Delivery Roadmap: Milestones, Dependencies, Risks
8 min read ยท 2026-11-20
A delivery roadmap is a time-based plan that shows what your team will ship, in what order, which milestones mark each step, and what each piece depends on. To build one that holds, start from the approved scope, map the dependencies, put the work in order across phases, then give every date an owner, its risks and a confidence level.
This guide is for project managers, PMOs, consultants and product leads who must hand a delivery plan to a steering group or a client. It starts with the four reasons these plans go off track, then gives a complete software implementation example you can copy phase by phase, with the fix for each problem already built in.
The roadmap at a glance
Goal: Take an approved software implementation from signed scope to a stable go live, with every dependency and risk visible to leadership. Duration: 16 weeks
Scope and baseline (Weeks 1-2)
Lock what is being delivered and what is not, so every later date has a fixed reference point.
- Confirm the business objectives with the sponsor and write them as two or three outcomes.
- List in-scope and out-of-scope items, and get both signed off in writing.
- Name an owner for each deliverable and one decision maker for scope changes.
- Record known constraints: budget window, blackout dates, contract deadlines.
Milestone: Scope document signed by the sponsor.
Dependency map and sequencing (Week 3)
Find out what blocks what before anyone commits to a date.
- List internal dependencies, such as configuration before testing and data cleanup before migration.
- List external dependencies: vendor environments, security review, access requests, other teams' releases.
- Give each dependency an owner and a "needed by" date.
- Find the longest chain of dependent tasks and mark it as the critical path.
Milestone: Dependency map reviewed with every owner named on it.
Configure and build (Weeks 4-9)
Deliver the core system in short increments the business can see and react to.
- Split the build into two-week sprints, each ending with a demo.
- Configure the riskiest workflows first, while there is still time to change course.
- Keep a decision log with a due date for every open question.
- Review the confidence level of later dates at the end of each sprint.
Milestone: Core workflows demonstrated and accepted by key users.
Data migration and integrations (Weeks 7-11)
Move clean data and connect the systems the new tool relies on. This phase overlaps the build on purpose.
- Clean and map the source data with the data owner.
- Run trial migrations and reconcile record counts against the source system.
- Build and test each integration against a real test environment.
- Agree the data freeze window for cutover.
Milestone: Full trial migration reconciled and signed off by the data owner.
Testing and training (Weeks 11-14)
Prove the system works for real tasks and that people can use it on day one.
- Run user acceptance testing with scripts written from actual daily work.
- Sort defects into "must fix before go live" and "fix after go live".
- Train by role, starting with super users who will support their colleagues.
- Write the cutover runbook and the rollback plan.
Milestone: Acceptance testing signed off and the go or no-go decision recorded.
Go live and hypercare (Weeks 15-16)
Switch over safely and keep close support in place until the system is stable.
- Run the cutover step by step from the runbook.
- Hold a short daily check-in with support and super users.
- Log every issue with an owner and a target date.
- Hand over to the regular support team with up-to-date documentation.
Milestone: Handover accepted by the support team, with every open issue owned and dated.
Why a delivery roadmap goes off track
Most delivery plans look fine on the day they are presented. They fail later, and nearly always for one of four reasons.
Keep these four in mind as you read the example. Each phase is written to block at least one of them.
- Dates without dependencies. Each phase has a date, but nothing shows what that date waits for. When a vendor environment arrives three weeks late, the whole plan moves and nobody saw it coming. The fix is in the dependency map phase and in the "Dependencies and confidence levels" section below.
- Milestones nobody can verify. "Migration progressing" or "testing ongoing" cannot be confirmed or denied, so the plan looks green until the week it turns red. The fix is the yes or no test in step 6 of the build steps, and the milestone wording in every phase of the example.
- Risks discovered too late. The team finds out about poor data quality or a security review during cutover week, when there is no room left to react. The fix is to put the riskiest work first, as the build phase does, and to agree risk responses before the first presentation.
- A plan that is never updated. The roadmap shown at kickoff stays in a slide deck while the real work moves on. Within a month, people are working from different versions. The fix is in "Keeping the roadmap current" below.
What a delivery roadmap must contain
A delivery plan takes approved requirements and shows how they reach production: which releases, in what order, blocked by what, owned by whom. If you still need to settle the direction of the product itself, do that first. The delivery view sits underneath it.
A complete delivery plan holds these items:
- Objectives and scope, including what is out of scope
- Deliverables and the release structure
- A timeline with milestones and dependencies
- Resources and the people assigned
- A risk register with agreed responses
- Governance roles: who decides, who approves, who is informed
- A review rhythm with leadership, the team and the client
How to build it in 7 steps
The order matters more than the tool. Each step feeds the next, and skipping one usually shows up later as a missed date.
