The model
ADDIE is the backbone framework of the whole course: Analyse, Design, Develop, Implement, Evaluate. It has real pedigree — it was developed in 1975 at Florida State University's Center for Educational Technology for the U.S. Army, and it has survived fifty years of instructional-design fashion because it is less a recipe than an operating system. The workbook's diagram is drawn with Evaluate in the centre, arrows running both ways to every other phase — evaluation is not the last step, it is the hub that every phase reports into and learns from. Analyse: identify the target audience, their needs and requirements, and the competency gap — the distance between what people can do today and what the job requires. Design: set learning outcomes, select training and assessment methods, identify learning resources. Develop: create a prototype, develop the course materials, run a try-out session, refine the draft. Implement: facilitate the training and engage the participants. Evaluate: conduct formative evaluation (while the programme is being built and delivered) and summative evaluation (after, to judge whether it worked).
What each phase is really for
Analyse is the phase everyone skips. Most corporate training starts with a solution — "we need a communication workshop" — instead of a competency gap. Without analysis you cannot even say what success looks like, because you never named what was missing. Design is where learning outcomes get written, and outcomes discipline everything downstream: the methods, the assessments, even the room layout. Our module's own outcomes are a working example — after this module I should be able to define learning and training, relate adult learning principles to delivery, and reflect on my own practice as a trainer. Observable verbs: define, relate, reflect. Develop surprised me by including a try-out session and refine the draft. Training materials deserve the same treatment as software: prototype, test on a small audience, iterate before shipping. Implement is delivery — but the workbook reduces it to two essentials, facilitate and engage, which is a quiet argument that a trainer is a facilitator of other people's learning, not a presenter of their own slides. And Evaluate runs formative (is this working while I build and deliver it?) and summative (did behaviour change afterwards?).
Case Study
designing a service-recovery workshop with ADDIE
To make ADDIE concrete I sketched how I would build a real programme for my own team: service recovery for front-line staff across our suites — what to do when a guest arrives at 11pm and the room is not ready. Analyse: pull the last 90 days of complaints and OTA reviews. The gap is not product knowledge — staff know the policies. The gap is confident recovery under pressure: apologising without admitting fault wrongly, offering the right compensation tier, escalating at the right threshold. Audience: front desk and duty managers, mixed experience levels, most of whom learned by imitation, never by design. Design: outcomes such as "given an overbooking scenario, the staff member can execute the 4-step recovery script and choose the correct compensation tier within 5 minutes". Methods: short demo, then role-play drills with rotating guest personas — not a lecture. Assessment: observed role-play against a checklist, not a quiz. Develop: write three scenario cards from real incidents (names removed), prototype the session with one property's team as a try-out, refine the compensation matrix after their feedback — they will find edge cases HQ never imagined. Implement: run it property by property in 90-minute sessions during shift overlap, trainer facilitating drills rather than presenting: brief, drill, debrief, repeat. Evaluate: formative — did participants pass the observed role-play? Summative — 60 days later, compare complaint-resolution ratings and compensation cost per incident against the baseline from Analyse. The baseline exists precisely because we did Analyse first; skip it and there is nothing to measure against.
Reflections
ADDIE is not a waterfall, it is a loop with Evaluate in the middle — every phase both feeds and consumes evaluation. My instinct as a builder is to jump straight to Develop because materials feel like progress. Today's discipline: no slides until the competency gap is written down and the outcome sentence passes the observability test. If I cannot say what the learner will be able to do, I am not ready to open the slide editor.