Learning Design Series (6): Problem-Centered
Problem-Centered e-learning is the mode to use when the course should start with the real problem, not with the theory. It is useful for troubleshooting, operations, product education, customer onboarding, analytical workflows, and any topic where learners need to solve something, not just understand a model.
Many courses get this backwards. They begin with definitions, background, framework, terminology, and only later reveal the practical problem. By then, the learner has already decided the course is a content dump with nicer formatting.
By the end of this article, you should be able to decide when Problem-Centered design is the right mode, choose the right e-learning blocks for it, and prompt an AI course builder with enough detail that the problem becomes the spine of the course.
This article builds on What is Instructional Design in E-Learnings?. Start there if you want the overview and comparison of the different learning design modes before going deep on Problem-Centered.
What is Problem-Centered e-learning?
Problem-Centered e-learning is a course design mode where learners begin with a realistic problem and learn the concepts, steps, tools, or decisions needed to solve it. The problem is not a decorative example at the end. It is the organizing structure of the course.
In a weak course, the learner hears:
First, here are the five principles of customer onboarding.
In a Problem-Centered course, the learner sees:
A new customer has completed kickoff, but adoption is stalled after two weeks. What do you check first?
Now the concepts have a job. Health signals, stakeholder alignment, activation events, handoff notes, and escalation rules are no longer abstract content. They are tools for solving the problem.
This is the main difference from Scenario-Based design. Scenario-Based often focuses on decisions inside a situation. Problem-Centered design uses the problem as the backbone for the whole learning path. The learner keeps returning to the problem as new information becomes useful.
When should you use Problem-Centered e-learning?
Use Problem-Centered e-learning when learners need to diagnose, solve, improve, or explain a real work problem. It works best when the topic feels abstract until it is attached to a situation that needs fixing.
This mode works especially well when the topic falls into one of these families:
| Troubleshooting and diagnosis | Operations and process improvement | Product and customer education | Technical or analytical workflows |
|---|---|---|---|
| Login failures | Bottleneck analysis | Failed customer setup | Dashboard investigation |
| Support triage | Quality defects | Feature adoption gaps | Report interpretation |
| Root-cause analysis | Handoff problems | Onboarding health issues | Data-cleaning decisions |
| Error-message diagnosis | Rework reduction | Customer activation blockers | Configuration review |
| Incident analysis | Process exceptions | Plan or permission confusion | Metrics anomaly review |
| Known-issue playbooks | SLA misses | Renewal risk signals | Workflow automation issues |
It is weaker when learners only need a quick update, a job aid, or repeated guided practice on a fixed procedure. For concise updates, use the learning design mode "Clear & Focused". For aids learners keep open while working, use the design mode "Performance-Support". For step-by-step skill building, use Scaffolded Practice. For ambiguous judgement and consequence, use Scenario-Based.
The watch-out is fake problem framing.
If the problem disappears after the opening paragraph, the course is not problem-centered. It just has a problem-shaped hook.
What are the advantages of Problem-Centered design?
Problem-Centered design makes e-learning feel useful faster because the learner sees why the content matters before the explanation begins. It reduces theory-first friction and gives every concept a reason to exist.
| Advantage | How it helps the learner | What the designer must still do |
|---|---|---|
| Immediate relevance | Learners see the work problem from the start. | Choose a problem they recognize. |
| Better attention | The problem creates a reason to keep going. | Avoid mystery for mystery's sake. |
| Stronger retention | Concepts attach to a concrete situation. | Revisit the same concept in variations. |
| Practical transfer | Learners practise using knowledge to solve. | End with a fresh transfer problem. |
| Intrinsic motivation | The task feels like real work, not school. | Make the problem consequential but not theatrical. |
| Cleaner scope | Content is included only if it helps solve. | Cut background that does not serve the problem. |
The last point is the hidden benefit. A real problem is a ruthless editor. If a section does not help learners understand, diagnose, decide, or act, it probably does not belong in this course.
