Examples & FAQ: personas, scenarios, criteria & reports
This article is a companion to the Creating & Configuring Simulations series. The other articles explain where each setting lives and what it does; this one is a cookbook of worked examples, design principles, and frequently asked questions you can adapt to your own simulations. For step-by-step setup, see The simulation editor, explained, Simulation Instructions, Feedback Instructions, and Simulation Display.
The fastest way to get all of this is to let Prof Cesura (the AI builder) draft it for you and then refine in the editor — see Create a simulation with AI. The examples below show what "good" looks like when you do refine.
Design principles
Focus and depth — the golden rule
The single most important rule: one learning goal, one character, one situation. Focused, single-scenario simulations lead to better learning than sprawling ones that try to cover everything. If you need to teach several situations, build a separate simulation for each, and let them add up to a library.
A simulation that tries to cover four healthcare conversations (intake, bad news, discharge, complaints) with 20+ criteria leaves learners overwhelmed and unsure what to improve. A single simulation on "Delivering difficult news to a patient" with four or five criteria lets them practise intensely, get clear guidance, and build real confidence.
Make the character believable
Give the character a real role, clear motivations, and a consistent way of speaking, and write the way people actually talk. Natural speech makes the simulation feel grounded:
- Instead of: "I am experiencing significant apprehension regarding the implementation timeline."
- Write: "I'm worried this is going to take way longer than you're saying."
Set a vivid scenario with real stakes
The scenario is the learner's stage: make clear who they are, what is happening, and what is at stake. Give enough context to feel real without burying them in detail.
- Too vague: "A customer is unhappy. Talk to them."
- Clear: "You're a support specialist for a cloud-hosting company. A customer's website went down for two hours this morning — your company's fault. They're on the phone now, frustrated, and they're up for renewal in two weeks. Rebuild trust and understand their needs before that renewal."
Write feedback that helps
Aim for two to five criteria, each describing an observable behaviour, not a trait:
- Weak: "Communicated well", "Was empathetic", "Handled the call well".
- Strong: "Acknowledged the customer's frustration before offering a solution", "Explained the cause of the outage in non-technical language", "Offered a concrete next step".
When a simulation is the right tool
Simulations shine for skills that require practice with unpredictable human behaviour: sales calls, customer service, difficult conversations, negotiation, conflict resolution, delivering bad news, clinical interviews, and crisis response. They are less suited to factual-knowledge testing (use a quiz), physical procedures (use video or hands-on practice), simple compliance checks, system navigation (use an interactive tutorial), or isolated cognitive drills like maths or coding.
Worked persona examples
A strong persona profile (the Detailed Characteristics field) covers background, personality, communication style, current emotional state, and the character's relationship to the learner. Aim for roughly 200–400 words. Then add three to five sample statements — a "verbal fingerprint" in the character's own voice.
Example 1 — Nurse manager (healthcare)
Name: Patricia Chen Age: 45 Gender: Female
Professional background - Registered Nurse for 22 years; Nurse Manager for 5 - Manages a 24-bed medical-surgical unit - Known for high patient-satisfaction scores and staff retention Personality - Calm under pressure; rarely flustered - Pragmatic; patient safety first, then efficiency - Empathetic but not sentimental; comfortable with hard conversations - Values reliability and follow-through over excuses Communication style - Speaks clearly and directly; explains clinical terms to non-clinical staff - Listens actively before responding - Often asks "What do we need to do to keep this patient safe?" Current emotional state - Tired from staffing shortages; protective of her team - Concerned about quality of care during busy periods Relationship to the learner - Sees the learner as a new nurse on the unit - Wants them to succeed but expects them to learn quickly
Sample statements:
- "Patient safety is our priority. Everything else comes after that."
- "Tell me what you're seeing and thinking — I want to understand your clinical reasoning."
- "I appreciate you bringing this to me. Let's figure out how to handle it."
Example 2 — Skeptical VP of Operations (business)
Name: Marcus Johnson Age: 52 Gender: Male
Professional background - VP of Operations at a mid-size manufacturer; 28 years in the field - Burned by two failed software implementations in the last five years - Responsible for a $50M budget Personality - Risk-averse after past failures; wants proof over promises - Detail-oriented; asks tough questions - Respects competence and honesty; dismisses hype Communication style - Blunt and direct; asks "What's the business case?" repeatedly - Uses data and metrics as proof; dry sense of humour Current emotional state - Frustrated with current bottlenecks; pressured to improve margins - Skeptical about proposed solutions Relationship to the learner - Sees the learner as a vendor/consultant trying to sell something - Will only trust them if they understand manufacturing realities
Sample statements:
- "The last two systems we bought promised the world and delivered headaches."
- "I need to see the data. Show me how this actually improves our margins."
- "You're speaking in generalities. I need specifics about our operation."
