Marta KravsIdentity signals come from different sources, including publisher first-party data, browser cookies, mobile advertising IDs, and identity providers. Publisher-side technology and SSPs can collect or receive permitted identifiers and include them in OpenRTB bid requests.
Depending on the integration, a bid request may be sent directly to a DSP or routed through an exchange. Identifiers may be mapped between platforms along the way, and identity resolution can happen through connected identity providers or DSP-side systems before the bid decision.
Identity resolution in programmatic advertising links identities that may refer to the same user, device, or household. Deterministic identity resolution relies on direct matches, while probabilistic identity resolution uses various cues to determine the likelihood of a match.
Google's 2025 policy shift did not resolve the industry's identity challenges. Earlier, marketers identified potential risks if third-party cookies disappeared: 66% of those surveyed by Supermetrics were concerned about cross-channel tracking, and 57% expected targeted advertising to become less effective.
For SSPs and DSPs, the technical challenge is broader: identifier availability varies across browsers and platforms, while identity ecosystems remain fragmented and only partially interoperable.
TL;DR
Identity signals include cookies, mobile IDs, first-party IDs, and alternative identifiers.
Identity resolution links identifiers to users, devices, or households. These relationships can be established before an auction or used during bid evaluation.
Deterministic matching offers higher confidence.
Probabilistic matching uses multiple signals and statistical inference.
Programmatic identity remains fragmented, requiring multiple identity methods.
Consent and privacy rules determine which signals can enter the bidstream.
What counts as an "identity signal" in a bid request
Identity signals in programmatic advertising are identifiers and device attributes that help an SSP, DSP, or identity provider recognize a user or device. They include user IDs, extended IDs, mobile advertising IDs, and device-related data in OpenRTB.
Deterministic identifiers are tied directly to a known browser, device, or authenticated account, which lets platforms recognize the same one across requests. Linking these identifiers to the same individual across environments requires additional verified relationships. Common deterministic identifiers include:
cookie-based IDs,
mobile advertising IDs such as Apple's Identifier for Advertisers (IDFA) or Google Advertising ID (GAID),
publisher first-party IDs, and
IDs derived from authenticated data, such as a hashed email.
Authenticated first-party data, such as an email address, can be processed through a supported identity framework to generate a pseudonymous identifier. An identity provider can then map it to a provider-specific ID that a connected DSP recognizes, and the DSP can match the result against an existing identity graph.
Probabilistic identity resolution can work when no direct identifier is available. The signals include:
IP address,
user agent,
device attributes,
operating system, and
location.
These signals can support probabilistic matching where platform rules and privacy requirements allow it. Device fingerprinting is restricted differently across environments. Apple prohibits using device signals to uniquely identify a user or device, while in the EU/EEA fingerprinting generally falls under ePrivacy consent requirements.
OpenRTB carries identity data through fields such as user.id, buyeruid, extended IDs (eids), and the device object. These signals can then support identity resolution in programmatic advertising by helping platforms match requests to known users, devices, or households.
Where identity signals enter the bid stream
Identity signals usually enter the bid stream from the sell side, when a publisher, header-bidding wrapper, SSP, or exchange creates or enriches the bid request. In OpenRTB, the user object can contain an exchange-specific user ID, a buyer-specific ID, or extended IDs from an identity provider. Device identifiers can appear in the device object when applicable.
The signal may come from a publisher's first-party ID, a cookie sync, a mobile advertising ID, or an identity resolution platform. It is not always passed to every DSP unchanged. An SSP or exchange may first map its own identifier to one recognized by the buyer, so different buyers may receive different identifiers for the same user.
For a cookie-based ID, the flow can look like this:
The publisher or SSP has a cookie ID that is available and permitted for use.
The SSP creates the bid request and includes the relevant identifier in the user object.
If the SSP and DSP have a cookie-sync mapping, the buyer-specific ID can be passed through buyeruid.
The DSP looks up that ID in its own systems or through an identity provider.
If the lookup resolves to a known profile or segment, the corresponding audience data can inform the bid.
The DSP can use these identifiers to resolve identity, retrieve audience data, and apply its bidding logic. Identity signals in programmatic advertising can therefore be mapped at several points before the DSP decides whether to bid and how much.
Why Google's cookie reversal didn't simplify anything
The market was already moving toward multiple identity methods. Google's decision to keep third-party cookies in Chrome did not bring the industry back to a cookie-first identity model. In October 2025, Google retired many Privacy Sandbox technologies. Safari still blocks third-party cookies by default, while Firefox restricts cross-site tracking through its own privacy protections. Meanwhile, The Trade Desk and other companies continue to develop alternative IDs such as UID2.
By then, the industry had already spent years preparing for a cookieless future, which accelerated investment in first-party data, contextual targeting, and alternative IDs.
The result is a fragmented identity layer. Even today, DSPs need identity resolution tools that can work with cookies where available, mobile advertising IDs, publisher first-party IDs, and alternative identity solutions. Customer identity resolution can connect authenticated first-party records across systems before translating them into IDs or audience segments for programmatic activation. As a result, identity resolution advertising now operates across multiple signal types, browser policies, and platform rules.
Identity resolution: Matching the signal to a profile
By the time an identity signal reaches the DSP, it may already have been mapped or enriched. The platform then uses identity resolution solutions and identity graphs to match the identifier. A successful match connects it to a relevant profile or audience segment.
