. The Scrum Master teaches the sponsor why in a complex setting.
- Reject the hardening Sprint. Instead, the Scrum Team reviews its Definition of Done and the Developers improve automation to create a usable Increment each Sprint. Earlier delivery may be possible if the release requirements are met.
- Invite the sponsor to the Sprint Review. Seeing working software every Sprint builds trust better than a signed forecast.
Why the alternatives fail. Confirming the plan would mean misrepresenting the data. Increasing velocity through overtime hides quality problems. Letting the Scrum Master decide the scope removes the Product Owner's accountability.
Key terms
- Forecast: an informed estimate of what can be achieved, revised as data arrives.
- Commitment: the Sprint Goal, the Product Goal and the Definition of Done in their respective artifacts.
- Velocity: the amount of work finished per Sprint in the team's own unit; an optional technique.
- Burn-down chart: remaining work over time.
- Burn-up chart: completed work and total scope over time.
- Cumulative flow diagram: items per workflow state over time, used to find bottlenecks.
- Thin vertical slice: a small piece of functionality delivered end to end.
- Empirical product planning: planning from observed data and regular inspection.
Summary checklist
- I can explain the difference between a forecast and a commitment, and name the commitments for each artifact.
- I can compute a range forecast from past Sprints and explain its limits.
- I know velocity is an optional technique and must not be a target or used to compare teams.
- I can describe what burn-down, burn-up and cumulative flow charts show.
- I know Scrum has no hardening Sprint and each Sprint should produce a usable Increment.
- I can explain why an eligible Increment may be delivered during a Sprint.
- I can explain fixed-date and fixed-scope strategies and why quality (the Definition of Done) is not the variable.
- I can describe how the Scrum Master helps with empirical product planning without making release decisions.
Sources
- The Scrum Guide, November 2020: https://scrumguides.org/scrum-guide.html (opens in a new tab) (sections "Scrum Theory", "Sprint Planning", "Scrum Master", "Increment").
- Scrum.org, Professional Scrum Master I assessment: https://www.scrum.org/assessments/professional-scrum-master-i-certification (opens in a new tab)
- Scrum.org, Introduction to Forecasting and Release Planning: https://www.scrum.org/resources/introduction-forecasting-and-release-planning (opens in a new tab)
- Scrum.org, Release Planning Techniques: https://www.scrum.org/resources/release-planning-techniques (opens in a new tab)