Example 3 — Overwhelmed parent (education)
Name: Jennifer Rodriguez Age: 38 Gender: Female
Professional background - High-school Spanish teacher; mother of three - Youngest child has a learning difficulty; feels stretched thin - First-generation college graduate Personality - Intelligent and well-informed; anxious about her youngest's future - Grateful for help but fiercely independent Communication style - Speaks thoughtfully; often apologizes or second-guesses herself - Opens up when she feels heard and validated Relationship to the learner - Sees the learner as an educator who might judge her - Wants reassurance and concrete strategies
Sample statements:
- "I worry that he's falling behind and we're not catching it early enough."
- "I've tried some things at home, but I'm not sure if I'm doing it right."
- "What can I realistically do to help him? I want to do everything I can."
Sample statements across other domains
- Emergency physician: "Walk me through what you saw when the patient came in." / "We don't have time to be guessing here — what's your clinical reasoning?"
- Frustrated student: "I don't get why we're learning this. When will I ever use it?" / "I've been studying for hours and still don't understand."
- Empathetic support agent: "I totally understand how frustrating that must be." / "We're going to make this right. Here's what I can do."
Common persona mistakes
- Too generic. "She's friendly and professional, a software engineer." Better: "A Python specialist who asks tough technical questions and is impatient with vague answers; warm and funny outside meetings, focused on problems inside them."
- Inconsistent details. Don't describe a character as "patient and supportive" and then have sample statements full of dismissiveness. If the contrast is intentional, explain it: "Appears impatient in meetings under time pressure, but genuinely supportive one-on-one."
- Overly complex backstory. Skip the ten-paragraph life history. Include only what shapes behaviour in this scenario.
- Persona that doesn't match the scenario. A cautious, detail-obsessed persona fights a scenario that needs fast, risky decisions. Align traits with the learning goal.
Worked scenario examples
Write the scenario story in the second person ("you…"), roughly 150–300 words, with real stakes. A strong structure: the situation, what led to this moment, current context, what you need to accomplish, and the stakes.
Example 1 — De-escalating an angry family member (ER)
The situation A patient's family member, David, has waited four hours for test results. He's now in the hallway, voice raised, demanding to know why, and threatening to take his father to another hospital. What led to this moment - The ER is at twice its normal volume; the lab is backed up - The family was never updated about the delay; David feels ignored Current context - Other families are watching; staff are stretched thin - The patient is stable - the urgency is emotional, not medical - David's frustration is legitimate, even though the delay isn't your fault What you need to accomplish - Calm David without making promises you can't keep - Validate his frustration and explain the situation honestly - Give a realistic timeline; model calm de-escalation for your team Stakes - Dismiss his anger and the family leaves with a bad impression - Over-promise and you create expectations you can't meet - Handle it well and the family feels heard, and your team learns from it
Learning goal: manage intense emotions while keeping honesty and boundaries. Pairs naturally with the Patricia Chen persona.
Example 2 — Negotiating a delayed project timeline (business)
The situation You're meeting Marcus, VP of Operations, to tell him a software project will slip four weeks. He was already skeptical; this confirms his doubts. What led to this moment - The project started strong with no early red flags - A key vendor missed a delivery, cascading into other delays - You've known for two weeks but hoped to catch up; now it's unavoidable Current context - Marcus has little patience for software projects after two bad ones - The delay hits his manufacturing timeline and budget - He's fielding questions from his own boss and sees you as responsible What you need to accomplish - Deliver the news directly (he hates surprises) - Take accountability without excuses; explain the specific cause - Present options and rebuild his confidence that you can deliver Stakes - Sugarcoat it and trust erodes; blame the vendor and you lose respect - Handle it well and he respects your honesty and backs the revised plan
Learning goal: deliver bad news with clarity, accountability, and forward focus. Pairs with the Marcus Johnson persona.
Example 3 — Addressing a student's declining performance (education)
The situation You're a reading specialist meeting Jennifer, a parent, about her son's reading comprehension, which has dropped 15 points in two months. He's starting to disengage. What led to this moment - He progressed well in the autumn; the decline started after the break - Jennifer has been managing this alone and hasn't asked the school for help Current context - This feels personal to Jennifer; she may blame herself - She's worried about a "struggling" label and is intelligent but stretched thin - She wants help but may be defensive about being judged What you need to accomplish - Listen and validate without dismissing her concerns - Describe what you observe in neutral, specific language - Explore causes collaboratively and agree on realistic next steps Stakes - Diagnose too fast and you miss the real cause and she feels judged - Handle it well and she feels supported, and her son gets help sooner
Learning goal: have a supportive, partnership-building conversation with a parent. Pairs with the Jennifer Rodriguez persona.
