How We Know the Data Is Real
Every audience Prime Signal builds, every signal it acts on, and every match it delivers is backed by a deterministic identity graph with continuous verification. Here is exactly how it works and why it matters.
Where this page gives a number for the identity graph, that number is published by the third-party identity data provider that supplies the graph. Prime Signal does not measure those figures independently and does not publish figures for other vendors.
Why Most Identity Data Is a Placebo
Most providers score and sell identity data without ever checking whether the score was right.
Decay
Records Go Stale
Consumer records degrade as people move and change jobs. Without continuous refresh, an audience targets people who are no longer there.
IP inflation
Match Rates That Count People Twice
A match rate quoted off IP addresses counts everyone sharing the address, so the usable rate is a fraction of the headline.
Cookies
A Shrinking Signal
Browser cookie coverage keeps shrinking as platforms restrict third-party tracking, so cookie-only identity loses reach every year.
Generic
Low-Signal Domains
Most of the domains behind a huge weekly signal count are generic news and publishing sites, which say little about who is in market.
The Shared WiFi Problem
When a single visitor arrives from a coffee shop, airport, or coworking space, IP-based tools match everyone on that network to that one visit event. The result is a match rate that looks impressive on a pitch deck and delivers almost no precision in practice.
Open Inference vs. Closed-Loop Verification
Traditional Intent Platforms
- 1Signal Captured
- 2Model Scores It
- 3Score Sold to You
- 4No Verification
A one-way pipeline with no feedback step, so nothing ever confirms whether the score was right.
Our Closed-Loop Model
- 1Signal Captured
- 2Deterministic Identity Resolution
- 3Validation Against Outcomes
- 4Outcome Feeds Back Into Model
- 5Model Improves Continuously
Every classification strengthens the graph. Every outcome improves precision.
The 7-Layer Verification Stack
From raw signal capture to campaign activation, every layer is built to eliminate noise and verify truth. Open a layer to read how it works.
LAYER 1Identity Scale308M+ Verified US consumer profiles
The graph starts with scale, but scale alone means nothing without verification. 308 million profiles across 65+ data dimensions, continuously validated.
Data points per person
65+
UID2 hashed email mappings
8B+
Vetted domain partners
3.7M
LAYER 2Verification InfrastructureEvery 30 days Full database refresh cycle
This is the moat. The full record database is refreshed through NCOA every 30 days. Three-layer verification (online, offline skip-trace, and public data) holds verification rates in the 70 to 90 percent range. That range is the provider’s own reported figure, and it is the only accuracy number on this page.
Emails verified daily
10M+
Offline skip-trace datasets
20+
Verification rate (provider-reported)
70-90%
LAYER 3Signal Capture250B URLs processed per week
14 million domains watched against roughly 750,000 for the largest comparable network, about 18.7 times the coverage (derived: 14,000,000 divided by 750,000). Wider coverage means signals come from specialist, high-density sources instead of generic news and publishing noise.
Weekly intent signals
100B+
Domains watched
14M
Email engagement signals per day
400M+
LAYER 4Deterministic Identity ResolutionCookie to HEM to offline graph Resolution method
Browser-level cookie events resolve to hashed emails (HEMs), then match against a pre-built table of HEMs, IPs, and MAIDs cross-referenced with 20+ offline skip-trace datasets. Geo Frame Resolution assigns IPs to the most recently logged household coordinates via NCOA and discards geographically implausible matches, solving the shared WiFi problem.
Geo Frame Resolution
Eliminates shared WiFi inflation
True match rate
60%+
LAYER 5Signal Grading (A through D)28 across 6 groups Scoring features per identity
Every hashed email is scored across 28 features in 6 groups: Recency and Velocity, Intent Domain Signals, Email Engagement, Web Behavior, Centroid Similarity, and Cross-Feed Composites. Grade A signals require corroboration from 3+ independent sources. Grade D (generic news, low search referral) is excluded before the audience is ever delivered.
Grade A requirement
3+ independent sources
Grade D policy
Filtered out before delivery
Retrain cadence
Daily + weekly + monthly
LAYER 6Closed Feedback Loop100M+ Verification signals per month
This is what makes it a closed loop. Intent signals are not just scored and sold. They are validated against actual downstream conversions, engagement, and buying events. After 10 confirmed conversions, the model begins centroid similarity matching. After 50+, conversion fingerprint matching activates.
Conversion lift (verified vs generic)
3x
Conversions before centroid similarity matching
10
Conversion lookback window
30 days
LAYER 7Activation & Platform Match80%+ UID2 match rate (provider-reported)
Verified identities activate through UID2 hashed email mappings, which the provider reports match at 80 percent or better. Meta is the only ad platform Prime Signal activates against today; the other platforms on our roadmap will use the same mappings when they ship.
Live activation channel
Meta
Pre-built audience segments
50,000+
Verified Data vs. IP-Based Inference
- Verification Rate
Our infrastructure
70-90% (provider-reported)
Typical IP-based approach
Rarely published
- True Usable Match Rate
Our infrastructure
60%+ (provider-reported)
Typical IP-based approach
Rarely published
- NCOA Refresh
Our infrastructure
Every 30 days (full DB)
Typical IP-based approach
Unknown / stale
- Domains Watched
Our infrastructure
14M
Typical IP-based approach
About 750K for the largest comparable network
- Signal Verification
Our infrastructure
Closed feedback loop
Typical IP-based approach
None (score and sell)
- Grade D Filtering
Our infrastructure
Excluded pre-delivery
Typical IP-based approach
Included (inflates volume)
- Shared WiFi Handling
Our infrastructure
Geo Frame Resolution
Typical IP-based approach
Matched as-is (inflated)
- Ad Platform Match Rate
Our infrastructure
80%+ via UID2 (provider-reported)
Typical IP-based approach
Varies, cookie-dependent
| Metric | Our infrastructure | Typical IP-based approach |
|---|---|---|
| Verification Rate | 70-90% (provider-reported) | Rarely published |
| True Usable Match Rate | 60%+ (provider-reported) | Rarely published |
| NCOA Refresh | Every 30 days (full DB) | Unknown / stale |
| Domains Watched | 14M | About 750K for the largest comparable network |
| Signal Verification | Closed feedback loop | None (score and sell) |
| Grade D Filtering | Excluded pre-delivery | Included (inflates volume) |
| Shared WiFi Handling | Geo Frame Resolution | Matched as-is (inflated) |
| Ad Platform Match Rate | 80%+ via UID2 (provider-reported) | Varies, cookie-dependent |
What the Industry Needs Is Verified Signal
Every audience Prime Signal delivers stays traceable back to the record it came from.