Theory First or Practice First? Sequencing a Skills Workshop

Thirty minutes. That's the entire theory block in our second Operation Aetherfall pilot – a seven-and-a-half-hour training day built around four hours of tabletop gameplay and a 90-minute debrief. The concepts fit on two reference cards. The session plan even includes an explicit instruction about what not to teach before play begins.

That ratio is what the sequencing research supports. In the theory-versus-practice question for adult training, the evidence favors practice first: learners who encounter a realistic problem or worked example before the concepts, then get concise targeted theory, outperform learners who sit through the full conceptual treatment up front. They do better on immediate transfer tasks, build schemas faster, and retain more over the long term. Theory-first feels rigorous. Practice-first works better.

I'm a product manager, not an instructional designer. Before Adam White and I designed a single scene of Operation Aetherfall, I synthesized more than 200 studies on instructional design, and sequencing turned out to be one of the clearest threads in the corpus – and one of the most ignored. Front-loading the frameworks is a real piece of why workplace training doesn't stick: the concepts arrive before anyone has a reason to need them.

Should you teach theory before practice?

Mostly, no. Research comparing the two orderings directly – worked examples followed by problem solving, versus theory followed by problem solving – finds that novices who study a worked example before the theory outperform those who get the full lecture first. Three results recur across this literature (Renkl, 2014; Atkinson et al., 2000): higher immediate performance on transfer tasks, faster schema acquisition with deeper conceptual understanding, and greater long-term retention when examples are interleaved with brief, targeted theory segments. The broader practice-versus-lecture evidence points the same direction.

Cognitive load theory explains the mechanism. Sweller, Ayres, and Kalyuga's work shows that novices get overloaded when abstract principles arrive without any context to hang them on – the literature names "all the theory up front" as a specific pitfall. Start with a concrete example instead and the load profile changes: existing knowledge activates, the coming theory gets grounded in a familiar situation, and the instructor only needs to introduce as much theory as the example requires.

David Merrill's First Principles of Instruction arrives at the same ordering from a different direction. His five-phase cycle runs activation (present a real task or problem), demonstration (show how experts handle it), application (guided attempts), integration (apply it in varied contexts) – and only then explanation, the explicit theory for why the procedures work. Explanation is phase five. Not phase one.

Why does struggling first beat concept-then-apply?

The struggle creates the question the theory answers. Problem-based learning is built on exactly this: learners meet a complex, realistic scenario first and identify their own gaps – that moment of realizing you don't know what to say next. Theory then arrives just-in-time to fill a gap the learner already knows they have. The timing does something a lecture can't. It makes the theory immediately useful, which recruits motivation instead of demanding it. Anchored instruction pushes the idea further, embedding the concepts inside the case itself so theory emerges while learners explore the problem rather than before.

Adults raise the stakes on this. Everyone who walks into a conflict workshop carries years of real disagreements – with bosses, peers, vendors, the occasional contractor who stopped answering emails. That history is the raw material activation works on. Teach the framework first and it has to compete with all of that experience in the abstract. Let people play through a tense negotiation first and the framework lands on a fresh, shared, specific example everyone at the table just watched happen.

What should you still teach before practice?

Safety, a little vocabulary, and the mechanics of the exercise – almost nothing else. The research carries a warning label here: practice-first learning isn't guidance-free learning. Kirschner, Sweller, and Clark's much-cited paper – titled, without apology, "Why Minimal Guidance During Instruction Does Not Work" – found that true novices flounder when left to discover everything themselves. The designs that work pair experience-first sequencing with scaffolding: worked examples, support that fades as competence grows, short theory chunks delivered right after application. Pure discovery isn't the goal. Productive struggle inside a scaffold is.

Safety comes first, because nobody practices authentically while guarding themselves. Pilot 2 spends fifteen minutes on framing before a single concept gets taught: participation is voluntary, anyone can shift to an observer role at any point, and an X-Card sits in the center of the table – tap it and we adjust, no explanation required. There's a full walkthrough in safety tools for workplace roleplay.

