calendar
Edited on October 7, 2026

ADDIE Model: The 5 Phases, Real Examples and Templates

The ADDIE model explained: the five phases worked through one course, where ADDIE really came from, whether it is outdated, real examples and templates.

Carolina Martin, Learning Experience Manager at Coursebox AI
Carolina Martin
Customer Success Manager
Key takeaways

ADDIE's five phases are Analysis (find the real problem), Design (plan objectives and assessments), Development (build and test), Implementation (roll it out) and Evaluation (check it worked).

  • A 1975 military instructional-design process was an important precursor; ADDIE later became an umbrella term for five design functions.
  • ADDIE can be used iteratively. Formative evaluation informs every phase and can send the team back to earlier decisions.
  • Choose deliverables and review points to suit the project. A formal sign-off after every phase is optional governance.
  • ADDIE, SAM and agile can be combined: use the design functions, prototype uncertain choices and organise feedback in short cycles.
Check your ADDIE planning priorities

Five prompts to guide a review of one project. This is an unvalidated editorial check, not a certification of model fit, completeness or learning effectiveness. Check each answer as well as the total.

1
2
3
4
5
Did you agree the business problem and a measure of success before designing anything?
Instructional Designers

Go from storyboard to an interactive course in minutes

Take a hypothetical hospital with five buildings and a fire warden in every ward. After a drill, the safety manager finds that wardens in two buildings do not know their evacuation route, and few wardens know how to move patients who cannot walk. The quick fix is a slide deck sent to everyone. The ADDIE fix starts with a question: what exactly do wardens need to do differently, in which building, and how will we know they can?

The ADDIE model is a five-phase process for designing training: Analysis, Design, Development, Implementation and Evaluation. I'm an instructional designer at Coursebox, where we make instructional design software, and most of the training projects I see follow some version of these five phases, whether the team calls it ADDIE or not.

What is the ADDIE model?

ADDIE is a systematic process for creating training, in five phases. Each phase answers one question and produces something the next phase depends on.

ADDIE analysis, design, development and implementation arranged around evaluation, with evaluation connected to every function and a recurring outer path showing revision.
PhaseThe questionTypical deliverable
AnalysisWhat must people do differently, and why aren't they doing it?Needs analysis summary and source material
DesignWhat will they learn, how will it be assessed, and in what order?Objectives, assessment plan and outline
DevelopmentIs the training built, accurate and tested?Built course, reviewed by a subject-matter expert and piloted
ImplementationWho takes it, when, and with what support?Rollout plan, enrolments and communications
EvaluationDid people learn it, use it, and did the problem improve?Evaluation report and list of changes
Clinical trainee working with a manikin.
Sahil Singh / Pexels

Case study: Knowledge and practical assessment in nursing training

A southern Indian teaching hospital used ADDIE to develop paediatric COVID-care training for 92 nursing officers. Mean knowledge-test scores rose from 10.3 to 14.1 out of 20. The study also used an OSCE, a practical clinical assessment.

The researchers did not find a statistically significant correlation between the post-test and OSCE scores. That finding does not prove that quiz performance can never relate to skill, or tell us what any individual nurse could do. It does show why this evaluation needed both knowledge and practical evidence.

For your plan, map each objective to the assessment it needs. A knowledge quiz can check knowledge; a practical objective needs observed performance. This was one programme, not a comparison establishing ADDIE’s superiority.

Michael Molenda describes ADDIE as "merely a colloquial term used to describe a systematic approach to instructional development" (Performance Improvement, 2003). That is the most useful way to think about it: less a strict model than the shared shape of most training projects.

ADDIE project plan template

Record the business problem and success measure, then plan each phase's key questions, deliverable, owner and sign-off, with the fire warden example. Editable Word file, no email required.

Where ADDIE came from

ADDIE is usually credited to Florida State University in 1975. That is close, but not quite right.

In 1975, Florida State's Center for Educational Technology delivered the Interservice Procedures for Instructional Systems Development (IPISD) for the US Army, later broadened for the Air Force, Navy and Marines. Its executive summary lists "five major phases, which are; analyze, design, develop, implement, and control." The fifth phase was control, not evaluate.

Molenda looked for an original ADDIE model and found none: "There is no original, fully elaborated model, just an umbrella term that refers to a family of models". He dates the term to around the late 1980s, when it became the common name for the instructional systems design process.

The 1975 version was not strictly linear either. Its fifth phase says collected data "can be used as quality control on instruction and as input to any phase of the system for revision." In my view, the linear, waterfall ADDIE that critics attack is a later simplification.

The five phases, worked through one programme

Our hypothetical hospital has five buildings and wardens who need to carry out their assigned fire-response duties under locally approved procedures. A drill found gaps in evacuation-route knowledge and assisting patients who cannot walk. The programme will combine instruction with supervised practice and observed checks of those duties. It is a useful ADDIE example because the drill findings must shape the objectives, materials, delivery arrangements and evaluation, rather than lead straight to a slide deck.

