An identity graph provider sells the relationships between identifiers: which hashed email, mobile ad ID, cookie, CTV device, household IP or postal address belongs to the same person. The market divides into three groups: adtech and consumer marketing graphs, CDP (first-party) graphs, and B2B person-level graphs. They are not interchangeable. A graph built to match a CRM file to TV households won’t tell a sales rep who read your pricing page, and the reverse is also true.
This list covers the main identity graph companies in each group, what each graph is keyed on, whether its links are deterministic or probabilistic, and how to evaluate one on your own data. Scale figures are the companies’ own published numbers, linked to the source. They are not independent audits. We checked facts and ownership against the linked pages on September 24, 2026.
In one sentence: Pick the identity graph category by your job (media activation, customer unification or visitor identification) before you compare providers within it.
Why the identity graph market changed
Three shifts are behind the current provider landscape:
- Consolidation. Many standalone graphs now sit inside credit bureaus, data companies or agency holding groups. Experian acquired Tapad in 2020. TransUnion closed its $3.1 billion acquisition of Neustar in December 2021. Verisk acquired Infutor in 2022. Omnicom completed its acquisition of Interpublic, Acxiom’s parent, in November 2025.
- Exits. Oracle ended its advertising business, including its data management platform, Digital Audiences (OnRamp) and cross-device services, on September 30, 2024. If an old vendor list still names Oracle’s ID Graph, that list is out of date.
- Cookies stayed, but signal loss didn’t reverse. Google said in April 2025 that it will keep third-party cookie choice in Chrome and not roll out a standalone prompt. Even so, graphs have moved toward hashed email, CTV and authenticated IDs such as UID2 because those identifiers hold up better across browsers and devices.
Adtech and consumer identity graph companies
These graphs connect offline consumer records (name, postal address, phone, email) to digital identifiers so brands and platforms can onboard CRM data, target audiences, cap frequency and measure. Most are keyed on the person or household, not the employee or company.
LiveRamp (RampID)
LiveRamp’s graph is the reference point for the search “LiveRamp identity graph”. Its documentation says LiveRamp “deterministically matches offline personally identifiable information (PII: name & address, email, phone) and online devices to people-based IDs” (RampID methodology). LiveRamp reports PII maintained on 245 million individuals in the U.S. and connects cookies, IDFAs, AAIDs and platform IDs to RampIDs. LiveRamp states the tradeoff openly: its fully deterministic graph has “less scale than probabilistic vendors” (source).
- Keyed on: name/postal, email (plain or hashed), phone → RampID → cookies, MAIDs, platform IDs
- Best for: CRM onboarding, clean-room collaboration and activation across a large partner ecosystem
Experian (Digital Graph, including Tapad)
Experian bought Tapad, a cross-device identity company, for about $280 million in 2020 (Experian). Experian lists 4.2 billion digital identifiers, 300 million CTV IDs and 250 million individuals in its U.S. Digital Graph, linking MAIDs, CTV IDs, universal IDs and hashed emails at the household or individual level with machine-learning models.
- Keyed on: household + individual; HEMs, MAIDs, CTV IDs, IPs, universal IDs
- Best for: cross-device and CTV targeting, household-level measurement
TransUnion (TruAudience, formerly Neustar)
TransUnion’s marketing graph comes largely from Neustar. In January 2024 TransUnion said its graph represents 98% of U.S. adults, covering 125 million+ households and 250 million adults, with 111 million IPs tied to U.S. households and more than 1 billion device IDs.
- Keyed on: individual + household; postal, phone, email, IP, device IDs
- Best for: household-level addressability, CTV and measurement, especially for brands already using TransUnion data
Epsilon (CORE ID, Publicis)
Publicis bought Epsilon for $4.4 billion in 2019. Epsilon describes CORE ID as a person-level identifier anchored to name and postal address and tied to transaction data.