Connecting scenario to persona
The scenario should activate the persona's traits. A skeptical VP needs something to be skeptical about; a stressed character needs a situation that justifies the stress. Before finalizing, check: does the scenario match the persona's role, make their emotional state make sense, and invite their key traits to emerge — without contradictions?
Worked theory examples
The Theory (Theoretical Background) is the professional framework behind the simulation. It is internal — learners never see it, but the AI uses it to ground the feedback, the feedback chat, and the Realtime Guide. A strong entry covers the model and its stages, core principles, what excellent vs. weak performance looks like, and your sources. 500–2,000 words is typical.
Example — Clinical reasoning (medicine)
Diagnostic process - History & physical: thorough history, focused exam, listen to the narrative - Differential: broad first, then narrow by probability and severity; keep "can't-miss" diagnoses in view - Testing: order to confirm/exclude/refine; avoid unnecessary tests; interpret in clinical context Performance standards - Excellent: complete history addressing red flags; focused, documented exam; 3-5 ranked differentials; 2-3 targeted tests; communicates reasoning - Adequate: most history; basic exam; unprioritized differential; some unnecessary tests; limited communication - Needs improvement: rushed history; minimal exam; vague differential; many tests without rationale; poor communication
Example — Interest-based negotiation (law/business)
Core principles (from "Getting to Yes") - Interests vs. positions: a position is what someone says they want; an interest is why. Skilled negotiators uncover the why. - Create value: separate people from the problem; look for win-win options; generate options before committing to one. - Ethics: candor, fiduciary duty, honesty in settlement discussions. Performance standards - Excellent: surfaces both sides' interests; proposes creative, interest-based solutions; stays professional and ethical - Needs improvement: argues positions; misses ethical obligations; proposes one-sided solutions
What goes where
It is easy to put content in the wrong place. As a rule of thumb:
- Theory — professional frameworks, standards of practice, evaluation criteria. For the AI, so it evaluates well.
- Activity steps before the simulation (the introduction the learner sees) — background, learning objectives, instructions, knowledge checks. Prepares the learner.
- Scenario — the immediate situation and stakes. Engages the learner in the moment.
- Documents — reference files the AI can draw on (accepted types: PDF, Word
.docx, and plain text.txt; save the simulation first).
Worked feedback-criteria examples
Each criterion is a 2–4 word Feedback Topic plus the rubric the AI scores by. Keep topics specific and observable, and aim for two to five criteria. Here are starter sets by domain.
- Customer service: Active listening without defensiveness · Ownership and empathy · Clear next steps.
- Medical (breaking bad news): Preparation and setting · Plain language and pacing · Emotional support · Plan and resources.
- Management (peer conflict): Balanced perspective-taking · Identifying the root cause · Collaborative problem-solving.
- Sales: Discovery approach · Objection handling · Value communication · Relationship building.
- Negotiation: Interest identification · Creative problem-solving · Win-win orientation · Relationship preservation.
A criterion's rubric is most reliable when it states what the criterion measures and why, what excellent looks like, a 0–100 scale broken into bands, and any conditions that force a failing score.
Common criteria mistakes
- Too many criteria. More than five (let alone ten) dilutes the feedback. Trim to the core competencies.
- Too vague. "Professionalism" or "Leadership" leaves too much to interpretation. Describe a specific behaviour.
- Overlapping criteria. "Demonstrates empathy through tone" and "Uses empathetic language" measure the same thing — combine them.
- Outcomes instead of behaviours. "Successfully resolved the issue" is an outcome; "Identified the root cause by asking targeted questions" is a behaviour.
For the overall grade, you can let the AI decide holistically (AI Generated) or set a Weighted Average with a weight per criterion. See Feedback Instructions.
Worked report (wrap-up) examples
After the conversation, the learner moves through the report. It always includes built-in steps (a start/landing step, the feedback assessment, the optional evaluator chat (the feedback chat), and an end step), and you can add your own steps. Keep it to three to five steps and end with something that points to action. Here are two example flows.
Customer-service simulation
- Start: "You've completed the customer interaction. Below you'll see how you did on the key skills, then a chance to discuss your approach."
- Feedback: overall plus per-criterion results for active listening, ownership and empathy, and clear next steps.
- Evaluator chat (feedback chat): let the learner ask "What could I have done differently with that frustrated customer?"
- Custom step (Report Step): quick-reference tips on tone, validation, and follow-through.
Leadership / difficult-conversation simulation
- Start: "Difficult conversations are a core leadership skill. This feedback is here to help you grow."
- Feedback: clarity and directness, emotional intelligence, collaborative problem-solving.
- Evaluator chat (feedback chat): "What was the hardest part of that conversation for you?"
- Custom step (Form): one strength to lean on, one behaviour to change, and how they'll know they've improved.
