Purpose
DBB heuristics determine when something is a decision worth recording. For SMBs, heuristics are intentionally:- Conservative
- Explainable
- Low-noise
Confidence-based capture model
Every potential decision signal receives a confidence score. Only signals above the persistence threshold are recorded. Confidence is derived from multiple lightweight features.Core heuristic signals
1. Explicit resolution language
Phrases indicating closure or commitment:- “We will…”
- “Let’s proceed with…”
- “Approved”
- “Final decision”
2. Convergence after discussion
Signals increase when:- Multiple viewpoints collapse into one
- Alternatives stop appearing
- Follow-up actions align
3. Approval or status change
Examples:- Document approval
- Merge or publish action
- Formal sign-off
4. Rejection of alternatives
Explicitly rejected options strengthen confidence:- “We won’t do X”
- “Option A is not viable”
5. Temporal persistence
If a topic:- Reappears across days or weeks
- Retains similar framing
Negative signals (noise reduction)
DBB suppresses capture when:- Discussions are exploratory
- Language remains hypothetical
- No convergence is detected
- The same topic oscillates without closure
False positives handling
When a captured decision is marked incorrect:- DBB records the correction as an event
- Original capture is preserved
- Confidence weighting is adjusted
SMB-specific constraints
Compared to Enterprise:- Fewer policy layers
- No manual approval workflows
- Faster decay of low-confidence signals
- Smaller evaluation windows
Summary
DBB heuristics for SMBs prioritize:- Precision over recall
- Trust over automation
- Persistence over activity tracking