- Keyed on: name + postal, with transactions; devices linked to the person
- Best for: retail and CPG brands, and Publicis clients running people-based media
Acxiom (Real ID, Omnicom)
Acxiom’s Real ID cites 260 million U.S. consumers and a 2.6 billion global audience. It’s built to let brands “build and control” their own interoperable identity graph inside their own cloud environment. That makes Acxiom a common answer to the search “private identity graph”.
- Keyed on: name, postal, email, phone, digital IDs; can be deployed as a brand-owned private graph
- Best for: enterprises that want a private graph and a single customer view without moving PII out of their own environment
Verisk Marketing Solutions (Infutor)
Infutor, now part of Verisk, describes its TrueSource Identity Graph as covering 260+ million U.S. consumers with 100 million daily updates, compiled from “deterministic, authoritative and permissible data”. It’s often used to verify identity on lead forms and to fill in consumer contact data, particularly in insurance and other considered-purchase categories.
- Keyed on: name, postal, phone, email
- Best for: consumer identity verification, contact data and lead validation
Stirista (OMNA)
Stirista’s OMNA identity graph links hashed emails, MAIDs, IPs, CTV identifiers and postal data. It advertises 1.5B+ hashed emails and deterministic identity signals, and it supports both B2C and B2B audiences. People searching “hashed email and MAID identity graph” often end up here.
- Keyed on: HEM, MAID, IP, CTV, postal
- Best for: programmatic and CTV activation that starts from email or device audiences
Unified ID 2.0 (The Trade Desk ecosystem)
UID2 is not a vendor graph. Its documentation calls it an open-source framework “that enables deterministic identity for advertising opportunities” on publisher websites, mobile apps and CTV apps. UID2 tokens are generated from hashed or unhashed email addresses or phone numbers, and consumers can opt out through a transparency portal. The Trade Desk currently serves as a public operator for the framework.
- Keyed on: hashed and encrypted email or phone
- Best for: SSP/DSP-compatible addressability without third-party cookies
CTV note: An identity graph is used in CTV advertising by linking a TV or streaming device (usually through the household IP and a CTV device ID) to the people and other devices in that household. That’s how a brand can reach its CRM audience on TV, cap frequency across screens and tie TV exposure to later website or store visits. Experian, TransUnion, Stirista, LiveRamp and UID2 all support CTV identifiers.
Try Leadpipe free with 500 leads →
CDP and first-party identity graph platforms
A customer data platform builds a first-party identity graph from identifiers you already collect: logins, emails, device IDs, loyalty numbers. It doesn’t add new people. It connects your existing records correctly.
| Platform | Matching approach (per vendor docs) | Best for |
|---|---|---|
| Twilio Segment (Unify) | “All Profile matching is deterministic and based on first-party data that you’ve collected. Segment doesn’t support probabilistic matching.” | Product and marketing teams stitching web, app and server events in real time |
| mParticle by Rokt (IDSync) | Deterministic identity strategies over customer IDs, emails and device IDs; Rokt acquired mParticle for $300 million in January 2025 | Mobile-first and commerce brands |
| Amperity (Stitch) | Patented machine-learning pairwise comparison plus rules you define, producing an Amperity ID per person | Retail and consumer brands with messy, multi-source customer records |
| Acxiom Real ID | Brand-owned private graph built in the client environment | Enterprises that want a private graph plus external reference data |
The difference matters for accuracy. A strictly deterministic CDP graph can miss links, but it rarely merges two different people. A graph that mixes in ML or probabilistic links finds more connections, so you should review a sample of its merges yourself.
B2B person-level identity graphs
Consumer graphs answer “which household is this device in?” B2B teams ask a different question: which professional at which company visited, and how do I reach them? A B2B identity graph links work and personal emails, LinkedIn profiles, job titles and employers to the device, cookie and IP signals that show up on a website.
Leadpipe
Leadpipe runs a proprietary, deterministic, person-level identity graph built for website visitor identification. It isn’t an adtech onboarding platform and doesn’t resell another vendor’s graph. When a visitor lands on your site, Leadpipe resolves the session against its own graph. A link enters primary serving only with verified evidence, such as a consented hashed email tied to a cookie. Probabilistic links never drive the primary identification. We describe the build in how we built the Leadpipe identity graph, including daily re-verification.