How do instructional design principles show up in this mode?
Instructional design principles matter in Problem-Centered courses because the problem can become either a useful anchor or a gimmick. Gagne's events still apply, but they are expressed through problem, hypothesis, information, attempt, feedback, and transfer.
| Principle | Problem-Centered implementation | Common mistake |
|---|---|---|
| Gain attention | Start with a real problem learners may face. | Opening with a dramatic but unrealistic crisis. |
| Inform objectives | Say what kind of problem learners will solve. | Listing vague objectives like "understand the process." |
| Stimulate recall | Ask what learners would check first or what they already know. | Providing all information before learners think. |
| Present content | Reveal concepts and tools when the problem needs them. | Teaching the full framework before the problem. |
| Provide guidance | Use hints, examples, data, and expert reasoning. | Withholding essential information to create false difficulty. |
| Elicit performance | Ask learners to diagnose, compare, choose, classify, or explain. | Making them read a problem and then click Next. |
| Provide feedback | Connect feedback to the problem logic. | Feedback that only says correct or incorrect. |
| Assess performance | Use a new but related problem. | Repeating the same case with different wording. |
| Enhance transfer | End with cues for recognizing similar problems at work. | Ending after the first solution is revealed. |
Merrill's first principles are especially relevant here because they put real-world problems at the center of instruction. The course should activate what learners know, demonstrate the solution, let them apply it, and help them integrate the approach into future work.
How do you structure a Problem-Centered course?
A Problem-Centered course should move from problem to investigation to useful content to solution attempt to feedback, then repeat with a variation. The course should not solve the problem for the learner too early. But it also should not starve them of the information they need.
| Course part | Purpose | Typical blocks |
|---|---|---|
| Problem setup | Present the real task, symptom, or failure. | TITLE, EXPLANATION, IMAGE, QUOTATION |
| Initial hypothesis | Ask what learners would check or assume first. | QUIZ_ONE_CHOICE, CARD_SORT, TABS |
| Information reveal | Provide the data, rule, concept, or artifact needed next. | TABLE, ACCORDION, DID_YOU_KNOW |
| Guided analysis | Help learners compare signals and narrow options. | MATCHING_CARDS, TABLE, TABS |
| Solution attempt | Ask learners to choose, sequence, classify, or explain. | QUIZ_MULTIPLE_CHOICE, ORDERING, CARD_SORT |
| Feedback and principle | Explain why the solution works and name the transferable principle. | EXPLANATION, PROCESS, BULLETED_LIST |
| Transfer problem | Use a fresh but related problem with less help. | QUIZ_ONE_CHOICE, TABLE, PROCESS |
The pacing matters. If the course reveals everything at once, the learner is not solving. If it hides everything, the learner is guessing. A good Problem-Centered course reveals information just in time.
Which e-learning blocks work best for Problem-Centered courses?
Use e-learning blocks to present the problem, reveal evidence, compare hypotheses, and support solution attempts. In Problem-Centered mode, the block earns its place when it helps learners understand the problem or make progress toward a solution.
Good primary blocks:
| Block type | How to use it in Problem-Centered courses |
|---|---|
| TITLE | Name the problem, case, investigation step, or solution checkpoint. |
| EXPLANATION | Provide concise context, instructions, and principle debriefs. |
| IMAGE | Show screenshots, artifacts, forms, dashboards, physical context, or evidence. |
| QUOTATION | Present customer, colleague, manager, or stakeholder statements as problem clues. |
| TABLE | Compare symptoms, causes, evidence, options, thresholds, or before/after states. |
| TABS | Separate hypotheses, roles, data sources, phases, or possible causes. |
| ACCORDION | Reveal clues, expert reasoning, hints, or optional detail in stages. |
| DID_YOU_KNOW | Highlight a rule, constraint, or hidden dependency that changes the solution. |
| QUIZ_ONE_CHOICE | Ask for the next best check, diagnosis, or action. |
| QUIZ_MULTIPLE_CHOICE | Select all relevant signals, risks, causes, or next steps. |
| CARD_SORT | Classify evidence, symptoms, cases, or actions. |
| MATCHING_CARDS | Match symptoms to causes, data points to interpretations, or problems to owners. |
| ORDERING | Sequence investigation steps or remediation actions. |
| PROCESS | Summarize the repeatable problem-solving approach. |
| FLIP_CARD | Pair misconception -> correction, symptom -> meaning, or clue -> implication. |
| TABLE_OF_CONTENTS | Support a multi-lesson problem-solving course. |
Usually avoid long front-loaded explanations, decorative images, and quizzes that ask for definitions before the learner has a reason to care. If the problem is real, the content should arrive when the learner needs it.
