What Is Accuracy Scoring?
Accuracy is one of six metrics in your BPI score. It measures how factually correct AI responses are about your brand.
A high accuracy score means AI is stating correct facts. A low score means hallucinations or outdated information.
Accuracy is weighted at 20% of your total BPI - the highest weight of any metric - because incorrect facts directly impact trust and conversion.
Scoring Methodology
For each AI response, VectorGap:
1. **Extracts factual claims** (numbers, dates, features, etc.)
2. **Matches against Knowledge Base** entries
3. **Scores each claim:**
- Verified correct: Full points
- Partially accurate: Partial points
- Incorrect/contradicted: No points
- Unverifiable: Neutral (no impact)
4. **Calculates weighted average** based on claim importance
The final accuracy score is 0-100, where 100 means every verifiable claim matched your Knowledge Base.
Improving Accuracy
If your accuracy score is low:
**First, check your Knowledge Base:**
- Is it complete? Missing info can't be verified
- Is it current? Outdated KB entries cause false flags
- Is it specific? Vague entries are hard to match
**Then, address public information:**
- Update your website with current facts
- Ensure key pages are crawlable
- Fix inconsistencies across your web presence
- Submit corrections to third-party sites
**Track progress:**
- Run audits after major content updates
- Expect gradual improvement (4-8 weeks)
- Set alerts for accuracy score changes
Accuracy vs. Other Metrics
Accuracy measures facts. Other metrics measure different things:
**Accuracy:** Is the information correct?
**Credibility:** Does AI sound authoritative?
**Coverage:** Are key details mentioned?
You can have high accuracy but low coverage (correct but incomplete). Or high coverage but low accuracy (comprehensive but wrong).
Balance matters. Prioritize accuracy first - incorrect information is worse than missing information.