ADDIE Model
Analysis · Design · Development · Implementation · Evaluation / Instructional Systems Design
What is the ADDIE model?
- Five stages: Analysis, Design, Development, Implementation, Evaluation
- The standard process in instructional systems design (ISD)
- The backbone of systematic training development
Using ADDIE isn’t about drawing a prettier training schedule. It changes the first question a manager asks when a training request comes in. Instead of “Which instructor should we bring in?”, you start by asking “Is the cause of our performance gap something training can actually solve?” The biggest difference you feel in practice is that every stage leaves a visible deliverable. As the needs analysis report, learning objectives and session plan, lesson plans and practice materials, operations checklist, and evaluation report accumulate in turn, you can review the entire program and run it again, even when the person in charge changes.
The five stages don’t carry equal weight. When building an in-house program, you might allot 2 weeks to analysis, 1 week to design, and 3–4 weeks to development. Putting most of your effort into the early stages, where changes get more expensive the later they happen, saves trouble down the line. Evaluation (E) also doesn’t mean a single satisfaction survey at the end. It splits into formative evaluation, which checks at the end of each stage whether it’s safe to move on, and summative evaluation, which looks at results after the program ends. The Kirkpatrick four levels answer “what do we measure?” for that summative evaluation; they are not a replacement for ADDIE. SAM, which prioritizes rapid prototyping and iteration, is a better fit for programs with frequently changing requirements and short development cycles.
The most common misconception is treating ADDIE as a straight line you pass through once. If the analysis stage relies on the opinions of two or three team leaders instead of interviews with the people doing the work, you only hear “this doesn’t match our work” after development is finished, and all the effort on lesson plans is wasted. Following the procedure only on paper causes the same problem. If you write a 30-page needs analysis report but the learning objectives say “understand” or “be aware of,” there is nothing to measure at the evaluation stage. The conclusion of analysis should be one line stating a measurable behavioral objective, not a page count, and if that one line doesn’t emerge, you aren’t ready to move on to design yet.
Kyungtae KimHR expert · Organizational culture and HR strategy consultant
John KimAuthor of B2B Marketing Design and CEO of Nanum ConsultingIt originated in the Instructional Systems Development procedure (IPISD) that Florida State University’s Center for Educational Technology developed for the U.S. Army in 1975, and later spread across all U.S. military branches and into industry.
Here is a hypothetical case. Company A, an auto parts manufacturer with 620 employees, saw customer quality claims climb to an average of 14 per month last year. Management ordered, “Retrain the inspectors,” but the training team took the 8-week window and used ADDIE to re-examine the root cause first.
- A manager on the training team followed the 22 inspectors across 3 lines for two days and broke down the 14 complaints. Nine of them came from people applying the appearance-judgment criteria differently. The other 5 were caused by aging equipment, so we excluded them from the training scope.
- The head of production insisted, “We can only stop the line for 4 hours,” so the training manager narrowed the learning objective to match pass/fail judgments at 90%+ agreement across 20 reference photos and designed a two-session, 4-hour program.
- We photographed 40 real returned-product samples and turned them into judgment cards. A quality-team manager reviewing them told us, “We disagree on these 6 types every time,” so instead of giving a single right answer, we built those 6 into separate discussion cases.
- We piloted with 7 inspectors on Line 1 first, but practice agreement reached only 62%. We added 30 minutes to the criteria explanation, moved the 6 confusing types to the very front, and then rolled it out to the other 2 lines.
- Three months later, judgment agreement among the 22 inspectors rose from 62% to 91%, and client complaints fell from 14 to 5 per month. Company A registered this judgment-card program as a required onboarding course for new inspectors.