Weak question:
What are the five stages of onboarding health analysis?
Better Problem-Centered task:
A customer completed kickoff but has not activated the reporting feature after 14 days. Which three signals should you check before calling this account "low adoption"?
The second question makes the framework useful.
How should AI be prompted to create a Problem-Centered course?
To prompt AI for a Problem-Centered course, give it the course ingredients, the real problem, the learner's likely first assumptions, the source material, the available block types, and the transfer problem. Do not assume the AI will naturally keep the problem alive. It may start with a strong scenario and then drift into a generic explainer.
Here is a detailed prompt template for AI-assisted e-learning course generation:
Create an e-learning course using the Problem-Centered instructional design mode.
Course inputs:
- Topic: [topic]
- Audience: [audience]
- Source material: [provided material, uploaded content, or model knowledge]
- Course goal: [what learners should know or be able to do by the end]
- Target depth: [Remember / Understand / Apply / Analyze-Decide / Create]
- Starting point: [Beginner / Novice / Intermediate / Advanced / Expert]
- Prerequisites: [what learners should already know]
- Evidence of learning: [how we will know they got it]
- Common pitfalls: [what learners usually misunderstand or do wrong]
- Relevance angle: [why learners should care]
- Spacing: [revisit key ideas across lessons or cover each once]
- Challenge ramp: [Ease in / Balanced / Push hard]
- Hands-on level: [Mostly reading / Balanced / Hands-on / Conversational]
- Tone: [tone]
- Course length: [minutes]
Your task:
Design a complete e-learning course, not a single lesson.
Use Problem-Centered design consistently. Start with a realistic work problem and make it the backbone of the course. Teach concepts, rules, tools, and procedures only when they help learners understand or solve the problem. Do not create a generic explainer with a problem in the introduction. Do not reveal all information before the learner has formed an initial hypothesis.
The course should follow this arc:
1. Present the real problem, symptom, failure, question, or case.
2. Clarify the role, context, stakes, and success standard.
3. Ask learners what they would check, assume, or try first.
4. Reveal only the information needed for the next step.
5. Teach the concept, rule, tool, or method when the problem needs it.
6. Let learners diagnose, classify, compare, sequence, or choose an action.
7. Give feedback that explains the problem logic and not only the correct answer.
8. Add a variation or complication that changes the solution.
9. Finish with a fresh transfer problem and a repeatable problem-solving approach.
For every lesson, specify:
- Lesson title.
- Lesson purpose.
- Estimated time.
- The problem or subproblem being worked on.
- The learner's likely first assumption or mistake.
- What information is visible at the start.
- What information is revealed later and why.
- Recommended e-learning blocks.
- What each block should contain.
- The learner action required.
- Correct answer or recommended next step.
- Plausible wrong answers based on common pitfalls.
- Feedback for each wrong answer.
- Which concept, rule, cue, or method is taught just in time.
- How this lesson connects back to the original problem.
- How difficulty increases from the previous lesson.
Use only these e-learning block types:
TITLE, EXPLANATION, IMAGE, BULLETED_LIST, NUMBERED_LIST, QUIZ_ONE_CHOICE,
QUIZ_MULTIPLE_CHOICE, ACCORDION, FLIP_CARD, SPACE, COLUMNS, CARD_SORT,
CAROUSEL, DIVIDER, CODE, TABLE_OF_CONTENTS, TABS, TIMELINE, PROCESS,
ORDERING, MATCHING_CARDS, TABLE, DID_YOU_KNOW, QUOTATION.
