{"site":{"name":"Koji","description":"AI-native customer research platform that helps teams conduct, analyze, and synthesize customer interviews at scale.","url":"https://www.koji.so","contentTypes":["blog","documentation"],"lastUpdated":"2026-05-18T13:48:30.076Z"},"content":[{"type":"documentation","id":"593b3435-79ac-47c2-942f-05a37455005e","slug":"mcp-best-practices","title":"MCP Best Practices — Getting the Most from Koji + Claude","url":"https://www.koji.so/docs/mcp-best-practices","summary":"Best practices guide for using Koji MCP with Claude. Covers prompting strategies, methodology selection, structured question design (scales, choices, rankings, probing depth), study design tips, analysis workflows with per-question aggregations, customization via koji_configure_study, common anti-patterns, token budget management, and efficient research workflow patterns.","content":"## Prompting Best Practices\n\n### Be Specific About Your Research Goal\n\nClaude creates better studies when you give it context about your business situation.\n\n**Less effective:**\n> \"Create a study about user feedback\"\n\n**More effective:**\n> \"Create a Mom Test study about why our trial-to-paid conversion rate dropped 15% last quarter. We think the onboarding flow is too complex but we want to validate this with actual users who recently churned.\"\n\nThe second prompt gives Claude enough context to generate a focused problem statement, relevant key questions, and a meaningful hypothesis to test.\n\n### Use Methodology Names\n\nWhen you know what methodology fits, say it explicitly:\n\n- \"Use the **Mom Test** approach\" — for validating ideas without bias\n- \"Apply **Jobs-to-be-Done** framework\" — for understanding user motivations\n- \"Run a **Customer Discovery** study\" — for exploring problem spaces\n- \"Set up an **Exploratory** study\" — for open-ended research\n- \"Create a **Lead Magnet** study\" — for research that doubles as lead gen\n\nIf you are unsure, just describe your goal and Claude will recommend one.\n\n### Ask for What You Actually Need\n\nInstead of pulling everything and then filtering, be targeted:\n\n| Need | Prompt |\n|------|--------|\n| Quick status check | \"How many completed interviews do I have?\" |\n| Theme overview | \"What are the top themes from my study?\" |\n| Quantitative snapshot | \"What is the average NPS score and satisfaction distribution?\" |\n| Deep dive | \"Show me the transcript from the interview with negative sentiment\" |\n| Stakeholder summary | \"Generate a report and share the executive summary\" |\n| Specific section | \"Get the recommendations section from my report\" |\n\n---\n\n## Study Design Best Practices\n\n### Start With a Hypothesis\n\nStudies with a clear hypothesis produce more actionable results:\n\n> \"Create a study to test whether users abandon checkout because of unexpected shipping costs. My hypothesis is that showing shipping estimates earlier would reduce cart abandonment by 20%.\"\n\nThis gives the AI interviewer a focused direction while still allowing for unexpected discoveries.\n\n### Mix Question Types for Richer Data\n\nThe best MCP-created studies combine qualitative and quantitative question types:\n\n- **2-3 open-ended questions** with probing depth 1-2 for qualitative depth\n- **1-2 scale questions** for benchmarkable metrics (NPS, satisfaction, effort)\n- **1 choice or ranking question** for segmentation or prioritization\n- **Optional yes/no** for screening or quick validation\n\nExample prompt:\n\n> \"Create a study with an NPS question (0-10 with anchor probing), a satisfaction scale (1-5), a multiple-choice question about feature usage, and 3 open-ended questions about onboarding experience with 2 follow-up probes each.\"\n\n### Design Structured Questions Effectively\n\n**Scale questions:**\n- Use established scales: NPS (0-10), CSAT (1-5), Likert (1-7)\n- Always label endpoints (e.g., \"1: Very dissatisfied, 5: Very satisfied\")\n- Enable anchor probing on important scales — the AI asks \"You said X, what would change that?