Analysis

Analysis defines the problem before anyone builds anything. For the hospital, that means walking each building with its warden, reviewing drill reports and talking to ward managers.

  • What it found: the gap is building-specific. Wardens in two buildings do not know their route, and few wardens know how to move a patient who cannot walk. Night shifts have fewer staff.
  • The decision: training covers building-specific routes and patient movement. A route map on every ward door, which is not training, fixes part of the problem without a course.
  • What to collect: the hospital's fire procedures, building plans and drill reports. These become the source material for development.

Collecting source material early shows up in our own data.

about 2.3x
observed publication-rate ratio: 19.6% among creators whose prompt included an upload versus 8.5% among prompt-only creators; this is a rate comparison, not a count or causal effect
Source: Coursebox product data, 5,708 creators who started a course between 6 and 28 September 2026, publishes tracked to 5 October, excluding internal accounts and bots

That is a correlation, not proof, but it matches what I see. In our support inbox, a children's English academy building a 16-week course about a novel got the result it wanted once it gave the builder a chapter-by-chapter summary of the book to work from. Analysis is where you collect that source.

Design

Design turns the analysis into a plan: objectives, assessments, structure and activities. For the hospital:

  • Objective: given a fire alarm in their building, the warden directs everyone in their zone to the correct exit and starts horizontal evacuation of patients who cannot walk, within the time set by the hospital's fire plan.
  • Assessment: a timed drill in each building, observed against a checklist, plus a short knowledge check on alarm codes.
  • Structure: a 30-minute online module per building, then a supervised drill.

For this hospital example, the safety manager approves the design before development. That is the project’s governance choice. ADDIE itself does not require a formal sign-off after every phase; choose review points that match the risk and keep formative evaluation active throughout.

Development

Development builds and tests the training. The modules are written from the hospital's own procedures, reviewed by the fire safety officer for accuracy, and tried by a few wardens before rollout.

This is the phase AI course builders change most. In my experience, a first draft built from the source documents now takes hours rather than weeks, which leaves more time for the review and pilot that catch mistakes. The subject-matter expert review still cannot be skipped.

Implementation

Implementation delivers the training. For the hospital: enrol every warden, tell their managers what to support afterwards, schedule the drills per building and shift, and name someone to handle questions during rollout.

Evaluation

Evaluation checks whether it worked, during the project (formative) and after launch (summative). The hospital checks whether wardens pass the timed drill, whether drill times improve over the next two quarters, and whether the route problems from the analysis are gone. The Kirkpatrick Model gives a common structure for the summative part.

ADDIE phase-gate checklist

Optional go or no-go review points for projects that need phase gates, from confirming the training need to planning evaluation. ADDIE does not require these gates. Editable Word file, no email required.

Healthcare colleagues talking at a clinical reception desk.
RDNE Stock project / Pexels

Case study: Retraining specialist nurses at a Zhejiang hospital in China

A Zhejiang hospital used ADDIE to develop retraining for 36 peritoneal dialysis specialist nurses, with input from 22 experts. The published evaluation reports higher examination scores and increases in independently operated nursing clinics, alongside improved patient indicators.

Peritonitis and catheter-related infection rates were reported as lower after the programme. There was no control group, so changes in patient care cannot be attributed to ADDIE alone. The programme also included support for putting the nurses’ new capabilities to use: implementation was part of the intervention alongside the course.

13 to 33
independent nurse-run dialysis clinics at a Zhejiang hospital, before and after an ADDIE-built retraining programme for 36 specialist nurses (2023 to 2025)
Source: First Affiliated Hospital of Zhejiang University School of Medicine, BMC Medical Education, 2026

Is the ADDIE model outdated?

The criticism is real. In 2000 a Training magazine article questioned whether instructional systems design was still relevant, and in the debate that followed, one critic called it "too generic and simplistic to be really helpful for the detailed design of instruction," while another said it "is the best thing we have, if we use it correctly" (Training, 2002).

In my experience, most complaints are about ADDIE done badly: months of analysis, a design document nobody reads, and evaluation that never happens. As ATD's Sophie Oberstein puts it, "ADDIE may be illustrated in a linear manner, but it is not meant to be performed that way" (ATD, 2022).

It is still in formal use. The Chicago Police Department's draft 2025 training plan notes that the Independent Monitoring Team, which oversees the department under a federal consent decree, "utilizes the ADDIE model as an evaluation standard when monitoring Department training development" (Chicago Police Department, 2025 training plan), and the UK Ministry of Defence runs an ADDIE-family process it calls the Defence Systems Approach to Training.

Colleagues listening to a presenter beside a flip chart.
Gustavo Fring / Pexels

Case study: Cybersecurity awareness at a marine manufacturer

Niroj Chaulagain's Oceanvolt training project used ADDIE and the NIST Cybersecurity Framework to build an awareness session for employees across a marine manufacturer's departments. Fourteen staff answered the baseline survey; sixteen attended the session and thirteen answered the post-training survey.

