Signal Pipeline
From raw signal to published score
VECTR runs a continuous ingestion pipeline across multiple independent sources. Here is exactly what happens from signal detection to score update.
01
Signal ingestion multi-source
Bots monitor public buyer feedback continuously across community forums, verified review platforms, and direct GunScore field reports. Each source carries a different base credibility weight.
02
Entity resolution NLP matching
Each signal is matched to a retailer in our database using fuzzy name matching, domain extraction, and alias lookup. Ambiguous matches are held for human review rather than auto-assigned.
input: "ordered from primaryarms.com" → entity: Primary Arms (id: 8, confidence: 0.97)
03
Data integrity scan nightly
A nightly PM agent checks for misattributed signals, thin-data anomalies, single-source outliers, and category mismatches. Flagged signals are reviewed before they influence any score.
04
Source weighting + recency decay weightedtime-decayed
Signals are weighted by source credibility and decayed by age. A verified purchase review from last month outweighs an anonymous forum post from 2020. Historical conduct signals (documented fraud, community bans) are permanent and carry 5x weight with no decay.
05
VECTR score calculation 5-dimension
Weighted signals feed into five independent scoring dimensions. A minimum of 5 verified signals is required before a score is published. Retailers below this threshold display "Insufficient Data" — never a fabricated low score.
Scoring Dimensions
Five independent signals, one trust score
VECTR scores five dimensions independently and combines them into a single 0–100 trust score. Each dimension is weighted differently based on its predictive value for buyer outcomes.
V
Verified
Cross-source signal consistency. Do multiple independent sources corroborate the same picture?
E
Experience
Buyer-reported outcomes: order accuracy, shipping condition, what arrived vs. what was ordered.
C
Consistency
Signal volume and variance over time. High variance with low volume scores differently than steady positive history.
T
Transparent
Undisclosed fees, bait-and-switch pricing, and checkout surprises. Zero undisclosed-fee reports is the single biggest score lever.
R
Responsive
Post-complaint behavior. Retailers who resolve issues publicly improve this dimension. Unclaimed profiles with unresolved complaints score lowest.
# Simplified scoring formula
base_score = 45 # silence is neutral, not a penalty
signal_weight = source_multiplier × recency_decay × volume_factor
vectr_score = CLAMP(base_score + weighted_avg(signals) × 30, 0, 98)
# historical_conduct signals: 5x weight, no recency decay
# minimum 5 verified signals required before score is published
base_score = 45 # silence is neutral, not a penalty
signal_weight = source_multiplier × recency_decay × volume_factor
vectr_score = CLAMP(base_score + weighted_avg(signals) × 30, 0, 98)
# historical_conduct signals: 5x weight, no recency decay
# minimum 5 verified signals required before score is published
Source Weighting
Not all signals count the same
Signal credibility is determined by how verifiable the source is. Verified purchase reviews carry more weight than anonymous posts by design.
Community Forums
Public posts from monitored firearms subreddits and forum threads. High volume, broad coverage, anonymous by nature.
Baseline — 1x
Verified Review Platforms
Reviews tied to confirmed purchases on third-party verified platforms. Purchase verification reduces fake review risk.
3x baseline
GunScore Field Reports
Submitted directly by buyers through GunScore, linked to a specific retailer and transaction details.
Highest trust tier
Score Tiers
What the number means
80–100
✓ Trusted
Consistent positive signal volume across multiple independent sources. Low or zero unresolved complaints.
60–79
Approved
Generally positive signal with moderate volume, or minor resolved issues. Worth buying from with normal due diligence.
40–59
Mixed
Baseline territory. Limited signal volume or a roughly even split of positive and negative reports.
0–39
⚠ Caution
Pattern of unresolved complaints, hidden fees, or documented conduct issues. Proceed with significant caution.
Data Infrastructure
Running 24/7
The VECTR pipeline runs continuously through an automated intelligence network. Scores reflect the current state of public buyer feedback, not a static snapshot.
Community Signal Monitor
Continuously indexes buyer feedback from across the firearms community, identifying retailer mentions and classifying signal sentiment in real time.
Continuous · Near real-time indexing
Trust Intelligence Engine
Synthesizes multi-source buyer signals into verified trust scores using weighted aggregation, recency decay, and cross-source validation.
Nightly · Full recalculation
Data Integrity Agent
Automated nightly audit of signal attribution, source diversity, and data quality. Anomalies are flagged for human review before influencing any score.
Nightly · Human-in-the-loop review
Score Change Monitor
Tracks daily score deltas across all retailer profiles. Significant movements trigger verified alerts to claimed retailer contacts.
Nightly · 5-point delta threshold
No retailer can pay to improve their VECTR score. Not for a higher tier, not to remove a complaint, not for preferred placement. Every signal on every profile links back to its original public source.
Browse Retailer Scores →