Identity Graph: MAID → HEM and Cross-Device Resolution
GSDSI's identity graph links mobile advertising identifiers, hashed emails, and CTV IDs so privacy-safe teams can onboard first-party audiences, stitch fragmented touchpoints, and measure across channels, with match rates and refresh cadence you can audit on your own seed file.
~200M+ US MAID→HEM
Deterministic device-to-email linkages for onboarding and measurement.
700M+ global pairs
International coverage across 150+ countries for multi-market programs.
Consent-aware posture
Documented sourcing, permitted uses, and opt-out handling for enterprise procurement.
Since 2018
Privacy-safe enterprise data
VOSB & SDVOSB
U.S. Veteran-Owned, Service-Disabled
SCC Transfers
EU/UK/CH per privacy policy
GDPR · CCPA · CPRA
Consent-first data sourcing
What is an identity graph?
An identity graph is a set of identifier linkages, not a single ID, that tells you when a MAID, hashed email, CTV household ID, or IP hash likely belongs to the same person or household. Enterprise buyers use graphs for CRM onboarding, cross-channel measurement, deduplication, and suppression. The graph is only as valuable as its measurable match rate on your audience, its refresh cadence, and the compliance envelope that governs activation.
GSDSI ships identity as operational data products: the MAID feed is the backbone for device-to-email resolution; the core email file and Euclidean feed extend matching for onboarding and clustering; CTV/ACR data bridges TV exposure into the same measurement stack.
Core identifiers & link semantics
Procurement teams should insist on a field dictionary and confidence tiers, not a deck average. Typical GSDSI identity deliveries expose the keys below; exact schema varies by license and activation surface.
| Field / edge | Role in the graph |
|---|---|
| maid | Mobile advertising ID (IDFA / GAID) for app and mobile web activation |
| hem | Hashed email (SHA-256) for CRM onboarding and deterministic matching |
| ctv_id | Connected-TV / ACR household or device identifier for TV exposure joins |
| ip_hash | Household- or device-level IP hash where deterministic email is unavailable |
| household_id | Resolved household cluster for dedupe, frequency caps, and reach |
| link_type | Deterministic vs probabilistic edge with disclosed confidence tier |
| last_seen | Freshness timestamp for decay analysis and suppression rules |
GSDSI identity stack
MAID feed
~200M+ US MAID→HEM linkages and 700M+ international pairs across 150+ countries.
View specs →Core email file
Verified HEM spine for CRM onboarding, deliverability-tested, identity-linked at scale.
View specs →Euclidean feed
Cross-device distance signal for dedupe, clustering, and household-adjacent matching.
View specs →CTV / ACR feed
~13-14M monthly CTV IDs to bridge TV exposure into the same graph.
View specs →How buyers use the graph
Activation workflows
- CRM / CDP onboarding: hashed customer list → MAID, CTV, and cookie-less activations.
- Suppression and dedupe across paid social, programmatic, and CTV buys.
- Householding for reach/frequency and loyalty-tier targeting.
- Partner audience extension with disclosed match tiers.
Measurement workflows
- Exposure → outcome joins across web, app, and TV (incl. ACR exposure).
- Incrementality and MMM inputs where identity quality gates lift.
- Cross-walled-garden reporting when DSPs will not share raw logs.
- Decay-aware cohorts: stale MAID/HEM pairs excluded by last_seen rules.
Evaluate before you license
Most identity contracts fail because buyers validate deck averages instead of their own seed file. Run a matched sample (10K-100K hashed emails or MAIDs), measure join yield by geography and cohort, then lock governance and refresh terms before production volume.
Request these artifacts
- Match-rate readout by cohort / geo
- Refresh cadence + decay curves
- Join-key dictionary + normalization rules
- Permitted uses + deletion SLAs
Red flags
- “100% match” without seed breakdown
- No downstream opt-out propagation
- Peak-cohort sizing after FTC panel deletions
- Vague “compliance” without use cases
Two-week pilot path
- Matched sample vs your CRM seed
- Confirm S3 / Snowflake delivery
- Lock governance + exclusions
- Production license
Delivery & integration
Batch & warehouse
- Parquet, CSV, or JSON on S3 / SFTP
- Daily or weekly refresh (scoped per product)
- Snowflake secure share on request
- Change-delta files for incremental merges
Activation surfaces
- Clean-room and warehouse joins (buyer-controlled)
- DSP / CDP onboarding via partner pipes
- Shared keys with mobility and CTV products
- Evaluation checklists in the developer hub
Use cases by buyer type
Adtech & agencies
CRM onboarding to DSP and CTV, suppression, and cross-channel frequency management.
Measurement & attribution
Exposure→outcome joins across web, app, and TV without walled-garden lock-in.
Financial & alt data
Panel joins and tickerized consumer signals where identity quality gates model lift.
Procurement & compliance
Match-rate diligence, decay economics, and governance before production license.
Request a matched sample
Send a hashed seed file and target use case. GSDSI typically scopes match keys, exclusions, and refresh cadence, then delivers a pilot file for join-yield analysis within a few business days.
Frequently asked questions about the GSDSI identity graph
What is GSDSI’s identity graph?
GSDSI’s identity graph links mobile advertising IDs (MAIDs), hashed emails (HEMs), and CTV identifiers so buyers can onboard CRM audiences, stitch cross-device touchpoints, and measure across channels. The commercial backbone is the MAID feed (~200M+ US MAID→HEM linkages and 700M+ international pairs), extended by the core email file, Euclidean distance signal, and CTV/ACR feeds where TV exposure must join the same graph.
What is MAID to HEM resolution?
MAID to HEM resolution links a mobile advertising ID (Apple IDFA or Google GAID) to a hashed email address (HEM) that belongs to the same person or household. It works in both directions: HEM to MAID lets a hashed CRM list reach the same people on mobile and CTV, and MAID to HEM ties a mobile audience or exposure log back to known customers for measurement. GSDSI’s MAID feed carries ~200M+ US MAID→HEM linkages and 700M+ international pairs; test join yield on your own hashed seed before licensing.
Can GSDSI match MAIDs to IP addresses?
Yes, for US programs. GSDSI licenses a feed that matches MAIDs to location and IP signals, and the identity graph schema carries an ip_hash field for household- or device-level linkage where a deterministic hashed email is unavailable. IP-based links are probabilistic, so ask for confidence tiers and last_seen dates, and confirm permitted uses before activation.
How is match rate measured?
Match rate should be measured on your own hashed seed file. CRM emails or MAID lists: broken out by geography, cohort, and confidence tier. Deck-level “average match” numbers are not a substitute for a matched-sample readout during procurement.
How often is the identity graph refreshed?
Refresh cadence is product-specific; MAID→HEM linkages are updated on a daily cadence for US programs. Buyers should request last_seen fields, decay documentation, and change-delta delivery so stale edges can be filtered in warehouse merges.
What delivery formats does GSDSI support?
Typical delivery is Parquet, CSV, or JSON via S3 or SFTP, with Snowflake secure share available on request. Activation may also flow through clean rooms, DSP onboarding pipes, or warehouse joins controlled by the buyer.
How does GSDSI handle privacy and opt-outs?
Identity products are sourced under consent-aware programs with documented permitted uses, retention boundaries, and opt-out/delete handling. Procurement should align activation and measurement use cases with the privacy policy, sourcing methodology, and any state or sector exclusions before production scale.
What is the fastest way to evaluate GSDSI identity data?
Request a matched sample against your hashed seed (10K-100K rows), measure join yield and freshness, confirm delivery path and governance terms, then move to a scoped production license. See the enterprise pilot checklist and MAID identity-graph diligence guide in resources.