Mirror Question Technique for AIAS

Implement conversational mirroring to extract specific budget and timeline intel from evasive prospects without breaking rapport.

Implementation Levels

• L1 -- Note it: Document the mirror question technique and question variants in the knowledge base for reference.
• L2 -- Build it: Update AIAS system prompt with mirror question protocol and alternative phrasings for budget/timeline discovery.
• L3 -- Go deep: Build automated evasive-answer classification using GPT-4.1-mini and structured question variant banks for all qualification topics.

What This Video Covers

Key Insights

• {
  "category": "sales",
  "theme": "Mirror questions technique for evasive prospects",
  "summary": "Andres Contreras breaks down the 'mirror question' technique for handling prospects who talk around your questions or give vague answers. The two-step process involves calling out the non-answer, then immediately rephrasing the same question differently to extract specific intel without making the prospect feel interrogated.",
  "video_breakdown": {
  "hook": "Opens with the relatable frustration of prospects who 'won't stop fucking talking' and 'keep talking about the wrong things'",
  "main_points": [
  "Define the problem: prospects either skirt questions intentionally or don't understand them, giving wrong answers",
  "Introduce 'mirror question' technique with two distinct parts",
  "Part One - The Callout: You must explicitly acknowledge when someone doesn't answer, otherwise you train them that they can dodge questions",
  "Part Two - Reframe: Ask the same question in a different way so they don't feel treated like a toddler, but you still get the intel needed",
  "Practical example using fitness sales: Instead of accepting vague answers about weight loss goals, prepare multiple question variants to uncover the real motivation (speed vs confidence vs appearance)",
  "Preparation is key: anticipate the 3-4 possible answer variants to your question, then create different questions that could surface each specific answer"
  ],
  "key_quotes": [
  "What are they learning? That you're a bitch they can not give you the answer and you'll just move on like nothing ever happens",
  "Mirror question is whenever you ask me the question and rather than give me the correct answer they either skirt it whether it's intentional or they just don't understand the questions then they give you the wrong answer",
  "Ask the same question in a different way that way they feel like you're not treating them like a toddler however you're still asking the question you need to ask to get the answer you needed"
  ],
  "creator_context": "Andres Contreras appears to be a sales coach/infopreneur based on the course pitch in comments. The setting suggests established authority (professional studio, expensive watch visible). Teaching direct response sales tactics likely aimed at fitness coaches or high-ticket service providers."
  },
  "detailed_notes": {
  "what_it_is": "A conversational control technique for sales calls/prospecting where you (1) explicitly flag when a prospect hasn't answered your question, and (2) immediately ask a differently-phrased version of the same question to extract specific qualifying information without breaking rapport.",
  "how_useful": "Directly applicable to AIAS conversation flows. Our AI currently qualifies leads via SMS - we can add 'mirror question' logic to the prompt chain when the LLM detects vague/non-answers about budget, timeline, or business pain points. Also useful for Dylan's TFWW sales calls when prospects say 'I need a website' without specifying why/what's broken.",
  "how_not_useful": "The aggressive 'callout' framing ('they learn you're a bitch') needs softening for SMS-based AI conversations - too confrontational for text. The technique assumes real-time voice conversation where tone can soften the callout; in SMS this requires careful phrasing to avoid seeming robotic.",
  "target_audience": "Dylan for TFWW sales calls, AIAS dev team for prompt engineering improvements"
  },
  "key_insights": [
  "We can implement 'answer validation' logic in AIAS where Claude checks if the prospect's response actually contains the requested data point (budget/timeline/pain) before proceeding the conversation flow",
  "Create 'question variant banks' for common qualification questions - if the prospect doesn't answer version A about budget, the AI switches to version B (e.g., 'What's your investment range?' vs 'Have you set aside funds for this project?')",
  "The specific example reveals we should train AI to probe for emotional/specific outcomes rather than accepting functional answers - 'I need a website' becomes 'Is this about getting more leads, looking more professional, or saving time on admin?'",
  "Add a 'non-answer detected' classification step in the AI pipeline using GPT-4.1-mini to classify responses as 'answered', 'partial', or 'evasive' before determining the follow-up prompt",
  "For TFWW discovery calls: When prospects say 'just looking' or 'gathering info', use the mirror technique - 'Totally understand you're exploring - to make sure I point you to the right resources, are you comparing us to other builders, or trying to figure out if you actually need a new site right now?'"
  ],
  "business_applications": [
  {
  "area": "AI conversation engineering",
  "recommendation": "Update AIAS system prompt to include mirror question logic: First acknowledge non-answers ('I want to make sure I understand...'), then ask alternative formulations of the same qualification question. Add example question variants for budget, timeline, and decision-making authority to the prompt context.",
  "target_system":

Analysis Notes

Cost Breakdown →