When should a scenario program go deep instead of broad?

Every topic has many aspects and a few hard nuances. The design question that follows is where your decision points go: which ones spread out across the topic, and which ones stack up on a single gap. Most programs never ask it explicitly, and most land, correctly, on spreading out. This guide is about that default and its exceptions: what breadth actually does, what depth is for, the three cases where stacking decision points on one gap is the right call, and how the data shows a deep design produces measurable improvement. It is the allocation companion to designing decision points: that guide owns the craft of a decision inside a scenario, while this one owns how decisions are distributed across scenarios.

Breadth is the normal shape of covering a topic

Most training exists to cover a topic, and topics have many aspects. A program on consultative selling touches discovery, stakeholder mapping, the pricing conversation, the objection moment, the close. A clinical program touches diagnosis, therapy selection, monitoring, escalation. When a content developer turns that material into scenarios, the design naturally contains many different situations and decision points, each covering a different aspect of the topic. That is breadth. It is not a compromise. It is simply the shape a program takes when its job is to cover a topic, which is most programs' job most of the time.

Breadth also matches how gaps usually distribute. In most topics the struggles are spread across the terrain: one gap in how managers set priorities for a quarter, another in how they delegate a stretch assignment, another in when they escalate a project risk. Each gap becomes one decision point, each decision point gets a situation that shows it off, and the program walks the learner across the whole territory while the ground keeps changing and the learner stays interested.

What depth is for

Sometimes one or two of a program's critical learning objectives are different in kind. The gap is nuanced and hard to detect, and no single encounter can convey it. Consider a manager learning to detect early disengagement in a team member. Disengagement almost never announces itself. Instead, one person goes quiet in meetings, another starts missing small deadlines, another stops volunteering ideas. The underlying issue is the same; only the surface differs, and the manager's real skill is recognizing it beneath those different presentations. Or the clinical version: different patient types present the same underlying issue differently, and the clinician's skill is seeing through the presentation to the condition.

For objectives like these, the design changes. More than one decision point focuses on the same gap across different situations, and across those meetings learners start to recognize the key aspects, the nuance, that no single situation could carry. This is the distinction Syandus' founder draws in his published framework as flexible breadth and flexible depth: breadth builds the capability to handle different types of situations, while depth builds mastery of the variations within similar situations, so the learner knows when and how to adapt (Seifert, 2024). His observation at the center of the depth case is that we often face the same decision point while different situations make different options optimal.

The mechanic behind that observation matters, because it is what makes depth coherent rather than repetitive. A decision point represents the gap itself: the distance between what people do now and what they should do. When a program goes deep, that decision point stays constant, and so does its option set. The learner meets the same decision, with the same options in front of them (a given situation may hide one or two), in situation after situation. The situation is the variable, and what moves with it is which options are optimal. Holding the options fixed is not a limitation. If every situation brought a fresh set of options, each encounter would measure a different gap, and nothing could be compared across them. The constant decision against the changing situation is what lets nuance become visible.

The three cases where depth earns its decision points

Depth is a deliberate allocation, and it belongs in three cases.

1. Rehearsal. In Stage 3 of a learning program, the learner has already decided what to do and is refining how to deliver it: phrasing, tone, composure. Delivery genuinely improves with repetition, so depth is the natural mode there, and nobody experiences it as redundant. The learning-science account agrees that strategy should follow stage: Kim, Ritter, and Koubek's skill-retention theory recommends different training approaches at different stages of learning (Kim, Ritter, & Koubek, 2013).

2. Detection. Here the skill is recognizing which type of situation you are in. Is this quiet stretch a workload crunch or the start of disengagement? Is this presentation the typical form of the condition or the atypical one? Detection is learned only by meeting the variations. A list of the ways disengagement can surface produces a learner who can recite it; a learner who has faced each surface in a live situation, misread one or two, and been coached on the tells can spot the next variation in the wild.