Vocabulary second, and minimal. Participants need just enough shared language to notice what they're doing during play – recognition, not mastery. Pilot 2's core-concepts block runs 30 minutes and covers three things: noticing your own defensiveness (five to seven minutes), positions versus interests (ten to twelve minutes), and how to open a disagreement with someone who outranks you (eight to ten). Everyone gets a reference card of conversation-opening phrases – "Help me understand...", "What problem are we trying to solve?" – to glance at mid-game. Nobody's asked to memorize anything.

Mechanics third, stripped to essentials. Rules teaching gets its own half hour, and the goal is deliberately modest: players learn enough of the game system to roll dice and use their character, and the GM handles the rest. That's cognitive load theory in the wild. Every unit of working memory spent decoding game mechanics is a unit not spent on the interpersonal dynamics the whole day exists to practice.

The session plan is equally explicit about what stays out. Consensus-building procedures, conflict-avoidance theory, escalation patterns – none of it gets pre-taught. All of it waits for the debrief, after participants have lived through an example worth explaining. The instruction in the plan is four words long: experience first, framework second.

How does Operation Aetherfall sequence theory and practice?

Short concept block, long play, and a debrief that carries the theory. After the 30-minute core-concepts block and rules teaching, the table plays for roughly four hours – a briefing scene where players hold information that contradicts their commander's plan, a string of team decisions under pressure, a finale shaped by the choices they made along the way. While they play, the GM keeps a notepad and records three to five specific moments: who challenged authority and how, how the group actually reached decisions, who held back when they had something to contribute. Those notes are the raw material for everything that follows. Generic debrief questions produce generic answers; "walk me through how the group made the sabotage decision" produces analysis.

The debrief runs 90 minutes in three phases – reactions, analysis, application – and the analysis phase alone takes 40 to 50 minutes, longer than the entire pre-game theory block. This is where the frameworks actually get taught. Escalation patterns come up when the group examines the moment a disagreement started feeling personal. Consensus-building gets named while unpacking how a decision was made and whose input never made it in. The theory that would've been abstract at 9 a.m. is concrete at 3 p.m., because now it explains something everyone watched happen. The full three-phase structure is in how to debrief a training exercise.

The day's last theory is the most practical piece. Participants write implementation intentions – specific if-then plans for one thing they'll try differently at work – and share them out loud. The session plan carries the numbers for why: specific plans roughly double transfer, from about 30% to 67%, and sharing a commitment publicly raises follow-through from around 43% to 76%. Two full pilot runs in, that end-loaded arc – play, then analyze, then commit – is the design decision I'd defend hardest in a room of skeptics.

How do you decide what to teach before practice?

Sort every concept by the job it does. If it helps people practice safely or notice what they're doing in the moment, it goes up front and stays short. If it explains what they'll have just experienced, it waits for the debrief, where it lands as an answer instead of a lecture.

Content	When to teach it	Why
Safety rules, opt-outs, consent tools	Before practice, first thing	Nobody engages authentically while guarding themselves
Minimal vocabulary – two or three named concepts	Before practice, briefly	Labels let participants notice a skill while using it
Exercise mechanics	Before practice, stripped down	Working memory spent on rules is stolen from the skill
Frameworks that explain behavior – escalation, avoidance	After practice, in the debrief	They answer questions the experience just raised
Models and procedures for doing better	After practice, in the debrief	Just-in-time theory attaches to a felt gap
Dynamics participants could discover by playing	Never up front	The discovery is the learning

When a concept could go either way, hold it. The research's recommended arc – scenario or demonstration first, brief targeted theory, scaffolded practice with fading support, then independent application and reflection – leaves plenty of room for theory after the experience and almost none for theory before it. A concept taught too early isn't neutral, either. It occupies working memory the exercise needs, and it spoils discoveries the game was built to deliver.

Thirty minutes of theory in a seven-and-a-half-hour day isn't a shortage. It's placement.