ADDIE Model Explained – Five Phases with Real Examples
ADDIE is the framework most training teams say they follow. It shows up in proposals, statements of work and project plans. But knowing the five letters is not the same as knowing what each phase looks like on a real project, what it should produce and where it tends to go wrong. How does the ADDIE model actually work, phase by phase, when you apply it to real training?
The ADDIE model is a five-phase framework for building training: Analysis (define the performance gap and audience), Design (plan objectives, assessments and strategy), Development (build and test the materials), Implementation (deliver to learners) and Evaluation (measure results and improve). Each phase produces specific outputs that the next phase depends on, and evaluation runs throughout rather than only at the end. ADDIE is not a strict, one-way sequence. Used well, it is a disciplined loop that keeps teams from building the wrong thing. The examples below show what each phase looks like in practice.
Where Does the ADDIE Model Come From?
ADDIE grew out of instructional systems design work for the U.S. military in the 1970s, most notably a model Florida State University developed for the armed services in 1975. The acronym itself seems to have come into common use later, and nobody can point to a single author. You can read more on the history of the ADDIE model if you are curious about how it evolved.
One historical point matters for how you use it. Early versions included feedback loops and revision, so the popular image of ADDIE as a rigid waterfall is more a story about how organizations applied it than about the model itself.
Phase 1: Analysis
Analysis answers one question: what problem are we solving, and is training the right way to solve it? You look at the gap between current and desired performance, the people involved, the tasks they perform and the constraints you will work within.
Typical analysis activities include interviews with managers and top performers, observation on the job, a review of existing data (error reports, customer complaints, audit findings) and a quick inventory of available time, budget and technology.
Example: A New Software Rollout
Suppose a company is moving its sales team to a new customer relationship management (CRM) system. The initial request is "a course on the new CRM." Analysis reveals that the team only uses about a fifth of the features, that most worry centers on logging calls and updating opportunities, and that reps have no more than 20 minutes at a time between appointments. Those findings reshape the whole project: short, task-focused lessons on the handful of things reps actually do.
Common mistake: skipping analysis because "we already know what they need." Teams that skip it often build comprehensive courses on features nobody uses.
Phase 2: Design
Design creates the blueprint. Here you write measurable learning objectives, plan how you will assess them, choose instructional strategies and decide on format and sequence. Design is where the analysis findings turn into decisions.
Design is also where learning research earns its place. Decades of studies on retrieval practice and spaced practice point the same way: people remember more when they recall information and use it several times, spread out, than when they read it once. A design that builds in practice and spaced follow-up will usually beat one that front-loads all the content into a single session.
The output is usually a design document or detailed outline that stakeholders review and approve. That sign-off matters because changes are cheap now and expensive later.
Example: Safety Procedure Training
Picture a manufacturer updating its lockout procedure for equipment maintenance. The design team writes objectives such as "Apply the lockout steps in the correct order on a training panel, with no missed steps." Because the task is physical and high-stakes, the design calls for a short eLearning module to introduce the steps, followed by hands-on practice with a supervisor and a performance checklist as the assessment. A multiple-choice quiz alone would not show whether someone can do it.
Common mistake: writing objectives with verbs like "understand" or "be aware of." If you cannot observe it, you cannot assess it, and the rest of the design loses its anchor.
Phase 3: Development
Development is where the materials get built: storyboards, scripts, graphics, video, interactions, eLearning modules, facilitator guides, job aids and assessments. It is often the longest phase and the one people picture when they think of course creation.
Good development includes testing along the way. Build a rough prototype of one lesson, show it to a few real learners and fix what confuses them before you produce the rest.
Example: Customer Service Onboarding
A contact center is building onboarding for new agents. The team storyboards three branching call scenarios, then builds the first one as a rough prototype with placeholder audio. Five new hires try it. Two get stuck on the same screen because the choices are worded too similarly. The team fixes the wording pattern before building the other two scenarios, which saves rework later.
Common mistake: treating stakeholder review as the only test. Reviewers know the content; learners don't. Only learners can show you where the design breaks.
Phase 4: Implementation
Implementation gets the training to learners and makes sure they can complete it. That covers technical setup, such as uploading courses to the learning management system (LMS) and testing tracking, plus the human side: scheduling, facilitator preparation and communication with managers.
Implementation is often underestimated. A good course that nobody has time to take, or that managers don't support, will not change performance.
