> ## Documentation Index > Fetch the complete documentation index at: https://docs.mavera.io/llms.txt > Use this file to discover all available pages before exploring further. # Focus Groups > AI-powered synthetic focus group research Focus Groups let you run simulated audience research at scale. Define your questions, choose your personas, set a sample size, and get quantitative scores plus qualitative reasoning from a synthetic panel — in minutes instead of weeks. A single Focus Group can simulate **N=25 to 200+** respondents across multiple audience segments, combining question types like NPS, Likert scales, open-ended responses, and ranking exercises into one cohesive study. Focus Groups are **asynchronous**. You create a study, then poll for results. Most studies complete in 30–120 seconds depending on sample size and question count. ## Key Features NPS, Likert, multiple choice, open-ended, rating, ranking, semantic differential, and more Mix multiple personas for diverse, segment-specific responses From 10 to 200+ respondents per study Get both statistical summaries and individual respondent reasoning ## Focus Group Lifecycle ```mermaid theme={"dark"} sequenceDiagram participant App as Your Application participant API as Mavera API participant Engine as Research Engine App->>API: POST /focus-groups (questions, personas, sample_size) API-->>App: { id, status: "PROCESSING" } loop Poll for completion App->>API: GET /focus-groups/{id} API-->>App: { status: "PROCESSING" } end API->>Engine: Generate N responses per persona Engine->>Engine: Aggregate scores, analyze themes Engine-->>API: Results ready App->>API: GET /focus-groups/{id} API-->>App: { status: "COMPLETED", results: [...] } ``` ## Creating a Focus Group Select 1–10 persona IDs that represent your target audience segments. The more diverse your panel, the richer the cross-segment analysis. Write 1–20 questions using the supported question types. Mix quantitative (NPS, Likert) with qualitative (open-ended) for well-rounded insights. Choose a sample size (10–200+). Larger samples give more statistical confidence but cost more credits. The study runs asynchronously. Poll the GET endpoint until `status` is `"COMPLETED"`. ### Full Creation Example ```python Python theme={"dark"} import requests response = requests.post( "https://app.mavera.io/api/v1/focus-groups", headers={"Authorization": "Bearer mvra_live_your_key_here"}, json={ "name": "Product Launch Feedback", "sample_size": 50, "persona_ids": [ "gen_z_consumer", "millennial_professional", "gen_x_parent" ], "questions": [ { "question": "How likely are you to recommend this product to a friend?", "type": "NPS", "order": 1 }, { "question": "The product design feels modern and appealing.", "type": "LIKERT", "order": 2 }, { "question": "Which features interest you most?", "type": "MULTIPLE_CHOICE", "options": ["Speed", "Design", "Price", "Sustainability", "Brand reputation"], "order": 3 }, { "question": "Rate the overall value for money.", "type": "RATING", "order": 4 }, { "question": "Rank these factors by importance when choosing this product.", "type": "RANKING", "options": ["Price", "Quality", "Brand", "Reviews", "Sustainability"], "order": 5 }, { "question": "Would you switch from your current product to this one?", "type": "YES_NO", "order": 6 }, { "question": "How innovative vs traditional does this product feel?", "type": "SEMANTIC_DIFFERENTIAL", "left_label": "Traditional", "right_label": "Innovative", "order": 7 }, { "question": "What would make you more likely to purchase?", "type": "OPEN_ENDED", "order": 8 } ], "workspace_id": "your_workspace_id" } ) focus_group = response.json() print(f"Focus Group ID: {focus_group['id']}") print(f"Status: {focus_group['status']}") ``` ```javascript JavaScript theme={"dark"} const response = await fetch("https://app.mavera.io/api/v1/focus-groups", { method: "POST", headers: { Authorization: "Bearer mvra_live_your_key_here", "Content-Type": "application/json", }, body: JSON.stringify({ name: "Product Launch Feedback", sample_size: 50, persona_ids: [ "gen_z_consumer", "millennial_professional", "gen_x_parent", ], questions: [ { question: "How likely are you to recommend this product to a friend?", type: "NPS", order: 1, }, { question: "The product design feels modern and appealing.", type: "LIKERT", order: 2, }, { question: "Which features interest you most?", type: "MULTIPLE_CHOICE", options: ["Speed", "Design", "Price", "Sustainability", "Brand reputation"], order: 3, }, { question: "Rate the overall value for money.", type: "RATING", order: 4, }, { question: "What would make you more likely to purchase?", type: "OPEN_ENDED", order: 5, }, ], workspace_id: "your_workspace_id", }), }); const focusGroup = await response.json(); console.log(`Focus Group ID: ${focusGroup.id}`); console.log(`Status: ${focusGroup.status}`); ``` ## Question Types | Type | Description | Response Format | Best For | | ----------------------- | ---------------------------- | ------------------------------- | --------------------------------------- | | `NPS` | Net Promoter Score (0–10) | Numeric score + reasoning | Measuring recommendation likelihood | | `LIKERT` | 5-point agreement scale | Scale value (1–5) + explanation | Gauging attitudes and opinions | | `MULTIPLE_CHOICE` | Select from defined options | Selected option(s) + reasoning | Feature preference, forced choices | | `OPEN_ENDED` | Free-form response | Detailed text | Exploratory insights, "why" questions | | `RATING` | Star rating (1–5) | Rating + explanation | Quick satisfaction scoring | | `YES_NO` | Binary choice | Yes/No + reasoning | Go/no-go decisions, adoption intent | | `RANKING` | Order options by preference | Ranked list + reasoning | Priority ordering, trade-off analysis | | `SLIDER` | Numeric scale (custom range) | Value + explanation | Fine-grained measurement | | `MATRIX` | Multiple items on same scale | Matrix of responses | Comparing multiple attributes | | `SEMANTIC_DIFFERENTIAL` | Between two opposites | Position (1–7) + explanation | Brand perception, positioning | | `CONJOINT` | Trade-off analysis | Preference data | Pricing and feature optimization | | `MAXDIFF` | Best/worst selection | Selection + reasoning | Finding most/least important attributes | **Recommended mix:** Start with 1–2 quantitative questions (NPS, Likert, Rating) to get scoreable data, then add 1–2 qualitative questions (Open-ended) for depth. Finish with a Ranking or Semantic Differential for nuanced insight. Keep total questions under 10 for best quality. ## Polling for Results Focus Groups run asynchronously. Poll the status endpoint until the study completes. ```python Python theme={"dark"} import time import requests focus_group_id = "fg_abc123" headers = {"Authorization": "Bearer mvra_live_your_key_here"} while True: response = requests.get( f"https://app.mavera.io/api/v1/focus-groups/{focus_group_id}", headers=headers ) result = response.json() if result["status"] == "COMPLETED": print("Focus group complete!") break elif result["status"] == "FAILED": print(f"Focus group failed: {result.get('error')}") break else: print(f"Status: {result['status']}... waiting") time.sleep(5) ``` ```javascript JavaScript theme={"dark"} const focusGroupId = "fg_abc123"; async function pollForResults() { while (true) { const response = await fetch( `https://app.mavera.io/api/v1/focus-groups/${focusGroupId}`, { headers: { Authorization: "Bearer mvra_live_your_key_here" } } ); const result = await response.json(); if (result.status === "COMPLETED") { console.log("Focus group complete!"); return result; } else if (result.status === "FAILED") { throw new Error(`Focus group failed: ${result.error}`); } console.log(`Status: ${result.status}... waiting`); await new Promise((r) => setTimeout(r, 5000)); } } const results = await pollForResults(); ``` **Status values:** | Status | Meaning | | ------------ | ------------------------------ | | `PENDING` | Study created, not yet started | | `PROCESSING` | Responses being generated | | `COMPLETED` | All results available | | `FAILED` | An error occurred | ## Interpreting Results ### Aggregate Scores Each question returns an aggregate summary alongside individual responses. ```python theme={"dark"} result = response.json() for question_result in result["results"]: print(f"\nQuestion: {question_result['question']}") print(f"Type: {question_result['type']}") print(f"Summary: {question_result['summary']}") if question_result["type"] == "NPS": print(f" NPS Score: {question_result['nps_score']}") print(f" Promoters: {question_result['promoters']}%") print(f" Passives: {question_result['passives']}%") print(f" Detractors: {question_result['detractors']}%") elif question_result["type"] == "LIKERT": print(f" Mean: {question_result['mean']}") print(f" Median: {question_result['median']}") print(f" Std Dev: {question_result['std_dev']}") elif question_result["type"] == "MULTIPLE_CHOICE": for option, count in question_result["distribution"].items(): print(f" {option}: {count} votes") elif question_result["type"] == "RANKING": for rank, item in enumerate(question_result["average_ranking"], 1): print(f" #{rank}: {item}") ``` ### Individual Responses Drill into individual respondent data for qualitative depth. ```python theme={"dark"} for question_result in result["results"]: print(f"\nQuestion: {question_result['question']}") for resp in question_result["responses"][:5]: print(f"\n Persona: {resp['persona_name']}") if "score" in resp: print(f" Score: {resp['score']}") print(f" Reasoning: {resp['reasoning'][:200]}...") ``` ### Cross-Segment Analysis Compare how different persona segments responded to the same question. ```python theme={"dark"} nps_results = result["results"][0] segment_scores = {} for resp in nps_results["responses"]: segment = resp["persona_name"] if segment not in segment_scores: segment_scores[segment] = [] segment_scores[segment].append(resp["score"]) for segment, scores in segment_scores.items(): avg = sum(scores) / len(scores) print(f"{segment}: avg NPS = {avg:.1f}") ``` ## Response Format ```json theme={"dark"} { "id": "fg_abc123", "name": "Product Launch Feedback", "status": "COMPLETED", "sample_size": 50, "personas": [ { "id": "gen_z_consumer", "name": "Gen Z Consumer" }, { "id": "millennial_professional", "name": "Millennial Professional" }, { "id": "gen_x_parent", "name": "Gen X Parent" } ], "questions": [ { "id": "q1", "question": "How likely are you to recommend this product?", "type": "NPS", "order": 1 } ], "results": [ { "question_id": "q1", "question": "How likely are you to recommend this product?", "type": "NPS", "nps_score": 42, "promoters": 55, "passives": 32, "detractors": 13, "summary": "Strong positive sentiment overall. Gen Z respondents showed the highest likelihood to recommend, driven by the product's sustainability messaging...", "responses": [ { "persona_id": "gen_z_consumer", "persona_name": "Gen Z Consumer", "score": 9, "reasoning": "I'd definitely recommend this — the sustainability angle is genuine, not performative, and the design feels like it was made for my generation." }, { "persona_id": "gen_x_parent", "persona_name": "Gen X Parent", "score": 7, "reasoning": "Good product, fair price. I'd mention it if someone asked but probably wouldn't go out of my way to recommend it." } ] } ], "usage": { "credits_used": 150 } } ``` ## When to Use Focus Groups vs Chat vs Mave | Scenario | Best Tool | Why | | ----------------------------------------------------- | ---------------- | ------------------------------------------------------------- | | "How would Gen Z react to this tagline?" | **Chat** | Single persona, single question — fast and cheap | | "Compare reactions across 4 segments with NPS scores" | **Focus Groups** | Multiple personas, quantitative data, statistical aggregation | | "What's the market size for EV charging in Europe?" | **Mave** | Research question needing web data and citations | | "Test 3 ad concepts with 50 respondents each" | **Focus Groups** | Structured comparison at scale | | "Get a quick gut-check from a millennial" | **Chat** | Conversational, low-cost, real-time | | "Deep competitive analysis with sources" | **Mave** | Multi-source research with validation | ## Credits and Pricing Credit cost depends on **sample size**, **question count**, and **question complexity**. | Sample Size | 3 Questions | 5 Questions | 10 Questions | | ----------- | --------------- | --------------- | ---------------- | | 10–25 | 30–50 credits | 50–75 credits | 75–120 credits | | 26–50 | 50–85 credits | 75–125 credits | 125–200 credits | | 51–100 | 85–140 credits | 125–200 credits | 200–350 credits | | 100+ | 140–250 credits | 200–350 credits | 350–500+ credits | Open-ended and conjoint questions cost more per respondent than NPS or yes/no questions because they require more generation. ## Listing Focus Groups Retrieve all focus groups in your workspace. ```python Python theme={"dark"} response = requests.get( "https://app.mavera.io/api/v1/focus-groups", headers={"Authorization": "Bearer mvra_live_your_key_here"} ) for fg in response.json()["data"]: print(f"{fg['name']} — {fg['status']} — {fg['sample_size']} respondents") ``` ```bash cURL theme={"dark"} curl https://app.mavera.io/api/v1/focus-groups \ -H "Authorization: Bearer mvra_live_your_key_here" ``` ## Best Practices Combine NPS or Likert questions (numbers you can chart) with open-ended questions (explanations you can quote). This gives you both statistical confidence and narrative depth. Include personas that represent your actual target audience segments. A panel of Gen Z + Millennial + Gen X gives generational breadth. Adding a Budget Shopper and a Luxury Consumer gives psychographic breadth. 5–8 well-crafted questions yield better results than 15+ rushed ones. Each additional question adds cost and can cause response fatigue in the model. Focus Groups are ideal for validating messaging, product concepts, pricing strategies, and feature prioritization before committing real-world budget. Create two focus groups with the same personas but different stimuli (e.g., two taglines, two product descriptions) to see which resonates more. After a Focus Group reveals a surprising result, follow up with a Chat conversation using the same persona to explore the "why" in depth. Focus Group results are simulated perspectives, not real consumer data. They are best used for directional insights, concept screening, and hypothesis generation. Validate high-stakes decisions with real audience research. ## Next Steps Step-by-step tutorial with complete scripts Get started in 15 minutes Choose and create personas for your panels Full API specification