Preferred blocks for Problem-Centered mode:
- TITLE for problem, investigation, and solution checkpoint headings.
- EXPLANATION for concise setup, instructions, and principle debriefs.
- IMAGE for screenshots, artifacts, forms, dashboards, or physical evidence.
- QUOTATION for customer, colleague, manager, or stakeholder clues.
- TABLE for comparing symptoms, causes, evidence, options, thresholds, or before/after states.
- TABS for hypotheses, roles, data sources, phases, or possible causes.
- ACCORDION for staged clues, expert reasoning, hints, or optional detail.
- DID_YOU_KNOW for rules, constraints, or hidden dependencies that change the solution.
- QUIZ_ONE_CHOICE for the next best check, diagnosis, or action.
- QUIZ_MULTIPLE_CHOICE for selecting relevant signals, risks, causes, or next steps.
- CARD_SORT for classifying evidence, symptoms, cases, or actions.
- MATCHING_CARDS for matching symptoms to causes, data points to interpretations, or problems to owners.
- ORDERING for investigation steps or remediation actions.
- PROCESS for the repeatable problem-solving approach.
- FLIP_CARD for misconception -> correction, symptom -> meaning, or clue -> implication.
- TABLE_OF_CONTENTS for multi-lesson problem-solving courses.
Avoid:
- Teaching the whole framework before the problem.
- Using the problem only as a decorative introduction.
- Hiding information learners would realistically have.
- Revealing all information before learners think.
- Creating fake mystery that depends on withheld facts.
- Asking trivia or terminology questions that do not help solve the problem.
- Leaving out the success standard.
- Feedback that does not explain the problem logic.
- A transfer task that is identical to the worked example.
Adjust the difficulty ramp:
- If Challenge ramp is "Ease in", use a simple problem, more visible cues, and more guided analysis.
- If Challenge ramp is "Balanced", use one guided problem, one variation, and one independent transfer problem.
- If Challenge ramp is "Push hard", use fewer hints, more ambiguous evidence, and faster movement to independent diagnosis.
Adjust the hands-on level:
- Mostly reading: include a worked problem and one light hypothesis check.
- Balanced: include staged evidence, one guided decision, and one short transfer task.
- Hands-on: include multiple diagnosis, sorting, matching, and sequencing tasks.
- Conversational: use Q&A prompts where the learner explains what they would check next and why.
Weight the course approximately:
- 10% problem setup, role, relevance, and success standard.
- 15% prior knowledge, first hypothesis, and likely misconception.
- 20% just-in-time content and evidence reveal.
- 25% analysis, diagnosis, solution attempt, or decision practice.
- 15% feedback, principle, expert reasoning, and variation.
- 15% transfer problem and repeatable problem-solving approach.
The prompt is detailed because AI often treats "problem-centered" as a writing style rather than a course architecture. It may write a case study, explain everything, and then add a quiz. The prompt has to force problem persistence, staged information, learner hypotheses, feedback, and transfer.
Where AI helps with Problem-Centered design
AI can help turn source material into realistic problems, likely misconceptions, staged clues, practice cases, and feedback variants. It is especially useful when the source material is dense and the designer needs to find the practical work problem inside it.
Useful AI tasks include:
| AI task | Why it helps |
|---|---|
| Drafting problem statements | Makes the course start from a concrete work issue. |
| Extracting signals and causes | Turns source material into evidence learners can analyze. |
| Creating plausible wrong answers | Helps feedback address real misconceptions. |
| Varying cases | Supports transfer instead of one-case memorization. |
| Mapping blocks to problem steps | Keeps the e-learning structure practical. |
Where human review still matters
Human review is essential because a problem can sound realistic and still be wrong. The reviewer should check whether the case reflects actual work, whether the evidence is available in the real situation, whether the recommended solution is allowed, and whether the feedback matches expert reasoning.