- Keyed on: person; hashed email, cookie, device and IP signals linked to name, work and personal email, LinkedIn URL, title and company
- Deterministic or probabilistic: deterministic for primary matches
- Match rate: about 30–40% of US traffic resolved to individuals, with international IP-to-company identification for visitors outside the US
- Delivery: dashboard, webhooks, CRM integrations and the identity API; person-level intent via Orbit
- Best for: B2B sales and marketing teams, agencies and platforms that need named visitor records
For a wider look at B2B tools, see person-level vs company-level visitor identification and our ten-tool comparison.
Identity graph companies compared
| Provider | Category | Primary keys | Match method | Primary use | Best for |
|---|---|---|---|---|---|
| LiveRamp | Adtech / onboarding | PII → RampID; cookies, MAIDs | Deterministic | CRM onboarding, clean rooms, activation | Large brands and publishers |
| Experian (incl. Tapad) | Consumer / adtech | HEM, MAID, CTV ID, IP, household | ML-driven linking | Cross-device, CTV | Household targeting and measurement |
| TransUnion TruAudience | Consumer / adtech | Postal, phone, email, IP, devices | AI clustering with confidence scoring | Addressability, measurement | Household reach |
| Epsilon CORE ID | Consumer / adtech | Name + postal, transactions | Person-based | Retail and CPG media | Publicis clients |
| Acxiom Real ID | Consumer / private graph | Name, postal, email, digital IDs | Brand-configured | Private enterprise graph | Enterprises keeping PII in-house |
| Verisk (Infutor) | Consumer data | Name, postal, phone, email | Deterministic sources | Verification, contact data | Insurance, lead validation |
| Stirista OMNA | Consumer + B2B adtech | HEM, MAID, IP, CTV, postal | Deterministic signals | Programmatic and CTV | Email and device audiences |
| UID2 | Open ID framework | Hashed email or phone | Deterministic | SSP/DSP addressability | Cookieless programmatic |
| Segment / mParticle | CDP first-party | Your IDs, emails, devices | Deterministic | Profile unification | Product-led and app brands |
| Amperity | CDP first-party | Your customer records | ML + rules | Profile unification | Retail and consumer brands |
| Leadpipe | B2B person-level | HEM, cookie, device, IP → person + company | Deterministic | Website visitor identification | B2B revenue teams |
How to evaluate identity graph providers
Match rate: define the denominator
“High match rate identity graph” claims are only comparable when the denominator is the same. Ask what was matched to what: CRM records to a person ID, IDs to a live device, or website sessions to a named person. Also ask whether “matched” means an ID came back or a usable record came back. LiveRamp’s documentation shows how much input quality matters: match rates can be “slightly lower with hashed email addresses than with plaintext emails” (LiveRamp).
Accuracy: test against ground truth
Accuracy varies a lot between providers. A Truthset and CIMM study analyzed 3.9 billion hashed email records and found email-to-postal linkage accuracy ranging from 32% to 69% depending on the provider. Best practices for testing a third-party business identity graph:
- Hold out known records. Send a sample of customers whose identity you’ve already verified (CRM contacts, authenticated users) and check how often the returned person, email and company are correct.
- Score each field separately. A correct company with a wrong contact is a partial miss, not a match.
- Log unmatched records separately. Don’t let a low-coverage provider look accurate just because it returns less.
- Run providers side by side on the same traffic or file and in the same time window.
See how visitor identification accuracy is measured for a scoring template.
Coverage and refresh cadence
Ask for coverage by geography, entity type (person, household, company) and identifier type. Ask how often links are re-verified, not just how often files are delivered. Stirista advertises real-time updates. Infutor reports 100 million daily updates. Leadpipe re-verifies on a daily cycle wherever a fresh signal exists. Job changes make freshness especially important in B2B, where a stale employer field sends outreach to the wrong company.
