OnTracAI

Note Capture & Normalization

Purpose

This step preserves truth.

The goal is to:

  • Capture exactly what was said
  • Avoid premature summarization
  • Create a clean bridge from raw conversations to structured analysis

High-level summaries are not enough. Direct quotes matter.

Raw Capture

What Must Be Captured

ArtifactRequirement
Full audioEntire interview
Verbatim transcriptWord-for-word, unedited
Speaker attributionWho said what
TimestampingOptional but helpful

Do not rely on memory or post-hoc notes.

Recommended Tooling

ToolWhy
GranolaRecords off local audio driver; no third bot
Native platform recordingAcceptable fallback
External note-taker botsAvoid if possible

Granola is preferred because it is invisible to the meeting and produces clean transcripts.

Recording Hygiene

Always:

  • Inform participants the meeting is being recorded
  • State purpose clearly: internal analysis and synthesis
  • Confirm comfort before proceeding

This builds trust and avoids downstream issues.


Why Raw Transcripts Matter

ReasonImpact
Exact languageReveals real pain
Direct quotesHigh credibility in exec readouts
Avoids biasPrevents over-interpretation
Cross-validationConfirms themes across interviews

Executive decks land harder when they include:

"We spend more time validating data than using it."

That power is lost in summaries.


Post-Interview Processing

Within 24 hours of the interview:

StepAction
Store transcriptCentral shared location
Label fileFunction + name + date
Flag quotesHighlight strong language
Add contextNote tone and emphasis

Do not wait until synthesis week.


Normalization Framework

Every interview is normalized into the same structure.

Normalized Note Template

SectionWhat Goes Here
Role & scopeAccountability and decisions
Core workflowsRepeated or manual work
Pain pointsFriction, failures, delays
Systems & dataTools, SoR vs SoT
Reporting and latencyFreshness, mismatches
ConstraintsSecurity, policy, risk
Desired outcomesWhat "better" means
Notable quotesVerbatim excerpts

This structure mirrors the interview framework.


Quotes Handling

Guidelines:

  • Use exact wording
  • Preserve context
  • Avoid paraphrasing
  • Attribute by role, not name in deliverables

Example:

"We don't trust the number until we've checked it three times." (Finance)


AI Assistance

AI may be used to:

  • Segment transcripts by topic
  • Suggest candidate quotes
  • Flag repeated phrases

AI may not:

  • Replace raw transcripts
  • Rewrite quotes
  • Decide what matters

Output of This Step

OutputUsed For
Raw transcriptsEvidence
Normalized notesAnalysis
Quote bankExecutive deck
Observation listOpportunity framing

If normalization feels tedious, it's working. This is where rigor is established.

Next step is Opportunity Identification, where normalized notes become scorable inputs.