The needs analysis found gaps in governance, recovery procedures and policy awareness. The session combined brief explanations, controlled demonstrations and discussion, with an internal wiki for follow-up. Post-training respondents rated relevance and confidence highly, and feedback requested more company-specific examples.

This is a small thesis case, not a controlled demonstration of fewer security incidents. Its value for an ADDIE plan is the connection from diagnosed gaps to training choices and follow-up support. Confidence ratings alone would not establish that staff can recognise or handle a real incident.

From analysis to course in Coursebox

ADDIE describes the work the training team does rather than the order learners study in. On 6 October 2026 I tested what happens when you hand the results of that work to Coursebox, the course builder we make. I gave it the fire warden example as a designer would have it after Analysis and Design: the drill findings, the objective, the planned structure, and the hospital's fire and evacuation procedure covering each building's exits, the alarm codes, the RACE warden actions (remove, alert, contain, extinguish or evacuate) and where the evacuation sleds are kept.

It built the course the design called for: 8 sections, 14 pages and about 54 minutes. A shared module on alarm codes and RACE comes first, then one module per building, then night-shift readiness and the supervised evacuation drill. Each building module used that building's own procedure. Building A covers acute-care evacuation, with a page on the evacuation sled at the nurses' station and a night-shift evacuation decision; Building C, where most people can walk, has a scenario where smoke blocks the main foyer. Every building ends with a short response check.

Coursebox fire warden outline built from analysis findings and the hospital procedure: a shared alarm and RACE module, then a module for each building with practice and a check

The analysis findings shaped what the course teaches and the procedure supplied the facts, so the draft was specific to this hospital from the first version. That is how the Development phase should work. My review then has a clear job. I check each building's routes against the fire plan with the safety manager, then pilot one building module with wardens before rolling out the rest.

ADDIE vs SAM vs Agile

ApproachHow it worksBest for
ADDIEFive design functions, with evaluation and revision throughout; teams choose their approval pointsA shared structure for planning training; usable with iterative development and formal approvals
SAMShort design and build rounds starting from a rough prototypeUnclear requirements and reviewers who can meet often
AgileFrequent feedback and usable increments; learner release follows required checksContent that changes often, such as product training

They are not either-or. In my projects, I often run ADDIE's analysis and evaluation around SAM-style prototyping in the middle, which is close to what rapid prototyping describes. If you plan individual lessons, the ASSURE model fits inside an ADDIE programme. Agile instructional design adds a way to organise that work around frequent feedback and usable increments.

Limitations of ADDIE

  • It can be slow and document-heavy when every phase waits for full sign-off before the next starts.
  • It describes phases, not methods. ADDIE tells you to design, not how; you still need models such as Gagné's nine events or Merrill's principles for that.
  • Evaluation is often cut. Many projects stop at a satisfaction survey and a quiz, which says little about whether people can do the job.
  • Late changes are expensive if the team treats design as frozen.

Use ADDIE to connect design decisions

ADDIE gives a team a shared structure for understanding the problem, designing the learning, building it, supporting delivery and evaluating the result. It is useful when those decisions remain connected and new evidence can change earlier work. Treating every phase as a document to approve once can make the process slow while leaving the real performance problem untouched.

Begin the project plan with the task people need to perform and the evidence that would show improvement. Align objectives, practice and assessment, choose review points that match the risk and plan implementation support before launch. Keep evaluation active throughout. If the pilot exposes a wrong assumption, revisit it while the design is still changeable. The hospital example needs locally approved procedures and observed performance; a finished course file is only one part of that job.

Frequently Asked Questions

Analysis, Design, Development, Implementation and Evaluation. Analysis defines the problem, design plans the objectives and assessments, development builds and tests the training, implementation delivers it, and evaluation checks whether it worked.

No single person is credited with the acronym. The five-phase process comes from Florida State University's 1975 Interservice Procedures for Instructional Systems Development for the US armed services, which named its last phase "control". The ADDIE name spread informally from the late 1980s.

Analysis, design, development, implementation and evaluation are connected design functions. Teams often start with analysis and revisit earlier work as evidence changes. Formative evaluation belongs throughout. A formal sign-off at each phase is a project governance choice.

ADDIE describes five connected design functions; SAM provides a more explicit structure for iterative design and development. Both can use early prototypes and repeated evaluation. Formal sign-off after every ADDIE phase is a project choice, not an inherent difference between the models.

Yes. It remains a common framework in corporate, healthcare, government and military training, often combined with faster prototyping in the development phase.

Carolina Martin, Learning Experience Manager at Coursebox AI

Carolina Martin

Customer Success Manager

Carolina is a highly experienced learning designer with tech start-up experience. She is based in Sydney and previously designed courses alongside Coursebox founder Travis in a previous job. She is passionate about delivering engaging learning experiences and brings the same perspective to customer experience too. She manages our customer support documentation and product testing processes.