\"\n\n**Choice questions:**\n- Keep options to 3-7 for single choice, 4-10 for multiple choice\n- Make options mutually exclusive for single choice\n- Include \"Other\" with allow_other for unexpected answers\n\n**Ranking questions:**\n- Limit to 4-7 items — more creates respondent fatigue\n- Use for feature prioritization, pain point ranking, or alternative comparison\n\n**Probing depth:**\n- 0 follow-ups: Use for demographic/screening questions\n- 1 follow-up (default): Good for most questions\n- 2-3 follow-ups: Reserve for core research questions where depth matters\n\n### Keep Interview Plans Focused\n\nThe best MCP-created studies have:\n- **5-8 total questions** mixing types (not 15 open-ended questions)\n- **Clear guardrails** from the methodology\n- **A specific target audience** defined\n- **An estimated duration** of 10-20 minutes\n\nLonger interviews lead to respondent fatigue and lower-quality insights.\n\n### Iterate Your Brief Before Publishing\n\nUse the review flow:\n1. Create the study with structured questions\n2. Ask Claude to show the full brief and question configuration\n3. Refine questions, adjust scales, tune probing depth\n4. Publish only when the brief feels right\n\nYou cannot un-publish easily, so take the time to get the brief right.\n\n---\n\n## Analysis Best Practices\n\n### Use Study Data Before Reports\n\n`koji_get_study_data` gives you structured summaries with per-question aggregations that Claude can analyze conversationally. Use this for:\n- Quick pattern spotting across themes\n- NPS averages, satisfaction distributions, and choice frequency counts\n- Feature ranking analysis\n- Hypothesis validation with both qualitative and quantitative evidence\n\nReserve `koji_generate_report` for when you need a formal, shareable document.\n\n### Ask Claude to Cross-Reference\n\nClaude shines at synthesis. Try:\n\n> \"Compare the themes from my pricing study with the themes from my onboarding study. Are there any overlapping pain points?\"\n\n> \"Looking at the NPS scores, do respondents with scores below 6 have different themes than those above 8? What do detractors have in common?\"\n\n### Use Section Filtering for Reports\n\nFor large reports, request specific sections to keep responses focused:\n\n> \"Just show me the key takeaways, charts, and recommendations from my report\"\n\nThis is more useful than pulling the entire report and scrolling through it.\n\n---\n\n## Customization Best Practices\n\n### Use koji_configure_study for All Customization\n\nAll branding, lead form, interaction mode, URL slug, and Open Graph settings are handled by a single tool: `koji_configure_study`. You can update any combination of settings in one call:\n\n> \"Set the headline to 'Customer Feedback Survey', enable voice as default, add a lead form with name and email, and set the URL slug to 'q1-feedback'.\"\n\nOnly include the fields you want to change — omitted fields keep their current values.\n\n---\n\n## Workflow Patterns\n\n### The Quick Pulse Check (2 minutes)\n```\n\"How are my active studies doing? Show me any new completed interviews since yesterday.\"\n```\n\n### The Quantitative Snapshot (3 minutes)\n```\n\"What is the average NPS score from my study? Show me the satisfaction distribution and top themes.\"\n```\n\n### The Stakeholder Prep (5 minutes)\n```\n\"Generate a report from my onboarding study and give me the top 3 takeaways I should present to the leadership team.\"\n```\n\n### The Research Sprint (30 minutes)\n```\n1. \"Create a discovery study about [topic] with NPS, satisfaction, and feature ranking questions\"\n2. \"Configure the landing page with our brand colors, a lead form, and a custom URL\"\n3. \"Import these 50 contacts from our CRM\"\n4. [Wait for interviews]\n5. \"Analyze the results — show me NPS average, feature rankings, and key themes\"\n6. \"Generate a publishable report\"\n```\n\n### The Continuous Discovery Loop (Weekly)\n```\nMonday: \"Any new interviews this week? Summarize the key themes and quantitative trends.\"\nWednesday: \"Show me transcripts from the 2 most insightful interviews with their structured answers.\"\nFriday: \"Update my report with the latest interview data.