Iterative and incremental delivery
8 min read
Lesson objectives
- The learner chooses a small usable Increment that exposes a value risk and distinguishes it from splitting phases without an inspectable result.
Please sign in to use all the tools
Once you sign in, the reading tools are yours: larger or smaller text, typeface, line spacing, focus mode, highlighting and bold, and notes saved to your account and gathered in My study.
Iterative and incremental delivery
A volunteer organization wants people to book shifts themselves. A team proposes a database in its first Sprint, interfaces in the second and testing in the third. This divides work, but postpones a real booking experience until the end. In Scrum, the team brings together what is needed for a usable Increment during a Sprint. The first Increment might support only one kind of shift or a small volunteer group, but it includes the necessary behavior and agreed quality. ‘Incremental’ means the product grows through integrated additions, not that each specialty hands over a technical fragment that cannot work alone. ‘Iterative’ means the team revisits its understanding and design based on learning from each addition.
A good choice begins with an explicit risk. If the main question is whether a volunteer can select a shift without a coordinator, one path that displays available shifts, confirms a booking and prevents duplicates may be enough. The first Increment does not need every future report. It does need a clear standard for Done. That may include permission checks, conflict prevention and transaction records under the product's Definition of Done. Leaving those out while calling the screen usable creates false learning; the user has not tested booking as it would actually work.
During inspection, the team asks more than how many features it added. Did the Increment reduce decision uncertainty? How many volunteers tried to book, where did they struggle, were the listed shifts accurate, and did coordinator workload fall or rise? Inspection may reveal that the problem is not the interface but the policy for publishing shifts. The Product Owner can then reorder the Product Backlog in light of discovered value. This does not require abandoning the Product Goal whenever new information appears. The goal may remain while the approach and ordering of work change.
Do not confuse ‘releasable’ with ‘released.’ An Increment that meets the Definition of Done supports a release decision when appropriate; the Sprint Review is not a mandatory release gate. A team may produce several Increments in one Sprint, each adding to prior product state after integration. If a fix is usable in the morning, it need not wait for the end-of-Sprint presentation to become an Increment. Conversely, an approaching presentation cannot make unfinished work Done. This distinction teaches that product value is not measured by the calendar alone, but by the product's condition and effects that can be inspected.
When slicing work, start from a user journey that can be completed and then narrow it without breaking its meaning. You can limit the type of shift, region or number of users. You cannot omit booking confirmation if its absence prevents the user from knowing the outcome. ‘Smallest’ depends on the risk under examination, not the number of screens. If the risk is whether the system prevents conflicting bookings, testing only the button appearance will not answer it. If the risk is whether volunteers understand the labels, a limited trial with real users after building a minimal usable flow may suffice.
Readiness questions often offer a choice to build everything before inspection, or ask the Scrum Master to dictate how Developers split work. Check whether the option creates a Done Increment and early inspection of a specified value or risk while respecting accountabilities. An empirical approach does not guarantee that the first design is right. Its strength is that a mistake can be discovered at a lower cost and early enough to change direction. Each Sprint should therefore be connected to a learning opportunity and a meaningful goal, rather than just a collection of disconnected tasks.
When slicing a returns service, follow value from the customer's request to knowing its outcome. Limit a trial to one product and purchase method while keeping approval, refund and confirmation integrated. This can expose a specific risk, such as two refund requests for the same transaction. Describe success and failure behavior, then inspect dependencies on the rest of the product. If a customer needs a manual call for an essential step, make that boundary explicit and check whether it meets the intended purpose. Manual assistance can be an openly designed part of a trial, or contradict a self-service Goal. The Goal and agreed boundary determine which. After completing a slice, extend product type or payment method. Check integration with prior behavior because the second route may reintroduce a defect repaired in the first. Test records should show both retained and added behavior.
Lesson summary
The learner chooses a small usable Increment that exposes a value risk and distinguishes it from splitting phases without an inspectable result.