What is the Voice Corpus and how do I record it?

Overview

The Voice Corpus is a set of recorded sessions where you narrate your real decision-making on live records from your pipeline. It is the primary input for Phase 1 setup and typically takes 3–5 hours across two or three sessions.

You do not need to prepare anything. A Fulcrum consultant runs the session and brings the records. You work from real contacts, accounts, and opportunities in your CRM — not templates or hypotheticals.

During the sessions, the consultant asks you to think out loud:

• Why you would pursue one record
• Why you would pass on another
• What would need to change for your answer to be different

All sessions are recorded and transcribed, then converted into a structured extraction of your reasoning — not just a raw transcript.

---

What the Voice Corpus captures

The output focuses on how you actually evaluate pipeline, including:

• Tier definitions
  • Tier 1 opportunities
  • Tier 2 opportunities
  • Fully disqualified records
• Signal weighting
• Disqualifier list
• ICP signals
  • Industry
  • Company stage
  • Role and seniority
  • Timing
  • Trigger events

---

How to record the Voice Corpus

You do not need to prepare anything in advance.

1. Scheduling
   • Plan for 3–5 total hours, usually split across 2–3 sessions.
2. Session setup
   • A Fulcrum consultant joins the call and shares or accesses real CRM records (contacts, accounts, opportunities).
   • You confirm you’re looking at live pipeline data, not sample or test records.
3. Thinking out loud
   • Would you pursue this opportunity? Why?
   • Would you deprioritize or disqualify it? Why?
   • What specific signals are driving your judgment?
   • What would need to change (data, context, history) for your answer to be different?
4. Focus on edge cases
   • Situations where your usual rules conflict
   • Cases where context overrides your normal pattern
   • Deals that are exceptions to your standard ICP or rules

These edge cases are often the most informative training data, because they reveal how you actually trade off signals in real life.

1. Recording and extraction
   • The sessions are recorded and transcribed.
   • Fulcrum converts the raw transcript into a structured pattern library, including:
     • Tier definitions
     • Signal weights and combinations
     • Disqualifiers
     • ICP and context signals

---

Why honesty about edge cases matters

The goal is not to capture an idealized, textbook version of your process. The system needs your real-world judgment, including:

• When you break your own rules
• When a strong positive signal is overridden by a disqualifier
• When a weak-fit account becomes attractive due to timing or triggers

This is what allows Vantage to mirror your true evaluation standard, not just your stated one.

---

What happens after recording

• During Phase 1, the Voice Corpus is processed into the initial pattern library that powers scoring and recommendations.
• After Phase 1, the Voice Corpus is not reused for ongoing training.
  • Continued accuracy improvements come from the correction cadence — your ongoing feedback on Vantage’s scores and recommendations.

Most clients only need to record the Voice Corpus once. A partial re-recording may be useful if your evaluation standard changes significantly, for example:

• Entering a new market
• Targeting a new ICP
• Launching a new service line

Related: How Vantage learns your standard · The correction cadence · How long setup takes