Privacy and consent basis
Ask where identifiers came from, what consent or notice covers them, how opt-outs spread through linked IDs, and whether a Data Processing Agreement (DPA) is available. For US data, check CCPA/CPRA handling. For EU and UK visitors, have counsel confirm your GDPR lawful basis for the specific use. Our GDPR and CCPA guides cover visitor identification specifically.
Connecting CRM data to a third-party identity graph
There are three common patterns:
- Hashed email onboarding. Normalize emails first. LiveRamp’s spec is to “convert all characters to lowercase, remove all whitespace, and perform UTF-8 encoding” before hashing, and it accepts SHA-256, MD5 or SHA-1 (docs). Then upload the file and receive a pseudonymous ID for activation.
- Clean rooms. Both sides match inside a controlled environment on hashed PII or IDs, and only aggregate or approved outputs leave it.
- API. Send an identifier (email, IP, session) and get a resolved profile back in real time. This is the usual pattern for B2B and visitor identification. See identity graph APIs and the Leadpipe API docs.
Multi-source and real-time needs
A multi-source identity graph combines several input feeds, so ask how the provider resolves conflicts when two sources disagree on the same field. If your use case is real time, such as bidding, personalization or routing a live website visitor to sales, test latency on real traffic rather than accepting a batch SLA.
Which identity graph should you choose?
- You buy media and need to onboard a CRM or reach households on CTV: LiveRamp, Experian, TransUnion, Epsilon, Stirista or UID2-compatible partners.
- You want a private graph under your own governance: Acxiom Real ID or a CDP (Amperity, Segment, mParticle).
- You need to know which people are on your website right now: a B2B person-level graph like Leadpipe Identification.
Many teams use more than one. A CDP owns the customer profile, an adtech graph handles activation, and a visitor identification graph finds new, in-market buyers the other two never see.
FAQ
What are the top identity graph providers in 2026?
For advertising and consumer marketing, the major identity graph companies are LiveRamp, Experian (including Tapad), TransUnion (including Neustar), Epsilon, Acxiom, Verisk (Infutor) and Stirista, plus the open UID2 framework. For first-party graphs: Amperity, Twilio Segment and mParticle. For B2B person-level visitor identification: Leadpipe.
What is a deterministic identity graph?
A deterministic identity graph links identifiers only when there’s direct evidence, such as the same email used on a login and a device. LiveRamp, Segment, UID2 and Leadpipe describe their core matching as deterministic. Probabilistic graphs infer links from patterns like IP, device and timing, which gains reach at the cost of more false positives.
Is Oracle still an identity graph provider?
No. Oracle ended its advertising products, including its data management platform, Digital Audiences (OnRamp) and cross-device services, on September 30, 2024.
What is a private identity graph?
A private identity graph is built and governed inside your own environment from your first-party data, sometimes enriched with a provider’s reference data. Acxiom Real ID and CDPs such as Amperity support this model.
How do I connect my CRM data to a third-party identity graph?
Normalize and hash emails (lowercase, trim, SHA-256), then use the provider’s onboarding upload, a clean room or an API. Check match rates on a held-out sample before you activate.
What is the best B2B identity graph for website visitors?
It depends on whether you need people or just companies. Leadpipe resolves US visitors to named individuals with a deterministic proprietary graph (about 30–40% match rate) and identifies non-US visitors at the company level with IP-to-company.
Related Articles
- What Is an Identity Graph? Definition and Examples
- How Identity Graphs Work
- How We Built the Leadpipe Identity Graph
- What Is Identity Resolution?
- Identity Resolution APIs Compared
- B2B Identity Resolution API Guide
See what a B2B person-level identity graph finds on your own traffic. Start free with 500 identified leads, no credit card →
Sources and review basis
Published by Leadpipe. Vendor references support product descriptions; recommendations reflect our assessment of fit. Confirm current plan and contract terms before buying.