Tip: use the evaluator chat for live exploration and reflection, and custom steps for fixed resources, best practices, or next-step guidance.
Worked opening (introduction) examples
The steps above the Simulation marker are what the learner sees before the conversation. Aim for about five to ten minutes of content: a short welcome, the context (without giving away the character), how the simulation runs, and what they'll be assessed on. You can mix Section, Quiz, Media, Image, Document, and Form segments.
Clinical assessment
- Section (welcome): "In this simulation you'll run an initial consultation with a patient presenting with chest pain and a mild fever. Gather a thorough history, perform a focused exam, and recommend appropriate testing."
- Media: a three-minute background video of the patient's arrival.
- Document: a patient-history form for reference.
- Section (instructions): how the interface works and that they'll be assessed on clinical reasoning, efficiency, and communication.
- Quiz: a short knowledge check on cardiology differentials.
Customer-service role-play
- Section (welcome + scenario): "You're a support rep for an online retailer. A loyal customer's order arrived damaged; they're frustrated and considering a refund and a competitor. Resolve the issue while keeping the relationship."
- Section (instructions): respond naturally as in a real call; you'll be assessed on empathy, problem-solving, and retention.
- Form: a one-question pre-survey on the learner's customer-service experience.
Worked opening statements (the character's first line)
The opening statement (the Starting Sentence) is the character's very first message — the moment the learner enters the scene. Leave it blank for an auto-generated greeting, or write one to three sentences in the character's voice that set the scene and invite a response without giving everything away.
- Support (matter-of-fact): "Hi there — I'm Sarah from technical support. I can see you've had some trouble with your account. What's going on, and how can I help you sort it out?"
- Negotiation: "Thanks for making the time. I've read your proposal and I have a few questions about the timeline. Before we dig in — what's driving the four-week deadline?"
- Sales: "Welcome! I'd love to hear how our platform might fit your needs. Tell me a bit about your current situation — what brought you to explore solutions like ours?"
Worked Realtime Guide coaching tips
The Realtime Guide gives short, supportive advice during the live conversation; the level controls how detailed it is (see Enable & configure the Realtime Guide). Here is the same moment coached at two levels.
Beginner — detailed, with a dialogue example (up to about five sentences):
Good question - you're working within their timeline, which is practical. To go deeper, find out what's negotiable. Try: "What matters most for an on-time launch - the full feature set, the polish, or something else?" That helps you spot where there's room to move.
Regular — concise, one to three sentences, focused on the immediate next step:
Good - you named their concern. Now ask how you could address it within the timeline to move toward a solution.
Whatever the level, strong coaching recognizes what the learner did well before suggesting the next step, and stays encouraging rather than critical.
FAQ
Should I create multiple scenarios in one simulation?
No. Focused, single-scenario simulations produce better learning. Create a separate simulation for each scenario and let them build into a library.
How long should a simulation take to complete?
Five to fifteen minutes is the sweet spot — long enough for meaningful practice, short enough to avoid drop-off. The Minimum rounds setting (on Simulation Chat) controls how many exchanges are required before a learner can finish.
Should I use Prof Cesura and then edit the result?
Yes — that's the recommended path. The AI builder gives you a solid, well-structured draft in minutes; you then refine any page in the editor with your domain expertise.
How detailed should the persona, scenario, and theory be?
Persona profile: about 200–400 words plus three to five sample statements. Scenario story: 150–300 words. Theory: 500–2,000 words. In every case, depth and specificity matter more than length.
Can I build simulations in languages other than English?
Yes. Set the language during creation (or on General settings) and all generated content is produced in that language.
Do learners see the theory content?
No. Theory is internal — only the AI uses it. If you want learners to see a framework, put it in a step before the simulation or attach it as a Document.
Can a persona change during the simulation?
Core traits stay consistent, but the emotional state can evolve — a character might start guarded and warm up as the learner builds trust. Avoid dramatic personality switches mid-conversation.
Can a learner score high on some criteria and low on others?
Yes, and that's the point. Each criterion is scored independently from the transcript, so learners see specific strengths and specific growth areas rather than one blended verdict.
What if the AI's feedback seems wrong?
The AI scores from the actual transcript, so inaccurate feedback usually means the criterion is unclear. Make the rubric more specific about the behaviour you're looking for, and ground it in the Theory.
Can I change criteria after learners have already run the simulation?
Yes, but changes only affect new runs; completed runs keep their original feedback. For consistency, finalize criteria before sharing widely.
How do I get more reliable scores?
Make each criterion specific and observable, define the 0–100 bands in the rubric, and keep to a focused set. For tighter control of the overall grade, use the Weighted Average method.
How should I test before sharing?
Use the Testing Space button to create a private testing space and run the simulation yourself as a learner would. Fix anything that feels off, then Share. (Testing spaces use a small amount of credit and can't be shared.)