Depending on the signal, the match may be deterministic or probabilistic. Some providers, including ID5, combine both approaches within the same identity graph.
Match rate affects the share of bid requests a DSP can associate with audience data. When resolution succeeds, the DSP may apply relevant audience segments, frequency controls, or other user-level bidding logic, depending on data availability, permissions, and platform configuration. For identity resolution advertising, this determines whether audience data can inform the bid. Without a usable identifier, the DSP may rely on contextual or other ID-less signals.
Identity graph advertising depends on this lookup: the graph stores relationships between users, devices, households, and other known IDs, and incoming identifiers are checked against them.
What gets stripped before the signal reaches the DSP
Not every identity signal collected on the sell side reaches the DSP. Consent and privacy rules are the most common reason, but identifiers can also be missing because of absent ID mappings, unsupported identity providers, technical configuration, or buyer-specific requirements.
Under TCF v2.3, the Consent Management Platform records the user's privacy choices in the TC String and makes them available to participating vendors. SSPs, exchanges, and identity resolution vendors then use these signals, along with applicable legal requirements and their own policies, to decide which identifiers they can process or include in the bid request.
The other causes are operational. A DSP without an ID mapping for the SSP can't recognize the SSP's identifier, a buyer may not use an identity provider it doesn't support, and some buyers accept only certain ID types or configurations.
An identity resolution platform or identity resolution API may therefore receive fewer identifiers than the publisher originally had available. This directly limits the inputs available to downstream identity resolution technology.
That does not make the remaining bidstream anonymous. OpenRTB can still carry pseudonymous user IDs, IP addresses, device attributes, and geolocation. Individually or in combination, these identity signals in programmatic advertising can help connect a request to an identity graph.
Publishers increasingly turn to consented first-party data to feed their identity resolution capabilities. It gives publishers a more direct source of audience data amid fragmented third-party identifier availability across browsers and platforms.
Comparing the main identity signal types
Some identity signals are more persistent than others; the less persistent ones depend on temporary browser, device, or network conditions. This affects match confidence and audience activation.
Signal type | Persistence & match accuracy | Where it breaks down |
Authenticated first-party / hashed email-derived IDs | High match confidence; persistence varies | Limited to authenticated users and consented data |
Mobile advertising IDs (IDFA/GAID) | Deterministic device-level ID when available | Resettable and restricted by mobile privacy controls |
Third-party cookie IDs | Deterministic for that browser/vendor while the cookie persists | Blocked in some browsers and lost when cookies are cleared |
Alternative IDs such as UID2 or ID5 | Depends on the provider's identity methodology, available signals, and ecosystem coverage. | Coverage depends on adoption and available data |
Probabilistic device or household signals | Variable and confidence-based | Weaker when device, IP, or network conditions change |
What identity support means for white-label SSP operators
For a white-label SSP operator, identity support is mainly an infrastructure question. The platform needs to support multiple identity frameworks, pass relevant OpenRTB identity fields, and stay compatible with the demand partners that consume them.
Identity and audience data integrations serve different purposes. Identity solutions such as ID5 support addressability and identity resolution, helping platforms recognize users across supported environments. Audience data integrations such as Eyeota provide audience segments for activation, including curated deal workflows.
TeqBlaze's configurable white-label infrastructure supports both, with implementation requirements depending on the provider and use case.
Final thoughts
Programmatic identity has not consolidated around a single standard. Platforms still need to process mobile IDs, first-party identifiers, alternative IDs, cookies where available, and probabilistic signals in parallel.
Platforms that can process multiple identity types depend less on any single vendor or identifier's coverage or availability. This gives them more opportunities to match audiences across different environments and maintain addressability as individual signals become unavailable.
On the supply side, supporting multiple data sources often requires SSP-level integrations. TeqBlaze’s Eyeota integration shows how audience data can be activated at this level rather than added only downstream. For a broader view of TeqBlaze’s work, see client feedback on Clutch.
FAQ
What is an identity signal in programmatic advertising?
Identity signals are identifiers or device characteristics that enable adtech platforms to identify users, devices, browsers, or households. Examples include cookie IDs, mobile advertising IDs, publisher first-party IDs, alternative IDs, and signals used for probabilistic matching.
Did Google's Privacy Sandbox shutdown bring back third-party cookies?
No. Chrome retained third-party cookie choice, but this did not restore cookies as a universal identity layer. Safari and Firefox still restrict them, while first-party data and alternative IDs remain in use.
What is identity resolution, and how do deterministic and probabilistic methods differ?
Identity resolution connects identifiers and signals that may belong to the same user, device, or household. Deterministic identity resolution uses direct identifiers such as authenticated first-party IDs or identifiers derived from hashed emails. Probabilistic resolution analyzes signals such as IP address, device characteristics, and location to estimate whether records likely belong to the same user or household.
How does a DSP match an identity signal to an audience segment?
The DSP checks the identifier against its own identity graph or a connected identity provider. If it resolves to a known profile, the DSP can associate the impression with relevant audience segments for bidding.
Is bidstream data considered personal data under GDPR?
Yes, it can be. Bid requests may contain pseudonymous IDs, IP addresses, device information, and geolocation that relate to or can help single out an individual. Under GDPR, data does not need to include a name or email address to qualify as personal data.