Example: A Regional Rollout
A retail chain launches new point-of-sale training across many stores. The team pilots it in a few locations first, finds that some store computers block audio, and adds captions plus a printed quick-reference card before the wider rollout. Store managers get a short brief explaining why the training matters and when staff should complete it.
Common mistake: launching everywhere at once with no pilot. Small technical problems become large support problems when they hit every location on the same day.
Phase 5: Evaluation
Evaluation asks whether the training worked and how to make it better. It has two forms. Formative evaluation happens during the project (reviews, prototypes, pilots). Summative evaluation happens after launch and looks at results: learner reactions, assessment scores, on-the-job behavior and business measures.
The most useful evaluation plans are written during design, not after launch. If you know you will compare error rates before and after, you can make sure that data is collected.
Example: Back to the CRM Rollout
For the CRM project, the team agreed in design to track the share of sales calls logged within a day. Two months after launch, logging improved for most reps but stayed low in one region. Follow-up showed that region's managers were still asking for updates by email, so reps saw no reason to change. The fix was a conversation with those managers, not more training.
Common mistake: stopping at satisfaction surveys. Learners can enjoy a course and still not change how they work.
Who Does What in Each ADDIE Phase?
ADDIE is easier to run when everyone knows their role. On a small team, one person may wear several hats. On a large project, these roles often belong to different people.
- The instructional designer leads analysis and design, writes objectives and assessments and keeps the project tied to the performance goal.
- The subject matter expert (SME) supplies accurate content, reviews drafts and helps write realistic scenarios.
- The eLearning developer or media producer builds the course, video and graphics during development.
- The project sponsor or business owner defines success, approves the design and removes obstacles during rollout.
- The LMS administrator and facilitators handle implementation, from tracking to live sessions.
- Managers support learners before and after training, which matters more for evaluation results than most teams expect.
A short kickoff that names these roles, with review dates for each phase, prevents the most common source of delay: waiting for feedback from someone who did not know it was their turn.
Can You Scale ADDIE Down for Small Projects?
Yes. ADDIE does not require a thick binder. For a short module or a job aid, each phase can be very light while still doing its job.
Analysis might be a 30-minute call with a manager and a quick look at the error that triggered the request. Design might be a one-page outline with three objectives and a note on how you will check them. Development might take a couple of days, with one learner test. Implementation might be an email from the manager with a link. Evaluation might be a check of the same error report a month later.
The point is not the size of the documents. It is that each question gets asked. A ten-minute module built on a clear objective will do more good than an hour-long course built on a guess.
How Do the ADDIE Phases Connect?
The phases connect through their outputs. This table summarizes what each phase hands to the next.
| Phase | Main output | Who uses it next |
|---|---|---|
| Analysis | Performance gap, audience profile, task list, constraints | Designer writing objectives and strategy |
| Design | Objectives, assessment plan, outline, approved design document | Developers and media producers |
| Development | Tested courses, guides, job aids, assessments | LMS administrators, facilitators |
| Implementation | Live training, completion data, rollout notes | Evaluators and stakeholders |
| Evaluation | Results, insights and a revision list | The whole team, for the next cycle |
Use the interactive overview below to step through each phase and see what it produces and where the key decisions sit.
Is ADDIE Still Relevant?
Yes. I'd argue it works best when you treat it as a set of questions rather than a set of forms. The phases describe work that has to happen on any serious training project. Whether you move through them once over several months or many times in short cycles is a separate choice.
That choice is the heart of the debate between ADDIE and more iterative approaches such as the Successive Approximation Model. If you are weighing the two, our comparison of ADDIE vs SAM walks through when each one fits.
For most teams, the practical answer is a blend: keep ADDIE's discipline around analysis and evaluation, and borrow iteration and early prototyping during design and development.
Making the ADDIE Model Work on Your Next Project
The ADDIE model gives you five phases, each with a clear job. Analysis finds the real problem. Design plans how people will learn and show it. Development builds and tests the materials. Implementation gets them to learners with the support they need. Evaluation checks whether performance changed and feeds the next round.
The examples above share a pattern. The biggest wins came from asking good questions early and testing with real learners, not from producing more content. That is the real value of ADDIE.
My advice for a first try: open a blank document, type the five phase names as headings and write two or three lines under each before the kickoff meeting. If any heading stays empty, that is the part of the project most likely to cause trouble later.