A good test for step 6: every milestone should be a fact someone can confirm with a yes or a no. "Migration progressing" fails the test. "Trial migration reconciled and signed off" passes.
- Step 1: clarify the objectives, written as outcomes the sponsor can check.
- Step 2: define the scope, with a clear out-of-scope list.
- Step 3: identify dependencies, internal and external, each with an owner.
- Step 4: sequence the work so blocked tasks never start before what they wait for.
- Step 5: estimate effort with the people who will do the work, not for them.
- Step 6: build the timeline, placing milestones where something can be verified.
- Step 7: agree risk responses before the plan is presented, not after the first slip.
Dependencies and confidence levels on a delivery roadmap
Most templates stop at phases and dates. Two extra layers make the plan honest: dependency labels and confidence levels.
Label each dependency where it bites. In the example, integrations cannot be tested until the vendor test environment exists, so write "Blocked by: vendor test environment, owner IT, needed by week 6" directly on the integration phase. The data freeze depends on the business agreeing a quiet period, so it carries the sponsor as owner. When a reader sees a date, they also see what it waits for.
Then give each date a confidence level: high, medium or low. In week 3 of the example, the scope sign-off and build start are high, migration is medium because the data quality is still unknown, and go live is low until the trial migration is reconciled. Raise the levels as milestones close. This turns the roadmap into a forecast instead of a promise, and it tells leadership exactly which risk to help with.
Delivery roadmap in agile teams
In an agile team, sprints do the work and the delivery plan shows where they lead. Each phase holds a set of sprints, and each milestone usually maps to a release or a demo. Keep sprint tasks in the backlog tool. The roadmap only needs the release level, or it becomes unreadable for anyone outside the team.
Teams that ship continuously do better with a roadmap people can open at any time, plus short regular updates on what has shipped, than with one big monthly presentation.
Keeping the roadmap current
A delivery roadmap loses value the moment it stops matching reality. Set a fixed review point, such as the end of every sprint or every milestone, and change three things each time: confidence levels, dependency status and the top risks.
The GOV.UK Service Manual gives a useful benchmark from government delivery teams. It describes a roadmap as "a plan that shows how a product or service is likely to develop over time", says that at the Government Digital Service roadmaps are iterated on at least a quarterly basis, usually led by the product manager, and asks that the roadmap be available for other civil servants to view. A delivery plan moves faster than that, but the habit is the same: a regular update, one clear owner and a version anyone can read.
Presenting dates to leadership without promising them
Leadership and clients want dates. Your job is to give them dates with honest uncertainty attached. Two techniques work well together.
First, use horizons. Put committed work in "Now" with specific weeks, planned work in "Next" with a month or a quarter, and options in "Later" with no date at all. Second, show the confidence level next to every date in Now and Next. A sponsor who sees "Go live, week 15, low confidence, waiting on trial migration" knows what to ask and where to help.
For dependencies and risks, keep it to one view. Draw arrows only for the dependencies on the critical path, and list the top three risks under the timeline with owner and response. Send a fixed version for the steering pack and keep a live version the team updates after each milestone, so nobody works from an old copy.
Common mistakes to avoid
- Hiding external dependencies inside phase names makes them invisible, so label each one with its owner and "needed by" date.
- Writing milestones as activities ("testing ongoing") makes progress impossible to verify, so phrase each one as a result someone can sign off.
- Leaving risk responses until something slips forces rushed decisions at cutover, so agree the response for each top risk before the first presentation.
Frequently asked questions
How is a delivery roadmap different from a product roadmap?
A product roadmap shows the direction of a product and the problems it will solve over time. A delivery roadmap shows how approved work reaches production: releases, sequence, milestones, dependencies and owners. Most teams keep both, with the delivery view sitting under the product view.
What should a delivery plan include?
At minimum: objectives and scope, deliverables, a timeline with milestones and dependencies, the release structure, resources, a risk register, governance roles and a review rhythm. If one of these is missing, someone will ask for it at the first steering meeting.
How do you show dependencies on a roadmap?
Draw arrows only for dependencies on the critical path, and write the others as short labels on the phase they block, with an owner and a "needed by" date. That keeps the view readable for leadership while still exposing every blocker.
How often should a delivery roadmap be updated?
At every milestone or sprint review, at least. Each update should change the confidence levels, the dependency status and the top risks, and show leadership what moved since the last version.