\"\n```\n\n---\n\n## Common Anti-Patterns\n\n### Creating Too Many Studies\n\nEach study should answer one research question. Do not create a study for every feature idea — group related questions into a single study with the right methodology and use structured questions to capture multiple data points.\n\n### Skipping the Brief Review\n\nPublishing a study without reviewing the interview questions and structured question configuration leads to unfocused conversations and misconfigured scales. Always review before you publish.\n\n### Overloading Structured Questions\n\nDo not add 10 scale questions. Structured questions should complement the qualitative conversation, not replace it. Aim for 2-3 structured questions mixed with open-ended questions.\n\n### Reading Every Transcript\n\nUse summaries, theme analysis, and per-question aggregations for pattern detection. Only deep-dive into transcripts when you find something worth exploring.\n\n### Generating Reports Too Early\n\nWait until you have at least 5 completed interviews. Reports from 1-2 interviews will not show reliable patterns or meaningful quantitative distributions.\n\n### Overloading Lead Forms\n\nKeep pre-interview forms to 2-3 fields. Every additional field reduces completion rates.\n\n---\n\n## Token Budget Tips\n\nMCP responses are constrained by context window size. Here are strategies for efficient use:\n\n1. **Use section filters** when getting reports — request only the sections you need\n2. **Paginate exports** — transcripts are limited to 10 per request for a reason\n3. **Use study data** instead of individual transcripts for overview analysis\n4. **Ask specific questions** rather than requesting \"everything\"\n5. **Request specific per-question metrics** rather than all study data\n\n---\n\n## Next Steps\n\n- **[Tool Reference](/docs/mcp-tool-reference)** — Detailed parameter reference for all 15 tools\n- **[Structured Questions Guide](/docs/structured-questions-guide)** — Deep dive into question types and probing\n- **[PM Workflow Guide](/docs/mcp-workflow-product-managers)** — Role-specific workflow\n- **[MCP Overview](/docs/mcp-overview)** — Full integration overview\n\n## Further reading on the blog\n\n- [Getting Customer Feedback That Actually Drives Product Decisions](/blog/getting-customer-feedback-that-actually-drives-product-decisions) — Customer feedback is only valuable when it leads to action. Learn proven methods for collecting, analyzing, and acting on customer insights \n- [Getting Started with Customer Research: A Beginner's Guide](/blog/getting-started-with-customer-research-a-beginner-s-guide) — A practical, step-by-step guide for Product Managers, UX Researchers, and Founders who want to start doing customer research today and build\n\n<!-- further-reading:blog -->\n","category":"Claude & MCP Integration","lastModified":"2026-05-13T00:26:36.807295+00:00","metaTitle":"MCP Best Practices — Effective AI Research Workflows | Koji Documentation","metaDescription":"Tips and patterns for using Koji MCP effectively with Claude. Better prompts, methodology selection, analysis workflows, anti-patterns to avoid, and strategies for efficient AI-powered research.","keywords":["MCP best practices","AI research tips","Koji MCP tips","effective AI interviews","research workflow optimization","Claude research prompts","qualitative research AI best practices"],"aiSummary":"Best practices guide for using Koji MCP with Claude. Covers prompting strategies, methodology selection, structured question design (scales, choices, rankings, probing depth), study design tips, analysis workflows with per-question aggregations, customization via koji_configure_study, common anti-patterns, token budget management, and efficient research workflow patterns.","aiPrerequisites":["Koji MCP connected","Basic familiarity with MCP tools"],"aiLearningOutcomes":["Write effective prompts for Koji MCP tools","Choose appropriate research methodologies","Build efficient research workflow patterns","Avoid common anti-patterns","Manage token budgets for large studies"],"aiDifficulty":"intermediate","aiEstimatedTime":"10 minutes"}],"pagination":{"total":1,"returned":1,"offset":0}}