One practical review question helps:
Would a competent employee recognize this as a problem they actually need to solve?
If the answer is no, the course may be problem-themed, but it is not problem-centered.
Worked example: diagnosing a stalled customer onboarding
Imagine a customer education lead needs a 20-minute course for customer success managers. The source material is an onboarding playbook, health-score definitions, product activation data, and notes from three stalled accounts.
The course goal:
CSMs can diagnose why a new customer is not reaching first value and choose the next intervention.
The common pitfall:
CSMs treat every stalled account as a training problem, even when the real blocker is missing stakeholder ownership, unclear success criteria, or plan limitations.
A Problem-Centered version could look like this:
| Lesson | Problem step | Blocks | What the learner does |
|---|---|---|---|
| 1. The stalled account | Customer completed kickoff but has no activation event. | TITLE, QUOTATION, TABLE | Review the case and choose first hypothesis. |
| 2. The evidence | Usage data, stakeholder notes, and plan limits appear. | TABS, ACCORDION, QUIZ_ONE_CHOICE | Decide what to check next. |
| 3. The false lead | Low usage looks like training need, but admin access is missing. | DID_YOU_KNOW, CARD_SORT | Sort signals into cause categories. |
| 4. The intervention | Options include training, admin follow-up, success-plan reset, or escalation. | TABLE, QUIZ_MULTIPLE_CHOICE | Choose the best next actions. |
| 5. Transfer problem | A new customer has different symptoms. | PROCESS, QUIZ_ONE_CHOICE | Apply the diagnostic approach independently. |
The final check should not ask:
What are the four onboarding health categories?
It should ask:
A customer has attended training, but only one admin has logged in and no report has been scheduled. What is the most likely blocker, and what should the CSM do next?
That is the course doing its job. The learner is using the health framework to diagnose, not memorizing the framework label.
How an AI e-learning builder should support this mode
An AI e-learning builder should treat Problem-Centered as a problem-solving architecture. It should ask for the topic, audience, source material, goal, target depth, starting point, prerequisites, evidence, pitfalls, relevance, spacing, challenge ramp, hands-on level, tone, and course length.
Then it should produce a course with a realistic problem, staged evidence, learner hypotheses, just-in-time content, practice tasks, feedback, variations, and a final transfer problem. The recap should leave learners with a repeatable way to recognize and solve similar problems.
The important part is not that the course starts with a problem. It is that the problem keeps shaping what the learner sees, learns, and does.
Related reading
FAQ
Is Problem-Centered learning the same as Scenario-Based learning?
Not exactly. Scenario-Based learning usually focuses on choices inside a situation. Problem-Centered learning uses the problem as the structure for the whole course. They overlap, but the emphasis is different.
What makes a good problem for e-learning?
A good problem is realistic, recognizable, and solvable with the knowledge the course teaches. It should include enough context to think, enough uncertainty to require judgment, and a clear success standard.
Should the problem appear before the explanation?
Usually, yes. The problem gives learners a reason to care about the explanation. You can still provide orientation, but avoid teaching the entire framework before learners understand why it matters.
How many problems should a short course include?
For a short workplace course, one main problem, one variation, and one transfer problem is often enough. More problems can help, but only if each one teaches a meaningful difference.
When should I choose another design mode?
Choose Clear & Focused for concise explanations, Scaffolded Practice for guided skill-building, Scenario-Based for ambiguous decisions, and Performance-Support for job aids used during work.
Can AI create Problem-Centered e-learning well?
AI can create useful first drafts, but it needs strong instructions and source material. The riskiest parts are fake problems, tidy causes, and feedback that sounds plausible but does not match the real work.
Conclusion
Problem-Centered e-learning works when the problem drives the course from start to finish. Start with the real issue, reveal content only when it helps, let learners try a solution, and end with a new problem they can solve more independently.