3. Situation-dependent optimality. In the third case, the right action itself changes with the situation. The same option is optimal in one presentation and suboptimal in another: the direct pricing conversation that wins over the transparent buyer backfires with the one still building internal support, and the therapy that fits the standard presentation is the wrong call for the patient with a complicating history. This goes beyond detection. The learner is not just identifying the situation but discovering that the answer moves, and that has to be experienced, not described. And this case exists only because the option set stays fixed: the learner can watch an option shift from optimal to suboptimal only if it is the same option at the same decision.

Why depth is hard to build

The costs are real, and they are design reality rather than a deterrent. The central difficulty is that depth requires genuinely different situations for the same decision. If the situations are too similar, the later encounters measure memory: the learner recognizes the scenario rather than the situation type and repeats what worked last time. If the situations are too different, the encounters no longer land on the same gap, and comparing them measures nothing. Authoring the middle, situations different enough to mean something and similar enough to compare, is some of the hardest content work in scenario design.

Learners also have limited appetite for revisiting territory. A broad program keeps the ground changing; a deep sequence asks the learner to return to what can feel like the same neighborhood. The variations have to earn each return by being genuinely different worlds around the same decision.

There is a sequencing implication as well. Depth delivers most once learners have footing: research on learning from errors found that mixing incorrect approaches with correct ones improved far transfer, the ability to apply learning to situations unlike the training examples, only for learners with favorable prior knowledge (Große & Renkl, 2007). A depth sequence dense with tempting wrong turns belongs after the topic's basics are in place, not before.

What the data shows a deep design delivers

The same structure that makes depth hard also makes it measurable. A matched decision point is the same gap with the same option set, so encountering it in a different situation later in the program is a genuine second reading of the same skill. AliveSim has measured exactly this at scale in its medical-education programs: in a matched decision-point analysis of 31,673 decisions across seven programs, clinicians showed a 2.85x improvement in decision-making (d = 3.95, p < 0.001) when they encountered similar situations later in the simulation. Because the later decisions arrive in changed situations, the improvement is not scenario memory. What improved is the handling of the gap itself.

The broader simulation literature points the same direction. The McGaghie review of simulation research for continuing medical education reports a highly significant dose-response relationship: more hours of simulation-based learning produced higher outcome gains (McGaghie, Siddall, Mazmanian, & Myers, 2009). Depth is the design pattern that concentrates that dose on the one objective that needs it.

The allocation rule

In AliveSim's Guided Scenarios, both allocations are native. A broad program walks learners through the topic's aspects, one decision point per gap, each in its own situation, with mentoring at every choice. A deep program brings the same decision point back across scenarios with different prospects or different patients, options constant, optimality moving with the situation, and the mentoring at each encounter explains why this situation calls for a different choice. The founder's flexible-breadth and flexible-depth framework, with both the sales and the clinical examples in full, is published in AliveSim's Science of Guided Scenarios series.

The rule for an educational designer comes down to this: cover the topic broadly, one decision point per gap. Then look hard at the one or two objectives that are nuanced and hard to detect. If the skill is a delivery that improves with rehearsal, a detection that only variations can build, or an answer that moves with the situation, give that gap several decision points, hold the options constant, vary the situations, and let the matched later encounters show whether the nuance landed.

References

• Große, C. S., & Renkl, A. (2007). Finding and fixing errors in worked examples: Can this foster learning outcomes? Learning and Instruction, 17(6), 612–634.
• Kim, J. W., Ritter, F. E., & Koubek, R. J. (2013). An integrated theory for improved skill acquisition and retention in the three stages of learning. Theoretical Issues in Ergonomics Science, 14(1), 22–37.
• McGaghie, W. C., Siddall, V. J., Mazmanian, P. E., & Myers, J. (2009). Lessons for continuing medical education from simulation research in undergraduate and graduate medical education. Chest, 135(3 Suppl), 62S–68S.
• Seifert, D. (2024). The Science of Guided Scenarios, Part 4: The Power of Flexible Decision-Making. Syandus Blog.

Published July 18, 2026 · 8 min read